CN120011179B - Inspection method for multi-system operation and maintenance using artificial intelligence-based RPA inspection robots - Google Patents

Abstract

The present invention relates to the field of IT system operation and maintenance management, and discloses an inspection method for multi-system operation and maintenance by an RPA inspection robot based on artificial intelligence. The RPA inspection robot establishes connections with multiple target systems that need to be inspected through a pre-set interface protocol, and automatically identifies login information of the login system page based on process automation RPA software; automatically extracts verification code information from the login information, and calls a verification code recognition service interface to perform verification code recognition; after the verification code recognition is passed, the RPA robot automatically logs in to each target system that needs to be inspected, obtains page information, and stores the collected page information in the inspection directory of the RPA robot; according to the inspection directory, the information collected from each target system that needs to be inspected is unified and aggregated, and anomaly detection is performed, and the alarm information of the page alarm module is analyzed; the alarm information page is intercepted, and the RPA robot sends the alarm information to the operation and maintenance personnel; the present invention improves work efficiency.

Description

Method for carrying out multi-system operation and maintenance on RPA inspection robot based on artificial intelligence

Technical Field

The invention relates to the field of IT system operation and maintenance management, in particular to an inspection method for carrying out multi-system operation and maintenance by an RPA inspection robot based on artificial intelligence.

Background

1. The technical complexity and updating are rapid ‌, the operation and maintenance work relates to a plurality of technical fields including network, system, database, application, safety and the like, the complexity and the speciality of the fields require the operation and maintenance personnel to have comprehensive knowledge and skills, and meanwhile, the rapid development of the technology enables the operation and maintenance personnel to continuously learn new knowledge and master new skills, and the difficulty ‌ of the operation and maintenance work is increased.

2. The workload is large and details are complicated ‌, operation and maintenance personnel need to be responsible for a plurality of tasks such as daily inspection, monitoring, fault handling, system upgrading and the like, the tasks are heavy and high in concentration, in addition, the details of the operation are complicated, such as configuration management, log analysis, performance tuning and the like, a great deal of time and effort are required, and the operation and maintenance personnel are easy to make mistakes ‌.

3. Monitoring and troubleshooting ‌, along with the complexity of an IT system, the monitoring difficulty is increased, the running state of the system needs to be monitored in real time, and potential problems are found in time, however, the monitoring work becomes extremely difficult due to factors such as diversity of data sources, high real-time requirements and the like. Troubleshooting is also complicated by the complex system architecture and numerous components, which require a great deal of time and effort.

4. The automation and standardization are insufficient ‌, the operation and maintenance work of part of enterprises still depends on manual operation, so that the work efficiency is low and errors are easy to occur, the application of the automation operation and maintenance tool can obviously improve the operation and maintenance efficiency and reduce the risk of human errors, but the current automation degree is insufficient as a pain point. Furthermore, the lack of uniform standards results in a lack of standardization ‌ for operation and maintenance work.

5. Resource limitation and pressure ‌. When an operation and maintenance team faces a complex operation and maintenance task, the operation and maintenance team is often influenced by resource limitation, including human resources, technical resources, fund resources and the like. Meanwhile, the operation and maintenance team bears important responsibility for guaranteeing the stable operation of the business system, and the huge responsibility pressure and psychological pressure lead the operation and maintenance work to be harder ‌.

‌ 6, Cross-department collaboration and communication ‌. In large enterprises, operation and maintenance work relates to collaboration and communication among multiple departments and teams, however, due to asymmetric information among departments, poor communication channels and the like, the collaboration efficiency is often low.

Disclosure of Invention

The invention aims to solve the problems, and designs a method for carrying out multi-system operation and maintenance on an RPA inspection robot based on artificial intelligence.

The invention provides an artificial intelligence-based RPA inspection robot inspection method for multi-system operation and maintenance, which comprises the following steps:

the RPA inspection robot establishes connection with a plurality of target systems to be inspected through a preset interface protocol, and automatically identifies login information of a login system page based on flow automation RPA software, wherein the login information at least comprises login user name, password and verification code information;

Automatically extracting verification code information in login information, and calling a verification code identification service interface to identify the verification code;

After the verification code is identified, the RPA robot automatically logs in each target system to be inspected, acquires page information at least comprising a target page and system running state page information according to the operation and maintenance targets of each system, and stores the collected page information into an inspection catalog of the RPA robot, wherein the target system to be inspected at least comprises a whole network behavior management system, an operation and maintenance audit and risk control system, a network controller system, an enterprise data backup and recovery system, an internet behavior management system, an application delivery management system, a firewall system, a cloud platform system and an IT operation and maintenance monitoring platform;

According to the inspection catalogue, carrying out unified aggregation processing on information acquired by each target system needing inspection, carrying out anomaly detection, and analyzing alarm information of a page alarm module;

And intercepting an alarm information page, storing the alarm information page into the RPA robot, and sending alarm information to an operation and maintenance person by the RPA robot.

Optionally, in a first implementation manner of the present invention, the RPA inspection robot establishes connection with a plurality of target systems to be inspected through a preset interface protocol, automatically identifies login information of a login system page based on flow automation RPA software, and includes:

Loading and reading an RPA tool software package and a development tool package corresponding to the RPA tool software package, and integrating the RPA tool and the inspection rule;

starting an RAP engine, initializing engine parameters and initializing operation instructions, wherein the engine parameters at least comprise thread pool size, timeout time setting and retry time threshold;

The hierarchical structure of the inspection catalog is defined, the storage rules are analyzed, the storage formats of different types of data are determined, and the storage formats at least comprise that the performance index data are stored in a CSV format according to time sequence, and the log files are stored under a folder named by date in an original text form;

The RPA engine is called, and the website of the inspection system stored in the configuration library in advance is transmitted to the browser starting parameter, so that the browser automatically navigates to a login page of the target inspection system after starting;

when the browser loads a login page of the inspection system and encounters a certificate error prompt, the browser commands to ignore the webpage certificate error and enters the inspection login page through thisisunsafe;

And driving the RPA engine to automatically identify the page element, judging whether the page is loaded or not by judging whether the page element is loaded or not, and when the loading of the page element is completed, automatically inputting a user name by a user name input box for identifying the login page, automatically inputting a password by a password input box for identifying the login page, and automatically hooking a consent selection box for identifying the login page, then, completing the system login by an identification login button.

