CN118536638B

CN118536638B - Power optimization method and system based on intelligent data processing analysis and digital twin

Info

Publication number: CN118536638B
Application number: CN202410413517.7A
Authority: CN
Inventors: 朱国伟; 杨梅初; 李超; 刘思锐; 赖吉丰; 陈满圆
Original assignee: Jinggangshan Power Plant of Huaneng Power International Inc
Current assignee: Jinggangshan Power Plant of Huaneng Power International Inc
Priority date: 2024-04-08
Filing date: 2024-04-08
Publication date: 2025-03-18
Anticipated expiration: 2044-04-08
Also published as: CN118536638A

Abstract

The invention discloses an electric power optimization method and system based on intelligent data processing analysis and digital twin, and relates to the technical field of electric power system optimization, comprising the steps of collecting electric power system data and preprocessing; extracting features of the preprocessed data, identifying problems in the power system, classifying the problems, and matching solutions; a virtual model of the power system is built, and the priority of the solution is verified. According to the invention, the EMD algorithm is used for analyzing the data, so that the problems in the power system can be accurately identified and effectively classified. Improving the efficiency of problem diagnosis and providing reliable basis for selecting and matching the most suitable solution. By constructing a digital twin model of the power system, the invention enables the solution to be fully tested and validated prior to actual application. This step significantly improves the safety and reliability of the decision, avoiding potential risks and costs.

Description

Electric power optimization method and system based on intelligent data processing analysis and digital twin

Technical Field

The invention relates to the technical field of power system optimization, in particular to a power optimization method and system based on intelligent data processing analysis and digital twin.

Background

Under the current technical background, the optimal management of a power system is always a key research field, especially under the background that the global energy demand continuously grows and renewable energy resources are gradually integrated into a power grid. Traditional power system management relies on empirical rules and a simplified physical model for decision support, which has demonstrated its effectiveness in the past decades. However, as power systems become more complex, integration of emerging technologies including distributed generation, demand response, electric vehicle charging networks, etc., traditional approaches face significant challenges. To address these challenges, intelligent data processing and analysis techniques, including machine learning, big data analysis, and artificial intelligence, have been introduced in recent years into the research and management of power systems. These techniques enable processing and analysis of large-scale, high-dimensional datasets, providing more accurate and dynamic decision support.

Nevertheless, the existing related art still has some drawbacks in power system optimization. First, most of the prior art mainly focuses on optimization in a single dimension, such as cost minimization, power generation efficiency improvement or carbon emission reduction, and few methods can comprehensively consider multiple optimization targets and constraints of an electric power system. Second, while intelligent data processing provides a new perspective for power system management, existing approaches often ignore the physical nature and operational constraints of the power system, resulting in insufficient feasibility and security of the optimization results in practical applications. In addition, as a typical dynamic system, the state of the power system is continuously changed along with time, and the prior art often lacks an effective dynamic optimization strategy and cannot adapt to the real-time change of the state of the system. Finally, while digital twinning provides an innovative tool for simulating and predicting power systems, its use in power system optimization is still in the launch phase, lacking systematic research and practical verification.

In order to overcome the defects of the prior art, the invention provides a comprehensive solution for the intelligent data processing analysis and digital twin-based power optimization method and system. The method not only utilizes intelligent data processing technology to carry out deep analysis on the data of the power system to realize accurate classification of problems and effective matching of solutions, but also comprehensively considers the physical characteristics, operation constraint and dynamic change of the system by constructing a digital twin model of the power system to realize priority verification and dynamic optimization of the solutions. The method is hopeful to comprehensively improve the operation efficiency and the safety of the power system, realize multi-objective and dynamic system optimization and provide more scientific, accurate and flexible decision support for power system management. By combining intelligent data processing and digital twin technology, my invention can not only overcome the limitations of the prior art, but also provide an innovative optimization management strategy for future development of the power system.

Disclosure of Invention

The present invention has been made in view of the above-described problems.

Therefore, the invention solves the technical problems of one-sided attention to a single optimization target, neglecting the physical essence and safety of the system, lacking dynamic adaptability and the like in the prior art.

In order to solve the technical problems, the invention provides the following technical scheme that the power optimization method based on intelligent data processing analysis and digital twin comprises the following steps,

And collecting power system data and preprocessing.

And extracting features of the preprocessed data, identifying problems in the power system, classifying the problems, and matching solutions.

A virtual model of the power system is built, and the priority of the solution is verified.

As a preferable scheme of the intelligent data processing analysis and digital twin-based power optimization method, the invention comprises the following steps: the power system data comprises power generation data, transmission data, distribution and consumption data and equipment state data;

the preprocessing is to remove invalid, abnormal and repeated data records from the collected power system data, perform standardized processing on the data, and perform proper resampling or time window division on the time-sensitive data.

The invention relates to an intelligent data processing analysis and digital twin-based power optimization method, which is characterized in that the feature extraction of the preprocessed data is to model a power system into a graph model, nodes represent various entities in the power system, edges represent power connection or flow direction, and a feature extraction function is constructed, wherein the expression is as follows:

Wherein X _i is a feature vector of the node i extracted by integrating time series data and a change rate thereof, α, β and γ are coefficients contributed by the time series data, the first derivative and the self-defined nonlinear function ψ, Θ represents parameters of the nonlinear function ψ, S _i (t) is a power data sequence of the node i at time t, and IMF _n(S_i (t)) is an nth eigenmode function.

The IMF _n(S_i (t) is solved by using an EMD algorithm, which comprises the steps of detecting the preset power system time sequence data S _i (t) by using an extremum, identifying all local maximum value and minimum value points by using a differential and smoothing processing method, obtaining an upper envelope and a lower envelope by adopting an adaptive interpolation method, further calculating an envelope mean value, introducing a method for dynamically adjusting local energy and frequency characteristics based on signals to calculate the mean value, subtracting the envelope mean value from the original data S _i (t), introducing a dynamic filtering mechanism based on data characteristics to output a signal level, and judging whether the detail d (t) meets IMF conditions.

When the number of the maximum value points and the number of the minimum value points are unequal and the difference value is larger than 1, the fluctuation of the current signal is asymmetric, and the differential and smoothing processing method is further adjusted to decompose the signal.

When the number of the maximum value points and the number of the minimum value points are equal or the difference value is equal to 1, further judging whether the average value of the upper envelope and the lower envelope of the detail d (t) at any point is 0, if not, judging that the detail d (t) is unbalanced fluctuation and cannot be represented as a complete IMF, further performing deep analysis on the original data or the detail d (t), if the average value of the upper envelope and the lower envelope of the detail d (t) at any point is 0, judging that the detail d (t) is a complete IMF, further extracting the complete IMF from the original signal, subtracting the extracted IMFS _i(t)＝S_i(t)-IMF_n(S_i (t) from the original data, and repeating the EMD algorithm until a new IMF cannot be extracted from the rest data.

The differential and smoothing method has the expression:

D(t)=S_i(t+1)-S_i(t)

S _i (t) is the time series data of the ith node in the power system at time t, the first-order difference of the original time series data of D (t), the size of a k smoothing window, and the difference sequence of D' (t) after smoothing.

The extreme point is determined by checking the sign change of the smoothed differential sequence, and if D '(t-1) <0 and D' (t+1) >0, S _i (t) is the local minimum point, and if D '(t-1) >0 and D' (t+1) <0, S _i (t) is the local maximum point.

The signal decomposition method of the adjustment difference and smoothing processing is to increase or decrease the smoothing window size k of the difference sequence, when the noise of the original sequence is higher than a threshold value, the k is increased to output a smoother difference sequence, and when the change speed of the signal is higher than the threshold value, the k is decreased.

The invention relates to an optimal scheme of an intelligent data processing analysis and digital twin-based power optimization method, wherein the problem classification in an identified power system is to construct a classification function according to the output result of a feature extraction function, and the classification function is expressed as follows:

Y_i＝C(X_i,W)

wherein, the weight matrix in the W classification function C, Adjacent node sets of nodes i and j respectively,The normalized degrees of nodes i and j, respectively.