Optionally, in a second implementation manner of the present invention, the automatically extracting the identifying code information in the login information and calling the identifying code identifying service interface to identify the identifying code includes:

Starting an RPA browser plug-in running environment, loading a preconfigured page element identification rule set, traversing the HTML structure of the page through a DOM analysis technology, positioning the node where the verification code information element is positioned, and identifying the verification code;

Transmitting an HTTPS request to a server, monitoring the response state of the request in real time, if a successful response is received, acquiring a call token, and reading the picture identification API access information stored in the local configuration file through the call token;

And extracting the returned JSON value, analyzing the JSON value to obtain an identification result, transmitting the identification result to a system for login, automatically filling a verification code input box in the login process, and completing the whole verification code processing flow.

Optionally, in a third implementation manner of the present invention, after the verification code passes the identification, the RPA robot automatically logs in each target system to be inspected, obtains, according to the targets of the operation and maintenance of each system, page information including at least a target page and system operation status page information, and stores the collected page information in an inspection directory of the RPA robot, including:

after the verification code is successfully identified and the verification code is automatically filled, driving the RPA robot to simulate the operation of manually clicking a login button by utilizing the login account and password information acquired before, and sending a login request to a target system server;

Continuously monitoring response information returned by the server, confirming whether the login is successful or not by analyzing the response state code, the page jump identifier and the login success prompt element, and if the successful login signal is not detected within 10 seconds, restarting a login process by the RPA robot according to a preset retry mechanism until each target system needing to be inspected is successfully logged in;

After successful login, starting a built-in page acquisition module of the RPA robot, and allocating a memory space as a temporary data buffer area to read a page acquisition rule file aiming at the target system, wherein the page acquisition rule at least comprises acquisition priorities, acquisition frequencies and data storage formats of different page elements;

initializing parameters of a page acquisition module according to a page acquisition rule, performing depth analysis on a target page by using a built-in DOM analysis engine through an RPA robot, and positioning page elements where key information is located by identifying an HTML tag structure, a CSS style class name and a JavaScript event trigger area of the page;

The RPA robot automatically extracts the data content in the page element where the key information is located, and preprocesses the data content;

Driving the RPA robot to jump from the current page to the system running state page according to the URL address of the system running state page or the page jump path configuration information, and excavating detailed system running state data in the page by the RPA robot after the loading of the system running state page is completed;

And integrating the data content in the page element where the key information temporarily stored in the memory cache area is located and the detailed data of the system running state, and storing the integrated data into the patrol directory by using the RPA robot according to the storage path.

Optionally, in a fourth implementation manner of the present invention, the system operation status detailed data in the RPA robot mining page includes:

Capturing an AJAX request interacted with a server behind a page by using a network request sniffing technology of a bottom layer, analyzing returned data, and acquiring a system performance index pushed by a server in real time, wherein the system performance index at least comprises a CPU (central processing unit) utilization rate, a memory occupancy rate and a network bandwidth utilization rate;

for system operation trend data displayed in a visual chart form on a page, converting data points in the chart into a structured numerical sequence by combining an OCR technology and a chart analysis algorithm;

The collected system running state data is temporarily stored in a memory buffer area, and preliminary arrangement is carried out according to preset data classification rules, wherein the preset data classification rules at least comprise data type classification, time stamp classification and data source classification.

Optionally, in a fifth implementation manner of the present invention, according to the inspection directory, unified collection processing is performed on information collected by each target system to be inspected, and anomaly detection is performed, and alarm information of a page alarm module is analyzed, including:

Extracting key features from the processed data by using a data mining algorithm, inputting the key features into an abnormality detection model for abnormality detection, and outputting a detection result;

when an abnormality occurs, acquiring alarm information sent by a target system page alarm module, wherein the alarm information at least comprises an alarm code, an alarm level and an alarm description;

the alarm description is preprocessed by using natural language processing technology, key semantic information is extracted, and the alarm codes and the alarm level are combined to locate the problem.

Optionally, in a sixth implementation manner of the present invention, the extracting key features from the processed data by using a data mining algorithm includes:

Calculating the pearson correlation coefficient between each feature in the processed data and the system target operation and maintenance state, and obtaining the feature with the absolute value of the correlation higher than a set threshold value, wherein the threshold value is 0.3;

the variance analysis is adopted to test the mean value difference of each characteristic under different types of operation and maintenance states, and the characteristics under different system health conditions are screened out, so that high-correlation characteristic data are obtained;

The method comprises the steps of taking high-correlation characteristic data as input, importing the high-correlation characteristic data into an LDA model frame, setting target dimensions of an LDA model according to the actual category number of system operation and maintenance, projecting a high-dimensional characteristic space into a low-dimensional discrimination subspace, maximizing inter-category distances of different operation and maintenance state categories in the subspace, and minimizing intra-category variances, wherein the categories of the system operation and maintenance at least comprise normal operation, slight faults and serious faults;

and (3) through iterative optimization of a projection matrix, the different operation and maintenance state samples are divided in a new projected feature space, and key features are extracted.