The invention relates to an optimal scheme of an intelligent data processing analysis and digital twin-based power optimization method, wherein the matching solution is to construct a matching model, and the expression is as follows:

S_i＝M(Y_i)

Wherein Y _i represents a state class obtained by classifying the node i according to the feature vector, S _i is a solution policy set matched with the problem of the node i, and M is a rule mapping.

And determining a state category Y _i of the node i according to the probability distribution, selecting the category with the highest probability as the final state of the node i, and determining a corresponding solution strategy set S _i through a rule mapping M.

The power optimization method based on intelligent data processing analysis and digital twin is an optimal scheme, wherein the virtual model for constructing the power system comprises a load prediction model, a power system cost model, a stability evaluation model, a safety and reliability evaluation model.

The load prediction model has the expression:

Wherein, The predicted load at time t is represented by L _t-1,L_t-2,…,L_t-n, the historical load data, and the model parameters are represented by Θ.

The power system cost model has the expression:

Wherein c _i is the unit cost of the i-th power generation unit, P _g,i is the power generation amount, lambda is the penalty coefficient of the load prediction error, L _j and The actual and predicted loads, respectively.

The stability evaluation model has the expression:

Where S is the stability index, Δω (T) is the system frequency deviation, ω _max is the allowed maximum of the frequency deviation, α is the attenuation coefficient, and T is the evaluation duration.

The safety and reliability evaluation model has the expression:

Where R is a system reliability index, p _f,i is the probability of failure of the ith element, n _k is the number of elements involved in the kth failure mode, and σ _k is the system performance penalty ratio caused by the kth failure mode.

The invention relates to an optimal scheme of an intelligent data processing analysis and digital twin-based power optimization method, wherein the priority of the verification solution is that a comprehensive scoring model is built, the priority of the solution is determined according to the scoring, and the expression is as follows:

E_k＝ω_aA_k+ω_bB_k+ω_cC_k+ω_dD_k

Wherein L _j is the actual load, Is the predicted load of solution S _k, M is the total number of time points in the evaluation period, P _g,i,k is the generated energy of the ith generation unit in solution S _k, c _i the unit cost of the ith generation unit, λ is the penalty factor for the load prediction error, Δω _k (t) is the systematic frequency deviation caused by solution S _k, the allowed maximum value of the ω _max frequency deviation, n _k is the number of elements involved in the kth failure mode, the probability of failure of the ith element of P _f,i, a _k、B_k、C_k、D_k is the load prediction accuracy score, the cost optimization efficiency score, the stability score, the safety and reliability score of solution Sk, respectively, and E _k is the composite score.

Another object of the present invention is to provide an intelligent data processing analysis and digital twin-based power optimization method system, which can implement accurate load prediction and problem classification by deep integration and analysis of power system data, and perform omnibearing simulation and verification of a solution in cooperation with a digital twin model. By the method, the problems of multi-objective optimization, dynamic adaptation and safety verification of the power system in the prior art are solved.

In order to solve the technical problems, the invention provides a power optimization method system based on intelligent data processing analysis and digital twin, which comprises a data collection and preprocessing module, a feature extraction and classification module, a solution matching and priority assessment module, a virtual model construction and verification module and a comprehensive scoring module.

The data collection and preprocessing module is used for collecting data from the power system, preprocessing the data, removing invalid, abnormal and repeated data records, standardizing the data, and properly resampling or dividing time windows of time sensitive data.

The feature extraction and classification module extracts key features from the preprocessed data by using a feature extraction function, and classifies problems in the power system by using a classification function.

The solution matching and priority rating module is used for constructing a matching model, matching proper solutions for various problems identified in the power system, and rating the priorities of the solutions through a comprehensive rating model.

The virtual model construction and verification module is used for constructing a virtual model of the power system, comprising a load prediction model, a cost model, a stability and safety evaluation model, and verifying the effects of different solutions through the virtual model under the condition of not interfering the operation of the actual power system.

The comprehensive scoring module is used for predicting accuracy, cost optimization efficiency, stability scoring and safety scoring of comprehensive load, calculating a comprehensive score for each solution, and determining the basis of the priority of the final solution by scoring the comprehensive solution.

A computer device comprising a memory storing a computer program and a processor implementing the steps of the intelligent data processing analysis and digital twin based power optimization method as described above when the computer program is executed.

A computer readable storage medium having stored thereon a computer program which when executed by a processor implements the steps of a power optimization method based on intelligent data processing analysis and digital twinning as described above.

The method has the beneficial effects that the data are analyzed through the EMD algorithm, so that the problems in the power system can be accurately identified and effectively classified. This not only improves the efficiency of problem diagnosis, but also provides a reliable basis for selecting and matching the most appropriate solution. By constructing a digital twin model of the power system, the invention enables the solution to be fully tested and validated prior to actual application. This step significantly improves the safety and reliability of the decision, avoiding potential risks and costs. The quantitative evaluation is provided for the selection of different solutions through the comprehensive scoring model, so that the decision making process is more scientific and reasonable. This step ensures that the final implemented solution can comprehensively meet various optimization requirements of the power system, achieving the best comprehensive effect.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art. Wherein:

FIG. 1 is an overall flow chart of a power optimization method based on intelligent data processing analysis and digital twinning provided by a first embodiment of the present invention;

FIG. 2 is a block diagram of a power optimization system based on intelligent data processing analysis and digital twinning according to a second embodiment of the present invention.

Detailed Description

So that the manner in which the above recited objects, features and advantages of the present invention can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to the embodiments, some of which are illustrated in the appended drawings. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.

Example 1

Referring to fig. 1, in one embodiment of the present invention, a power optimization method based on intelligent data processing analysis and digital twin is provided, which is characterized in that:

And collecting power system data and preprocessing.

A virtual model of the power system is created, and the effectiveness and safety of the solution are verified.

And carrying out iterative optimization on the solution according to the result of the simulation test.

The power system data includes power generation data, transmission data, distribution and consumption data, and equipment status data.

Further described are:

the power generation data comprise the generated energy, the power generation efficiency and the running state of various power stations such as hydroelectric power, thermal power, wind power and solar energy.

The transmission data includes transmission line load, line loss rate, real-time grid load, etc.

Distribution and consumption data relates to end user power usage, peak-to-valley periods of power usage, power quality (e.g., frequency, voltage), etc.

Environmental data includes air temperature, wind speed, sunlight, etc., and these environmental factors have a direct impact on renewable energy generation.

The device status data includes operational status of critical devices, maintenance records, fault history, etc.

For power systems, invalid data may include data generated at the time of equipment failure, blank values at the time of data collection, or illegal values (e.g., negative power generation or consumption). Such data is useless for analysis, and must first be identified and removed, as for an outlier in the power system may be caused by equipment failure, an incident (e.g., a power outage, a short circuit), or a data transmission error. Outliers may be identified and processed using statistical methods (e.g., IQR, Z-score) or machine learning methods (e.g., isolated forests).

Repeated records may be generated during the collection of power system data due to system faults or human error. And identifying and removing repeated data, and ensuring the uniqueness and accuracy of the data.

Further described are:

Different power devices and sensors may report data in different units or dimensions (e.g., power may be in kW or MW). For unified analysis, all data needs to be converted into consistent units and dimensions. In addition, for data with large range difference (such as power generation amount and ambient temperature), normalization or normalization processing is performed so as to fairly treat all variables in the model.

Power system data is typically characterized by a time series, and the frequency of collection of the different data may be different (e.g., power generation may be recorded once per hour, and grid load may be recorded once per minute). Depending on the analysis requirements, it may be desirable to resample all data to a uniform time frequency, such as every 15 minutes or every hour.

Time window division for analyzing problems such as load prediction or consumption mode of the power system, a sliding window or fixed window method can be adopted to divide time sequence data so as to capture periodic changes and trends.

Through the refined preprocessing steps, the accuracy and the reliability of the data analysis of the power system can be improved, and the follow-up model can be ensured to better understand the characteristics and the modes of the data. The targeted preprocessing method provides a solid data base for solving the specific problems of the power system.