Optionally, in a seventh implementation manner of the present invention, inputting the key feature into an anomaly detection model to perform anomaly detection, and outputting a detection result includes:

inputting the key features into an anomaly detection model by adopting a bidirectional RNN (RNN) as a basic model of the anomaly detection model, sequentially reading key feature vectors one by a forward RNN unit from a starting time step of a sequence in a forward propagation stage of the model, and calculating hidden layer output at the current moment through an activation function according to the current input features and the output state of a hidden layer at the last moment;

the reverse RNN unit is started from the last time step of the sequence, reversely reads the characteristic vector, calculates and outputs according to the input and the reverse hidden layer state at the last moment, reversely transmits key information at the rear end of the sequence, and fuses the key information with the forward transmitted information;

And outputting a detection result through the anomaly detection model, and transmitting the detection result to the RPA inspection robot in real time through a communication protocol.

According to the technical scheme, the RPA automatic robot carries out login operation on different systems of an enterprise according to the conditions of the enterprise, pages which pay attention to importance and reflect the running states of the systems are automatically intercepted, unified collection processing and storage are carried out, so that operation and maintenance personnel review the pages, the operation and maintenance personnel can replace timing and fixed-point manual login of the systems for carrying out inspection operation, the automatic inspection tool is used for improving the working efficiency, the personnel can be ensured to quickly adapt to and improve the overall working efficiency, inspection screenshot can be carried out on the following systems, operation and maintenance operation is assisted by the operation and maintenance personnel, and the inspection system comprises a whole network behavior management system, an operation and maintenance audit and risk control system, a network controller system, an enterprise level data backup and recovery system, an online behavior management system, an application delivery management system, a firewall system, a cloud platform system and an IT operation and maintenance monitoring platform.

Drawings

Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention.

Fig. 1 is a schematic diagram of an embodiment of an inspection method for performing multi-system operation and maintenance by an RPA inspection robot based on artificial intelligence according to an embodiment of the present invention;

Fig. 2 is a schematic diagram of a logging step of an inspection method for performing multi-system operation and maintenance by an RPA inspection robot based on artificial intelligence according to an embodiment of the present invention.

Detailed Description

The terms "first," "second," "third," "fourth" and the like in the description and in the claims and in the above drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, apparatus, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.

For convenience of understanding, the following describes a specific flow of the embodiment of the present invention, please refer to fig. 1 for a schematic diagram of an embodiment of an inspection method for performing multi-system operation and maintenance by using an RPA inspection robot based on artificial intelligence, which specifically includes the following steps:

step 101, an RPA inspection robot establishes connection with a plurality of target systems to be inspected through a preset interface protocol, and automatically identifies login information of a login system page based on flow automation RPA software;

In this embodiment, the login information at least includes a login user name, a password, and verification code information;

In the embodiment, an RPA tool software package and a corresponding development tool package are loaded and read, an RPA tool and a patrol rule are integrated, an RAP engine is started, engine parameters and an initialization operation instruction are initialized, wherein the engine parameters at least comprise thread pool size, overtime time setting and retry time threshold, the hierarchical structure of a patrol catalog is defined, a storage rule is analyzed, storage formats of different types of data are determined, at least performance index data are stored in a CSV format according to time sequence, a log file is stored in an original text form under a folder named by date, the RPA engine is called, a patrol system website pre-stored in a configuration library is transferred to a browser starting parameter, so that the browser automatically navigates to a login page of a target patrol system after being started, when the browser loads a patrol system login page and encounters a certificate error prompt, the web page certificate error is ignored through a thisisunsafe command, the RPA engine is driven, page elements are automatically identified, whether the loading of the login page elements is judged to be completed or not is judged, when the loading of the page elements is completed, a user name input frame of the login page is automatically input, a password input frame is automatically identified, a password input is automatically, and a login system is identified after the login page is completed is selected.

102, Automatically extracting verification code information in login information, and calling a verification code identification service interface to identify the verification code;

In the embodiment, an RPA browser plug-in running environment is started, a preconfigured page element identification rule set is loaded, an HTML structure of a page is traversed through a DOM analysis technology, nodes where verification code information elements are located, verification code codes are identified, an HTTPS request is sent to a server, the response state of the request is monitored in real time, if a successful response is received, a call token is obtained, picture identification API access information stored in a local configuration file is read through the call token, a returned JSON value is extracted, the JSON value is analyzed to obtain an identification result, the identification result is transmitted to a system login, a verification code input box is automatically filled in the login process, and the whole verification code processing flow is completed.

Step 103, after the verification code is identified, the RPA robot automatically logs in each target system to be inspected, acquires page information at least comprising a target page and system running state according to the targets of operation and maintenance of each system, and stores the collected page information into an inspection catalog of the RPA robot;

In this embodiment, the target system to be inspected at least includes a whole network behavior management system, an operation and maintenance audit and risk control system, a network controller system, an enterprise data backup and recovery system, a network behavior management system, an application delivery management system, a firewall system, a cloud platform system, and an IT operation and maintenance monitoring platform.

In the embodiment, after verification code identification is successful and verification code is automatically filled, an RPA robot is driven to utilize a login account number and password information acquired before to simulate manual login button operation, a login request is sent to a target system server, response information returned by the continuous monitoring server is analyzed, whether login is successful or not is confirmed through a response state code, a page jump identifier and login success prompting elements, if a successful login signal is not detected within 10 seconds, the RPA robot reinitiates a login flow according to a preset retry mechanism until each target system needing to be patrolled is successfully logged, after login is successful, a page acquisition module arranged in the RPA robot is started, memory space is allocated as a temporary data cache area, page acquisition rule files aiming at the target system are read, wherein the page acquisition rules at least comprise acquisition priority levels, acquisition frequencies and data storage formats of different page elements, parameters of the page acquisition modules are initialized according to the page acquisition rules, a built-in analysis engine of the RPA robot is used for carrying out depth analysis on the target page through an HTML tag structure, a style class and a CSS trigger area of an identification page, the RPA robot is positioned to the page jump page information, the page jump information is automatically loaded in the page system, the page information is pre-arranged in the page system, the page information is loaded from the page jump system, the operation condition of the system is loaded, the operation condition of the RPA is completed, the operation condition of the system is finished, the operation condition of the RPA is carried out, and the operation condition of the system is finished, and integrating the data content in the page elements where the key information temporarily stored in the memory buffer area is located and the detailed system running condition data, and storing the integrated data into a patrol directory by using the RPA robot according to a storage path.