The feature extraction of the preprocessed data is to model the power system into a graph model, nodes represent various entities in the power system, edges represent power connection or flow direction, and a feature extraction function is constructed, wherein the expression is as follows:

Solving for IMF _n(S_i (t)) using the EMD algorithm:

S1, identifying all maximum values and minimum values, namely, for given power system time series data S _i (t), firstly identifying all local maximum values and minimum value points by utilizing an improved extremum detection method. This step can be optimized by taking into account the local rate of change of the data and the added noise tolerance to reduce false positives due to data fluctuations.

S2, interpolation to obtain an envelope, namely, adopting a self-adaptive interpolation method (selecting an interpolation algorithm by considering local characteristics, such as selecting cubic spline interpolation or Gaussian process regression based on the same part fluctuation) to obtain an upper envelope, and adopting the same method to obtain a lower envelope. The adaptive interpolation method can better reflect the nonlinear characteristics of the power data.

And S3, calculating a mean value, namely calculating a mean value m (t) of the upper envelope and the lower envelope, and introducing a method for dynamically adjusting based on local energy and frequency characteristics of the signals to calculate the mean value so as to more accurately capture transient and long-term trends of the power system.

S4, extracting details, namely subtracting the mean value m (t) from the original data S _i (t) to obtain details d (t) =S _i (t) -m (t). To better reflect specific phenomena in the power system, a dynamic filtering mechanism based on data features may be added at this step, such as dynamically adjusting the extraction process according to predefined power system state features.

S5, checking IMF conditions, namely checking whether detail d (t) meets IMF conditions, introducing a self-adaptive IMF condition checking mechanism based on signal characteristics, adjusting the strictness of the IMF conditions according to the local characteristics (such as the significance of transient response and noise level) of the signals, and ensuring that the extracted IMFs are more representative.

S6, an iterative process of subtracting the extracted IMFS _i(t)＝S_i(t)-IMF_n(S_i (t) from the original data, and then repeating the steps 1 to 5 until a new IMF can not be extracted from the rest data. Intelligent stopping conditions based on signal interpretation capability and analysis targets are introduced instead of simply iterating until new IMFs cannot be extracted, and automatically determining the optimal stopping point of the iteration according to the analysis requirements of the power system.

Through the optimized EMD process, the accuracy and applicability of feature extraction are improved, and the flexibility and pertinence of the process are improved, so that the method is more suitable for processing complex and non-stationary signals in a power system. The method provides a more reliable data basis for problem identification, classification and solution matching of the power system.

The differential and smoothing method has the expression:

D(t)=S_i(t+1)-S_i(t)

S _i (t) is the time series data of the ith node in the power system at time t, the first-order difference of the original time series data of D (t), the size of a k smoothing window and the difference sequence of D' (t) after smoothing;

Adjusting the difference and smoothing method to decompose the signal is to increase or decrease the smoothing window size k of the difference sequence, increase k to output a smoother difference sequence when the original sequence noise is above a threshold, and decrease k when the signal change speed is greater than the threshold.

Detail d (t)) refers to a specific frequency component extracted from the original signal S _i (t) in a certain step. The purpose of EMD is to extract stepwise fluctuating components of various frequencies from the signal, each step of extraction of components being considered as a "level of detail" of the original signal. This level of detail represents an eigenmode function IMF of the signal and contains information about the specific frequencies of the signal.

The problem in the power system is identified for classification, namely a classification function is constructed according to the output result of the feature extraction function, and the classification function is expressed as follows:

Y_i＝C(X_i,W)

The matching solution is to construct a matching model with the expression:

S_i＝M(Y_i)

Constructing a virtual model of the power system, wherein the virtual model comprises a load prediction model, a power system cost model, a stability evaluation model and a safety and reliability evaluation model;

the load prediction model has the expression:

Wherein, Representing the predicted load at time t, L _t-1,L_t-2,…,L_t-n representing historical load data, Θ representing model parameters;

the power system cost model has the expression:

Wherein c _i is the unit cost of the i-th power generation unit, P _g,i is the power generation amount, lambda is the penalty coefficient of the load prediction error, L _j and The actual and predicted loads, respectively;

the stability evaluation model has the expression:

Where S is the stability index, Δω (T) is the system frequency deviation, ω _max is the allowed maximum of the frequency deviation, α is the attenuation coefficient, and T is the evaluation duration;

The safety and reliability evaluation model has the expression:

The priority of the verification solution is that a comprehensive scoring model is built, the priority of the solution is determined according to the score, and the expression is:

E_k＝ω_aA_k+ω_bB_k+ω_cC_k+ω_dD_k

Example 2

Referring to fig. 2, a system of the power optimization method based on intelligent data processing analysis and digital twin is provided, and is characterized by comprising a data collection and preprocessing module, a feature extraction and classification module, a solution matching and priority assessment module, a virtual model construction and verification module and a comprehensive scoring module.