In the embodiment, an underlying network request sniffing technology is utilized to capture an AJAX request interacted with a server behind a page, returned data is analyzed, and system performance indexes pushed by the server in real time are obtained, wherein the system performance indexes at least comprise CPU (Central processing Unit) utilization rate, memory occupancy rate and network bandwidth utilization rate, system operation trend data displayed on the page in a visual chart form are converted into a structured numerical sequence by combining an OCR (optical character recognition) technology and a chart analysis algorithm, collected system operation condition data are temporarily stored in a memory cache area, and preliminary arrangement is carried out according to preset data classification rules, wherein the preset data classification rules at least comprise data type classification, time stamp classification and data source classification.

104, According to the inspection catalogue, carrying out unified collection processing on information collected by each target system needing inspection, carrying out anomaly detection, and analyzing the alarm information of the page alarm module;

in the embodiment, key features are extracted from the processed data by using a data mining algorithm, the key features are input into an anomaly detection model to detect anomalies and output detection results, when anomalies occur, alarm information sent by a target system page alarm module is obtained, wherein the alarm information at least comprises alarm codes, alarm levels and alarm descriptions, the alarm descriptions are preprocessed by using a natural language processing technology, key semantic information is extracted, and the alarm codes and the alarm levels are combined to locate problems.

In the embodiment, the pearson correlation coefficient between each feature in the processed data and the target operation and maintenance state of the system is calculated, the feature with the absolute value of the correlation higher than the set threshold value is obtained, the threshold value is 0.3, the mean value difference of the features in the operation and maintenance states of different types is checked by adopting analysis of variance, the features in the health condition of the different types are screened out to obtain high-correlation feature data, the high-correlation feature data are used as input and are imported into an LDA model frame, the target dimension of the LDA model is set according to the actual number of the operation and maintenance of the system, the high-dimensional feature space is projected into a low-dimensional judging subspace, the inter-class distance between the operation and maintenance states of the different types in the subspace is maximized, the intra-class variance is minimized, the operation and maintenance states of the system at least comprise normal operation, slight faults and serious faults, and the projection matrix is optimized through iteration, so that the samples of the different operation and maintenance states are divided in the new feature space after projection, and key features are extracted.

In the embodiment, a bidirectional RNN is adopted as a basic model of an anomaly detection model, key features are input into the anomaly detection model, key feature vectors are sequentially read one by a forward RNN unit from the beginning time step of a sequence in the forward propagation stage of the model, hidden layer output at the current moment is calculated through an activation function according to the current input features and the output state of a hidden layer at the last moment, a reverse RNN unit is started from the last time step of the sequence, feature vectors are reversely read, the calculation output is also carried out according to the input and the state of a reverse hidden layer at the last moment, key information at the rear end of the sequence is reversely transmitted and fused with forward transmitted information, a detection result is output by the anomaly detection model, and the detection result is transmitted to an RPA inspection robot in real time through a communication protocol.

And 105, intercepting an alarm information page, storing the alarm information page into the RPA robot, and sending alarm information to an operation and maintenance person by the RPA robot.

Another embodiment of an inspection method for performing multi-system operation and maintenance by an RPA inspection robot based on artificial intelligence includes the following steps:

The system login comprises the steps of automatically identifying login information of a login system page based on flow automation software, wherein the login information comprises login user name, password and verification code information;

identifying verification codes, namely automatically extracting page verification code pictures based on machine flow automation, and calling a universal verification code identification service interface to identify the verification codes;

Logging in each inspection system, capturing the concerned pages, reflecting the running state of the system and other page information, and storing the page information in a specified inspection catalog;

the data collection is to perform unified processing on the inspection data collected by a plurality of different systems;

And (3) after logging in the system, automatically analyzing the alarm information of the page alarm module, intercepting an alarm information page, storing the alarm information page in a specified inspection catalog, and storing pictures through a screenshot naming rule, so that operation and maintenance personnel can confirm the alarm information of the system through file names.

Referring to fig. 2, a schematic diagram of a logging step of an inspection method for performing multi-system operation and maintenance by an RPA inspection robot based on artificial intelligence includes the following steps:

Integrating and packaging an RPA engine, namely integrating an RPA tool and a patrol rule to realize automation of a patrol logic flow;

initializing an RPA engine, namely starting the RAP engine, initializing engine parameters and an initializing method;

initializing parameters of a patrol system, namely defining a patrol catalog and a storage rule;

The RPA calls a browser to open a patrol website, namely calling an RPA engine, opening a default browser of the system, and automatically inputting a VPN system website;

the method solves the problem of certificate errors, namely, using a command 'thisisunsafe', ignoring the certificate errors, skipping a page 'the security certificate of the website has a problem', and entering a VPN login page.

The RPA calls a browser check-in system, namely, opening a check-in website, calling an RPA engine, automatically identifying page elements through browser plug-ins, judging whether the page elements are loaded or not through judging whether the page elements are loaded, and when the loading of the page elements is completed, automatically inputting a user name by a user name input box for identifying a login page, automatically inputting a password by a password input box for identifying the login page, and automatically hooking an agreement selection box for identifying the login page, and then completing the login of the VPN system by an identification login button.

And then according to the difference of attention information and key points of different systems, the method and the steps of each system in the process of inspection are as follows:

1. inspection of the whole network behavior management system:

Calling an RPA engine, opening a default browser of the system, and automatically inputting the website of the whole network behavior management system;

the certificate error is solved by using thisisunsafe command, ignoring the certificate error, skipping the page of the website where the security certificate has problems, and entering the login page of the whole network behavior management system;

Calling an RPA engine, automatically identifying page elements through browser plug-ins, judging whether the page is loaded or not by judging whether the page elements are loaded, and when the loading of the page elements is completed, automatically inputting a user name by a user name input box for identifying the login page, automatically inputting a password by a password input box for identifying the login page, and automatically hooking an agreement selection box for identifying the login page, and then completing the login of the whole network behavior management system by an identification login button.

And intercepting the home page screen information, namely calling an RPA engine, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist or not, intercepting the home page screen information when the page elements are loaded completely, and storing the screen capturing picture into a patrol catalog. The screenshot information should include information reflecting network behavior such as device status icon information, management status information, traffic analysis-interface throughput line graph, user traffic ranking, application traffic ranking, etc.

2. And (3) inspection of the operation and maintenance audit and risk control system:

The operation and maintenance audit and risk control system is opened by using a default browser, namely, an RPA engine is called, the default browser of the system is opened, and the website of the operation and maintenance audit and risk control system is automatically input;

The certificate error is solved by using thisisunsafe command, ignoring the certificate error, skipping the page of the website with the problem of the security certificate, and entering the operation and maintenance audit and risk control system login page;

Calling an RPA engine, automatically identifying page elements through browser plug-ins, judging whether the page is loaded or not through judging whether the page elements are loaded, automatically inputting a user name by a user name input box for identifying the login page, automatically inputting a password by a password input box for identifying the login page, automatically marking an agreeing selection box for identifying the login page, automatically analyzing an identifying result of the identifying code by calling a hundred-degree picture identifying API, automatically filling the identifying code into the identifying code input box, finally identifying a login button, and completing operation and maintenance audit and risk control system login.

And intercepting the screen information of the main interface, namely calling an RPA engine, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist or not, intercepting the screen information of the main interface when the loading of the page elements is completed, and storing the screen capturing picture into a patrol catalog. The screenshot information comprises operation and maintenance times statistics charts, operation and maintenance statistics information, real-time monitoring information, system running states, licenses and the like.

And intercepting network configuration screen information, namely calling an RPA engine, automatically inputting a network configuration page website, opening the network configuration page, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist, intercepting the network configuration screen information when the page elements are loaded completely, and storing screen capturing pictures into a patrol catalog. The screen shot information should contain detailed interface information list and the like.

3. NAC network controller system patrol:

The network controller system is opened by using a default browser, namely, an RPA engine is called, the default browser of the system is opened, and the network address of the network controller system is automatically input;

The certificate error is solved by using thisisunsafe command, ignoring the certificate error, skipping the page of the website with the problem of the security certificate, and entering the operation and maintenance audit and risk control system login page;

The network controller system logs in, namely calling an RPA engine, automatically identifying page elements through browser plug-ins, judging whether the page is loaded or not through judging whether the page elements are loaded, automatically inputting a user name by a user name input box for identifying the login page, automatically inputting a password by a password input box for identifying the login page, automatically hooking an agreement selection box for identifying the login page, and finally identifying a login button to complete the network controller system login.

And intercepting the screen information of the main interface, namely calling an RPA engine, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist or not, intercepting the screen information of the main interface when the loading of the page elements is completed, and storing the screen capturing picture into a patrol catalog. The screen shot information should contain information such as equipment resources, system state, network port state, access point state, total throughput flow chart and the like.

Intercepting the state screen information of the address pool by calling an RPA engine, automatically inputting the address of the state page of the address pool, opening the state page of the address pool, automatically identifying page elements, judging whether the page is loaded or not by judging whether the page elements are loaded or not, intercepting the state screen information of the address pool when the page elements are loaded, and storing the screen capturing picture into a patrol catalog. The screen shot information should contain detailed information such as a wireless address pool status chart.

4. And (3) backing up enterprise-level data and recovering the system to carry out inspection:

the system is opened by using a default browser, namely, an RPA engine is called, the default browser of the system is opened, and the website of the enterprise-level data backup and recovery system is automatically input;

The certificate error is solved by using thisisunsafe command, ignoring the certificate error, skipping the page of the website with the problem of the security certificate, entering the enterprise data backup and recovery system login page;

Calling an RPA engine, automatically identifying page elements by browsing browser plug-ins, judging whether the page is loaded or not by judging whether the page elements are loaded, automatically inputting a user name by a user name input box for identifying the login page, automatically inputting a password by a password input box for identifying the login page, automatically typing a password by an agreement selection box for identifying the login page, and finally identifying a login button to complete enterprise-level data backup and system login.

And intercepting the screen information of the main interface, namely calling an RPA engine, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist or not, intercepting the screen information of the main interface when the loading of the page elements is completed, and storing the screen capturing picture into a patrol catalog. The screenshot information should include information such as system time, accumulated run time, accumulated protection data, virtualization center, backup storage, backup node graph, network traffic graph, current task list, etc.

And the warning information reminding step of calling an RPA engine, automatically analyzing the node content of the page warning information, naming the main screen screenshot information as a picture name containing warning when the warning is found in the system, and logging in the system for checking the system warning when the operation and maintenance personnel find the picture containing the warning file name.