The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. The storage medium includes a U disk, a removable hard disk, a Read-only memory (ROM), a random access memory (RAM, randomAccessMemory), a magnetic disk, an optical disk, or other various media capable of storing program codes.

Logic and/or steps represented in the flowcharts or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.

More specific examples (a non-exhaustive list) of the computer-readable medium include an electrical connection (an electronic device) having one or more wires, a portable computer diskette (a magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.

It is to be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or combination of techniques known in the art, discrete logic circuits with logic gates for implementing logic functions on data signals, application specific integrated circuits with appropriate combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.

Example 3

In this embodiment, in order to verify the beneficial effects of the present invention, scientific demonstration is performed through economic benefit calculation and simulation experiments. The present embodiment has been conducted by the conventional method and the method of the present embodiment.

In this embodiment, a test was designed to verify the effectiveness of the "intelligent data processing analysis and digital twin based power optimization method and system". Experiments were aimed at comparing the performance of the present invention with the prior art in terms of power system load prediction accuracy, cost optimization efficiency, system stability and safety.

The test preparation stage comprises the step of collecting power system operation data in two months, wherein the data comprise the generated energy, the transmitted energy, the distribution and consumption data and the state data of various devices. By cooperating with the electric company, the authenticity and accuracy of the collected data are ensured. Then, the data preprocessing method provided by the invention is used for denoising, standardization processing and time window division on the collected data so as to prepare subsequent analysis and model construction.

And constructing a feature extraction function and a classification model, and modeling the power system as a graph model, wherein nodes represent various entities in the power system, and edges represent power connection or flow directions. Features are extracted from the preprocessed data through intelligent data processing analysis technology, particularly a machine learning algorithm, and problems in the power system are effectively classified.

Next, a digital twin model of the power system is created and several pre-selected solutions are verified in this virtual environment. The solution scheme comprises power generation dispatching optimization, transmission line improvement scheme, a demand response strategy of the power distribution network and the like. Each solution was modeled in a digital twin model according to its expected impact.

To evaluate and compare the effects of different solutions, this embodiment devised a set of comprehensive scoring models. The model combines the scores of the aspects of load prediction accuracy, cost optimization efficiency, stability, safety and the like, and calculates the comprehensive score of each solution to determine the priority thereof.

Table 1 data comparison table

By analyzing the data in the table, the invention can obviously improve the load prediction accuracy, the cost optimization efficiency, the stability index and the safety grading compared with the prior art. Compared with the prior art, the load prediction accuracy of the scheme C is improved by 8%, the cost optimization efficiency is improved by 15%, and the stability index and the safety score are respectively improved by 13% and 12%. The improvement directly reflects the application of the intelligent data processing and digital twin technology, so that the dynamic change of the power system can be more accurately understood, and various problems in the system can be effectively predicted and dealt with.

The composite score further demonstrates the superiority of the present invention. By evaluating the solution taking into account a number of key performance indicators, it is ensured that the selected solution meets the optimization requirements of the power system in many ways. The scheme C has the highest comprehensive score, and shows the comprehensive advantages of improving the prediction accuracy, reducing the cost, enhancing the system stability and improving the safety.

These data support the innovative and novel effects of the present invention, demonstrating that the present invention can provide a more efficient power system optimization solution than the prior art. Through intelligent data analysis and digital twin model application, the invention improves the operation efficiency and safety of the power system.

Table 2 data comparison table

The invention shortens the system response time from 60 seconds to 15 seconds through an optimization algorithm and an efficient data processing flow through table data, and the quick response capability is important for real-time monitoring of the power system and quick response in emergency. The high-efficiency data processing architecture of the novel technology can improve the daily processing data volume from 5TB to 20TB, greatly enhance the capability of data processing and analysis of a power system, and provide richer and more accurate data support for power optimization and decision.