5. And (3) tour inspection of a vpn internet surfing behavior management system:

The Internet surfing behavior management system is opened by using a default browser, an RPA engine is called, the default browser of the system is opened, and the website of the Internet surfing behavior management system is automatically input;

The certificate error is solved by using thisisunsafe command, ignoring the certificate error, skipping the page of the website with the problem of the security certificate, and entering the log-in page of the internet behavior management system;

the login of the internet behavior management system is completed by calling an RPA engine, automatically identifying page elements by browsing browser plug-ins, judging whether the page is loaded or not by judging whether the page elements are loaded, automatically inputting a user name by a user name input box for identifying the login page, automatically inputting a password by a password input box for identifying the login page, automatically hooking an agreement selection box for identifying the login page, and finally identifying a login button.

And intercepting the screen information of the main interface, namely calling an RPA engine, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist or not, intercepting the screen information of the main interface when the loading of the page elements is completed, and storing the screen capturing picture into a patrol catalog. The screenshot information should include information such as a system information line graph, a real-time network throughput line graph, a real-time concurrent session line graph, a line status list, a real-time user concurrent trend line graph, a real-time stream buffer status line graph, and the like.

6. And (5) inspection by an application delivery management system:

the application delivery management system is opened by using a default browser, an RPA engine is called, the default browser of the system is opened, and the website of the application delivery management system is automatically input;

The certificate error is solved by using thisisunsafe command, ignoring the certificate error, skipping the page of the website with the problem of the security certificate, and entering the login page of the application delivery management system;

Calling an RPA engine, automatically identifying page elements by browsing browser plug-ins, judging whether the page is loaded completely by judging whether the page elements are loaded completely, automatically inputting a user name by a user name input box for identifying the login page, automatically inputting a password by a password input box for identifying the login page, automatically hooking an agreement selection box for identifying the login page, and finally identifying a login button to complete the login of the application delivery management system.

And intercepting application load screen information, namely calling an RPA engine, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist or not, intercepting the application load screen information when the page elements are loaded completely, and storing screen capturing pictures into a patrol catalog. The screenshot information comprises information such as an application load overall situation chart, equipment operation information, a CPU, a memory, a power supply, a temperature, a fan, a system throughput line graph, a system newly-built connection number line graph, a system concurrent connection line graph and the like.

And intercepting the link load screen information by calling an RPA engine, automatically inputting a link load page website, opening the link load page, automatically identifying page elements, judging whether the page is loaded or not by judging whether the page elements exist, intercepting the link load screen information when the page elements are loaded, and storing the screen capturing picture into a patrol catalog. The screen shot information comprises link state, link load real-time state line graph, stability monitoring line graph, application flow rate ratio analysis column graph, system throughput line graph, system newly-built connection line graph, system concurrent connection line graph and the like.

7. And (3) inspecting by a firewall system:

the default browser is used for opening a firewall system, namely an RPA engine is called, the default browser of the system is opened, and the website of the firewall system is automatically input;

The certificate error is solved by using a command thisisunsafe, ignoring the certificate error, skipping a page of 'the security certificate of the website has a problem', and entering a firewall system login page;

Calling an RPA engine, automatically identifying page elements by browsing browser plug-ins, judging whether the page is loaded or not by judging whether the page elements are loaded, automatically inputting a user name by a user name input box for identifying the login page, automatically inputting a password by a password input box for identifying the login page, automatically hooking an agreement selection box for identifying the login page, and finally identifying a login button when the page elements are loaded, thereby completing the login of the firewall system.

And intercepting the home page screen information, namely calling an RPA engine, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist or not, intercepting the home page screen information when the page elements are loaded completely, and storing the screen capturing picture into a patrol catalog. The screenshot information should include information such as business security-risk analysis, attack trends, hotspot event graph information, user security-risk analysis, attack trends, hotspot event graph information, device and system operation-device status, interface status, security capability information, network operation-concurrent sessions, interface throughput, etc.

And intercepting network interface screen information, namely calling an RPA engine, automatically inputting a network interface page website, opening a network interface page, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist, intercepting the network interface screen information when the page elements are loaded completely, and storing screen capturing pictures into a patrol catalog. The screen shot information should contain information such as a status list of physical interface details.

8. Cloud platform system inspection:

the cloud platform system is opened by using a default browser, namely, an RPA engine is called, the default browser of the system is opened, and the website of the cloud platform system is automatically input;

the certificate error is solved by using a command thisisunsafe, ignoring the certificate error, skipping a page of 'the security certificate of the website has a problem', and entering a cloud platform system login page;

Calling an RPA engine, automatically identifying page elements by browsing browser plug-ins, judging whether the page is loaded or not by judging whether the page elements are loaded, automatically inputting a user name by a user name input box for identifying the login page, automatically inputting a password by a password input box for identifying the login page, automatically hooking an agreement selection box for identifying the login page, and finally identifying a login button when the page elements are loaded, thereby completing the login of the cloud platform system.

And intercepting the home page screen information, namely calling an RPA engine, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist or not, intercepting the home page screen information when the page elements are loaded completely, and storing the screen capturing picture into a patrol catalog. The screenshot information comprises the information of the whole situation of cluster resource scheduling load balancing, entity machines and storage states, namely a CPU, a physical memory, a detailed list of configuration memories and the like.

Intercepting application delivery screen information, namely calling an RPA engine, automatically inputting an application delivery page website, opening the application delivery page, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist, intercepting the application delivery screen information when the page elements are loaded completely, and storing screen capturing pictures into a patrol catalog. The screenshot information should include information such as a device status list, a session trend graph, a connection status list, a network status trend graph, etc.

And intercepting the WEB management screen information by calling an RPA engine, automatically inputting a WEB address of a WEB management page, opening the WEB management page, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist, intercepting the WEB management screen information when the page elements are loaded completely, and storing the screen capturing picture into a patrol catalog. The screenshot information comprises information such as an application load overall situation chart, equipment operation information, a CPU, a memory, a power supply, a temperature, a fan, a system throughput line graph, a system newly-built connection number line graph, a system concurrent connection line graph and the like.