According to the invention, by introducing more flexible data analysis and optimization algorithm, the number of the adaptive adjustment times which can be carried out by the system per month is increased from 2 times to 8 times, and the adaptability and the flexibility of the power system to environmental changes are effectively improved. By optimizing the power distribution and the use efficiency, the energy saving and consumption reduction rate is improved from 5% to 20%, so that the operation cost is obviously reduced, the influence on the environment is reduced, and the sustainable development goal is realized.

It should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention may be modified or substituted without departing from the spirit and scope of the technical solution of the present invention, which is intended to be covered in the scope of the claims of the present invention.

Claims

1. A power optimization method based on intelligent data processing analysis and digital twins, characterized by comprising:

Implement the collection of power system data and pre-processing;

Extract features from preprocessed data, identify and classify problems in the power system, and match solutions;

Build a virtual model of the power system to verify the priority of the solution;

The feature extraction of the preprocessed data is to model the power system into a graph model, where nodes represent various entities in the power system, edges represent power connections or flows, and construct a feature extraction function, which is expressed as:

Where _Xi is the characteristic vector of node i extracted by integrating time series data and its rate of change, α, β and γ are coefficients for adjusting the contribution of time series data, first-order derivative and custom nonlinear function Ψ, Θ represents the parameter of nonlinear function Ψ, S _i (t) is the power data sequence of node i at time t, IMF _n (S _i (t)) is the nth intrinsic mode function, and the EMD algorithm is used to solve IMF _n (S _i (t));

The identification and classification of problems in the power system is to construct a classification function based on the output of the feature extraction function, which is expressed as:

_Yi ＝C( _Xi ,W)

Among them, W is the weight matrix in the classification function C, are the adjacent node sets of nodes i and j respectively, are the normalized degrees of nodes i and j respectively;

The matching solution is to construct a matching model, which is expressed as:

S _i ＝M(Y _i )

Among them, _Yi represents the state category of node i after classification according to the feature vector, _Si is the set of solution strategies matched to the problem of node i, and M is a rule mapping;

Determine the state category Yi of node _i according to the probability distribution, select the category with the highest probability as the final state of node i, and determine the corresponding solution strategy set _Si through the rule mapping M;

The virtual model of the power system includes a load forecasting model, a power system cost model, a stability assessment model, and a safety and reliability assessment model;

The load forecasting model is expressed as:

in, represents the predicted load at time t, L _t-1 , L _t-2 ,…, L _tn represent historical load data, and Θ represents model parameters;

The power system cost model is expressed as:

Where _ci is the unit cost of the ith generating unit, Pg _,i is the power generation, λ is the penalty coefficient for load forecast error, _Lj and are the actual and predicted loads, respectively;

The stability evaluation model is expressed as:

Where S is the stability index, Δω(t) is the system frequency deviation, ω _max is the maximum allowable frequency deviation, α is the attenuation coefficient, and T is the evaluation duration;

The safety and reliability evaluation model is expressed as:

Where R is the system reliability index, p _f,i is the probability of failure of the i-th component, n _k is the number of components involved in the k-th failure mode, and σ _k is the proportion of system performance loss caused by the k-th failure mode;

The priority of the verification solution is to build a comprehensive scoring model and determine the priority of the solution according to the score. The expression is:

E _k =ω _a A _k +ω _b B _k +ω _c C _k +ω _d D _k

Where, _Lj is the actual load, is the load predicted by solution S _k , M is the total number of time points in the evaluation period, P _g,i,k is the power generation of the i-th generation unit in solution S _k , c _i is the unit cost of the i-th generation unit, λ is the penalty coefficient for load forecast error, Δω _k (t) is the system frequency deviation caused by solution S _k , ω _max is the maximum allowable frequency deviation, n _k is the number of components involved in the k-th failure mode, p _f,i is the probability of failure of the i-th component, A _k , B _k , C _k , and D _k are the load forecast accuracy score, cost optimization efficiency score, stability score, safety and reliability score of solution S _k , respectively, and E _k is the comprehensive score.

2. The power optimization method based on intelligent data processing analysis and digital twins according to claim 1, characterized in that: the power system data includes power generation data, transmission data, distribution and consumption data, and equipment status data;

The preprocessing is to remove invalid, abnormal and duplicate data records from the collected power system data, standardize the data, and appropriately resample or divide the time window for time-sensitive data;

The use of the EMD algorithm to solve the IMF _n (S _i (t)) includes using extreme value detection on the preset power system time series data S _i (t), identifying all local maximum and minimum points through differential and smoothing processing methods, obtaining the upper envelope and the lower envelope using an adaptive interpolation method, further calculating the envelope mean and introducing a method based on the local energy and frequency characteristics of the signal to dynamically adjust the mean, subtracting the envelope mean from the original data S _i (t), introducing a dynamic filtering mechanism based on data characteristics to output the signal hierarchy, and judging whether the detail d(t) meets the IMF condition;

When the number of maximum and minimum points is not equal and the difference is greater than 1, the current signal volatility is asymmetric, and the difference and smoothing methods are further adjusted to decompose the signal;

When the number of maximum points and minimum points is equal or the difference is equal to 1, it is further determined whether the mean of the upper envelope and the lower envelope of the detail d(t) at any point is 0. If it is not 0, it is determined to be an unbalanced fluctuation and cannot be represented as a complete IMF. The original data or detail d(t) is further analyzed. If the mean of the upper envelope and the lower envelope of the detail d(t) at any point is 0, it is determined to be a complete IMF, which is further extracted from the original signal. The extracted IMFS _i (t) = S _i (t) - IMF _n (S _i (t)) is subtracted from the original data. The EMD algorithm is repeated until no new IMF can be extracted from the remaining data.

The difference and smoothing method is expressed as:

D(t)＝S _i (t+1)-S _i (t)

Where S _i (t) is the time series data of the i-th node in the power system at time t, D(t) is the first-order difference of the original time series data, k is the size of the smoothing window, and D′(t) is the difference sequence after smoothing;

The extreme points are determined by checking the sign change of the smoothed difference sequence. If D′(t-1)<0 and D′(t+1)>0, then S _i (t) is a local minimum point. If D′(t-1)>0 and D′(t+1)<0, then S _i (t) is a local maximum point.

The adjustment difference and smoothing processing method decomposes the signal by increasing or decreasing the smoothing window size k of the difference sequence. When the original sequence noise is higher than a threshold, k is increased to output a smoother difference sequence. When the signal change speed is greater than a threshold, k is reduced.

3. A system using the power optimization method based on intelligent data processing analysis and digital twin as described in any one of claims 1 to 2, characterized in that it includes a data collection and preprocessing module, a feature extraction and classification module, a solution matching and priority assessment module, a virtual model construction and verification module, and a comprehensive scoring module;

The data collection and preprocessing module collects data from the power system for data preprocessing, removes invalid, abnormal and duplicate data records, performs standardization processing, and performs appropriate resampling or time window division on time-sensitive data;

The feature extraction and classification module uses a feature extraction function to extract key features from the preprocessed data and classifies the problems in the power system through a classification function;

The solution matching and priority assessment module is to build a matching model, match appropriate solutions to various problems identified in the power system, and assess the priority of each solution through a comprehensive scoring model;

The virtual model construction and verification module is to construct a virtual model of the power system, including load forecasting, cost model, stability and security assessment model, and verify the effects of different solutions without interfering with the actual power system operation through the virtual model;

The comprehensive scoring module comprehensively considers load forecasting accuracy, cost optimization efficiency, stability score and safety score, calculates a comprehensive score for each solution, and determines the basis for the priority of the final solution through scoring the comprehensive solution.

4. A computer device comprising a memory and a processor, wherein the memory stores a computer program, and wherein when the processor executes the computer program, the steps of the power optimization method based on intelligent data processing and analysis and digital twins described in any one of claims 1 to 2 are implemented.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the power optimization method based on intelligent data processing analysis and digital twins as described in any one of claims 1 to 2 are implemented.