And intercepting firewall screen information, namely calling an RPA engine, automatically inputting a firewall page website, opening the firewall page, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist, intercepting the firewall screen information when the page elements are loaded completely, and storing the screen capturing picture into a patrol catalog. The screenshot information should include information such as a device status list, a session trend graph, a connection status list, a network status trend graph, etc.

9. IT operation and maintenance monitoring platform inspection:

The IT operation and maintenance monitoring platform is opened by using a default browser, namely, an RPA engine is called, a default browser of a system is opened, and the website of the IT operation and maintenance monitoring platform is automatically input;

The certificate error is solved by using thisisunsafe command, ignoring the certificate error, skipping the page of the website with the problem of the security certificate, and entering the login page of the IT operation and maintenance monitoring platform;

And calling an RPA engine, automatically identifying page elements by browsing browser plug-ins, judging whether the page is loaded or not by judging whether the page elements are loaded, automatically inputting a user name by a user name input box for identifying the login page, automatically inputting a password by a password input box for identifying the login page, automatically typing a password by an agreement selection box for identifying the login page, and finally identifying a login button when the page elements are loaded, thereby completing the login of the cloud platform system.

And intercepting the screen information of the instrument panel, namely calling an RPA engine, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist or not, intercepting the screen information of the instrument panel when the loading of the page elements is completed, and storing the screen capturing picture into a patrol catalog. The screenshot information should include summary information chart-SLA status, monitor status, alarm status, device status list, etc.

And intercepting the screen information of the core switch by calling an RPA engine, automatically inputting a page website of the core switch, opening the page of the core switch, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist, intercepting the screen information of the core switch when the page elements are loaded completely, and storing the screen capturing picture into a patrol catalog. The screen shot information should contain information such as core exchange name, network port, memory Pool, IP address, CPU, file list, etc.

And intercepting the screen information of the main exchanger by calling an RPA engine, automatically inputting a page website of the main exchanger, opening the page of the main exchanger, automatically identifying page elements, judging whether the page is loaded completely by judging whether the page elements exist, intercepting the screen information of the main exchanger when the page elements are loaded completely, and storing the screen capturing picture into a patrol catalog. The screen shot information should contain list information of main exchange name, main exchange location, network port, fan, stack module, module humidity, power supply, IP address, CPU, memory, etc.

By implementing the scheme, the inspection task is automatically executed, a large amount of time can be saved compared with manual operation, the probability of missed inspection and false inspection caused by human negligence is reduced, the system fine abnormality such as sporadic error information hidden in a system log can be timely found, real-time early warning provides comprehensive health condition insight of the system for operation and maintenance personnel, the problem root of the assistance accurate positioning problem is formulated, and the business continuity of enterprises is ensured.

The foregoing has shown and described the basic principles, principal features and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the above-described embodiments, and that the above-described embodiments and descriptions are only preferred embodiments of the present invention, and are not intended to limit the invention, and that various changes and modifications may be made therein without departing from the spirit and scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.

Claims (7)

1. The method for carrying out multi-system operation and maintenance on the RPA inspection robot based on artificial intelligence is characterized by comprising the following steps:

The RPA inspection robot establishes connection with a plurality of target systems needing inspection through a preset interface protocol, automatically identifies login information of a login system page based on flow automation RPA software, and comprises loading and reading an RPA tool software package and a development tool package corresponding to the RPA tool software package, integrating the RPA tool with an inspection rule, starting an RAP engine, initializing engine parameters and an initialization operation instruction, wherein the engine parameters at least comprise thread pool size, timeout time setting and retry time threshold, defining a hierarchical structure of an inspection directory, analyzing a storage rule, determining storage formats of different types of data, at least comprising performance index data stored in a CSV format in time sequence, and a log file stored in a file folder named by date, calling the RPA engine, transmitting a login system website pre-stored in a configuration library to a browser starting parameter, enabling the browser to automatically navigate to a login page of the target inspection system after being started, and automatically checking up the login page when the browser loads the inspection system, and meets a certificate error prompt, automatically inputting a login page through thisisunsafe command webpage certificate, automatically judging whether a login page identification element is stored in the login system page, automatically judges whether a login password is input by a password, a login frame identification element is input, and a login password is input by a password is input after a password is input by a password, and a user is input by at least a password is completely selected by a user name;

Automatically extracting verification code information in login information, and calling a verification code identification service interface to identify the verification code;

After the verification code is identified, the RPA robot automatically logs in each target system to be inspected, acquires page information at least comprising a target page and system running state page information according to the operation and maintenance targets of each system, and stores the collected page information into an inspection catalog of the RPA robot, wherein the target system to be inspected at least comprises a whole network behavior management system, an operation and maintenance audit and risk control system, a network controller system, an enterprise data backup and recovery system, an internet behavior management system, an application delivery management system, a firewall system, a cloud platform system and an IT operation and maintenance monitoring platform;

According to the inspection catalogue, carrying out unified collection processing on information collected by each target system needing inspection, carrying out anomaly detection, and analyzing alarm information of a target system page alarm module;

And intercepting an alarm information page, storing the alarm information page into the RPA robot, and sending alarm information to an operation and maintenance person by the RPA robot.

2. The inspection method for multi-system operation and maintenance of an RPA inspection robot based on artificial intelligence according to claim 1, wherein the steps of automatically extracting verification code information from login information and calling a verification code identification service interface to identify verification codes include:

Starting an RPA browser plug-in running environment, loading a preconfigured page element identification rule set, traversing the HTML structure of the page through a DOM analysis technology, positioning the node where the verification code information element is positioned, and identifying the verification code;

Transmitting an HTTPS request to a server, monitoring the response state of the request in real time, if a successful response is received, acquiring a call token, and reading the picture identification API access information stored in the local configuration file through the call token;

And extracting the returned JSON value, analyzing the JSON value to obtain an identification result, transmitting the identification result to a system for login, automatically filling a verification code input box in the login process, and completing the whole verification code processing flow.

3. The method for carrying out multi-system operation and maintenance on an RPA inspection robot based on artificial intelligence according to claim 1, wherein after the verification code is identified, the RPA robot automatically logs in each target system to be inspected, obtains at least information including a target page and system operation status page according to the targets of each system operation and maintenance, and stores the collected page information in an inspection catalog of the RPA robot, comprising:

after the verification code is successfully identified and the verification code is automatically filled, driving the RPA robot to simulate the operation of manually clicking a login button by utilizing the login account and password information acquired before, and sending a login request to a target system server;

Continuously monitoring response information returned by the server, confirming whether the login is successful or not by analyzing the response state code, the page jump identifier and the login success prompt element, and if the successful login signal is not detected within 10 seconds, restarting a login process by the RPA robot according to a preset retry mechanism until each target system needing to be inspected is successfully logged in;

After successful login, starting a built-in page acquisition module of the RPA robot, and allocating a memory space as a temporary data buffer area to read a page acquisition rule file aiming at the target system, wherein the page acquisition rule at least comprises acquisition priorities, acquisition frequencies and data storage formats of different page elements;

initializing parameters of a page acquisition module according to a page acquisition rule, performing depth analysis on a target page by using a built-in DOM analysis engine through an RPA robot, and positioning page elements where key information is located by identifying an HTML tag structure, a CSS style class name and a JavaScript event trigger area of the page;

The RPA robot automatically extracts the data content in the page element where the key information is located, and preprocesses the data content;

Driving the RPA robot to jump from the current page to the system running state page according to the URL address of the system running state page or the page jump path configuration information, and excavating detailed system running state data in the page by the RPA robot after the loading of the system running state page is completed;

And integrating the data content in the page element where the key information temporarily stored in the memory cache area is located and the detailed data of the system running state, and storing the integrated data into the patrol directory by using the RPA robot according to the storage path.

4. The inspection method for performing multi-system operation and maintenance by using an RPA inspection robot based on artificial intelligence as claimed in claim 3, wherein the RPA robot digs detailed system operation status data in a page, and the method comprises the following steps:

Capturing an AJAX request interacted with a server behind a page by using a network request sniffing technology of a bottom layer, analyzing returned data, and acquiring a system performance index pushed by a server in real time, wherein the system performance index at least comprises a CPU (central processing unit) utilization rate, a memory occupancy rate and a network bandwidth utilization rate;

for system operation trend data displayed in a visual chart form on a page, converting data points in the chart into a structured numerical sequence by combining an OCR technology and a chart analysis algorithm;

The collected system running state data is temporarily stored in a memory buffer area, and preliminary arrangement is carried out according to preset data classification rules, wherein the preset data classification rules at least comprise data type classification, time stamp classification and data source classification.

5. The inspection method for carrying out multi-system operation and maintenance on the RPA inspection robot based on artificial intelligence according to claim 1, wherein the method for carrying out unified collection processing on information collected by each target system to be inspected and carrying out anomaly detection according to the inspection catalog and analyzing the alarm information of the page alarm module comprises the following steps:

Extracting key features from the processed data by using a data mining algorithm, inputting the key features into an abnormality detection model for abnormality detection, and outputting a detection result;

when an abnormality occurs, acquiring alarm information sent by a target system page alarm module, wherein the alarm information at least comprises an alarm code, an alarm level and an alarm description;

the alarm description is preprocessed by using natural language processing technology, key semantic information is extracted, and the alarm codes and the alarm level are combined to locate the problem.

6. The inspection method for performing multi-system operation and maintenance on an RPA inspection robot based on artificial intelligence as set forth in claim 5, wherein the extracting key features from the processed data by using a data mining algorithm comprises:

Calculating the pearson correlation coefficient between each feature in the processed data and the system target operation and maintenance state, and obtaining the feature with the absolute value of the correlation higher than a set threshold value, wherein the threshold value is 0.3;

the variance analysis is adopted to test the mean value difference of each characteristic under different types of operation and maintenance states, and the characteristics under different system health conditions are screened out, so that high-correlation characteristic data are obtained;

The method comprises the steps of taking high-correlation characteristic data as input, importing the high-correlation characteristic data into an LDA model frame, setting target dimensions of an LDA model according to the actual category number of system operation and maintenance, projecting a high-dimensional characteristic space into a low-dimensional discrimination subspace, maximizing inter-category distances of different operation and maintenance state categories in the subspace, and minimizing intra-category variances, wherein the categories of the system operation and maintenance at least comprise normal operation, slight faults and serious faults;

and (3) through iterative optimization of a projection matrix, the different operation and maintenance state samples are divided in a new projected feature space, and key features are extracted.

7. The inspection method for performing multi-system operation and maintenance by an RPA inspection robot based on artificial intelligence as set forth in claim 5, wherein inputting the key features into an anomaly detection model for anomaly detection and outputting a detection result comprises:

inputting the key features into an anomaly detection model by adopting a bidirectional RNN (RNN) as a basic model of the anomaly detection model, sequentially reading key feature vectors one by a forward RNN unit from a starting time step of a sequence in a forward propagation stage of the model, and calculating hidden layer output at the current moment through an activation function according to the current input features and the output state of a hidden layer at the last moment;

the reverse RNN unit is started from the last time step of the sequence, reversely reads the characteristic vector, calculates and outputs according to the input and the reverse hidden layer state at the last moment, reversely transmits key information at the rear end of the sequence, and fuses the key information with the forward transmitted information;

And outputting a detection result through the anomaly detection model, and transmitting the detection result to the RPA inspection robot in real time through a communication protocol.