[go: up one dir, main page]

CN110696835B - An automatic warning method and automatic warning system for dangerous driving behavior of vehicles - Google Patents

An automatic warning method and automatic warning system for dangerous driving behavior of vehicles Download PDF

Info

Publication number
CN110696835B
CN110696835B CN201910963943.7A CN201910963943A CN110696835B CN 110696835 B CN110696835 B CN 110696835B CN 201910963943 A CN201910963943 A CN 201910963943A CN 110696835 B CN110696835 B CN 110696835B
Authority
CN
China
Prior art keywords
automatic
vehicle
cloud model
dangerous driving
data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN201910963943.7A
Other languages
Chinese (zh)
Other versions
CN110696835A (en
Inventor
向怀坤
曾松
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Polytechnic
Original Assignee
Shenzhen Polytechnic
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Polytechnic filed Critical Shenzhen Polytechnic
Priority to CN201910963943.7A priority Critical patent/CN110696835B/en
Publication of CN110696835A publication Critical patent/CN110696835A/en
Application granted granted Critical
Publication of CN110696835B publication Critical patent/CN110696835B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
    • B60W40/08—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to drivers or passengers
    • B60W40/09—Driving style or behaviour
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
    • B60W50/08—Interaction between the driver and the control system
    • B60W50/14—Means for informing the driver, warning the driver or prompting a driver intervention
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Automation & Control Theory (AREA)
  • Mathematical Physics (AREA)
  • Biomedical Technology (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Biophysics (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Mechanical Engineering (AREA)
  • Transportation (AREA)
  • Human Computer Interaction (AREA)
  • Traffic Control Systems (AREA)

Abstract

本发明提供一种车辆危险行驶行为的自动预警方法及自动预警系统，本发明自动预警方法包括如下步骤：构建车辆危险行驶行为的自动判别云模型；基于自动判别云模型，构建车辆危险行驶行为自动预测云模型‑Elman神经网络；实时采集并存储车辆运动姿态数据；利用实测数据对自动预测云模型‑Elman神经网络进行训练，得到满足精度要求的自动预警云模型‑Elman神经网络；基于自动预警云模型‑Elman神经网络进行自动预警。本发明能够对车辆的危险行驶行为进行准确的、快速的、可靠的分析和预判，实现对车辆危险行驶行为的主动预警，有效降低道路交通事故。

Figure 201910963943

The present invention provides an automatic early warning method and an automatic early warning system for dangerous driving behavior of vehicles. The automatic early warning method of the present invention includes the following steps: constructing a cloud model for automatically discriminating dangerous driving behaviors of vehicles; Prediction cloud model-Elman neural network; collect and store vehicle motion and attitude data in real time; use the measured data to train the automatic prediction cloud model-Elman neural network to obtain an automatic early warning cloud model-Elman neural network that meets the accuracy requirements; based on the automatic early warning cloud Model-Elman neural network for automatic warning. The invention can accurately, quickly and reliably analyze and predict the dangerous driving behavior of the vehicle, realize active early warning of the dangerous driving behavior of the vehicle, and effectively reduce road traffic accidents.

Figure 201910963943

Description

Automatic early warning method and automatic early warning system for dangerous driving behaviors of vehicle
Technical Field
The invention relates to the field of vehicle safety management, in particular to an automatic early warning method for dangerous driving behaviors of a vehicle and an automatic early warning system for realizing the automatic early warning method for the dangerous driving behaviors of the vehicle.
Background
The dangerous driving behavior of the vehicle is a main factor causing road traffic accidents, and how to quickly, accurately and reliably predict the dangerous driving behavior is one of the difficult problems in the research of the field of vehicle safety management. The dangerous driving behaviors of the vehicle mainly refer to the characteristics of states affecting lane safety keeping stability, vehicle relative distance control safety, vehicle speed/direction control stability and the like, and can be generally divided into two categories, namely longitudinal dangerous driving behaviors and transverse dangerous driving behaviors, wherein typical dangerous driving behaviors comprise that the vehicle is too close to the following vehicle, overspeed, rapid acceleration, rapid deceleration, rapid turning, frequent lane changing and illegal overtaking.
In the conventional research on dangerous driving behaviors, the driving behavior characteristics of a driver are mostly developed, or the dangerous motion state of a vehicle is judged by directly utilizing vehicle-mounted sensor data, so that the subjective feeling and judgment of vehicle safety conditions by vehicle-mounted passengers are rarely considered.
For the analysis of dangerous driving behaviors, research on classification of driving styles of drivers and identification methods thereof is mainly focused. The classification of the driving style of the driver can be generally divided into two major categories, a statistical-based method and a machine learning-based method. Constantinescu et al utilize vehicle-mounted GPS data to perform modeling analysis on a driver driving style, and Hong et al utilize a sensor platform composed of a vehicle-mounted android smart phone, an OBD and an IMU to collect driving behavior parameters; the identification of dangerous driving behaviors is mainly realized by detecting driving events related to safety, such as sudden acceleration, sudden braking, sudden turning and the like, and the dangerous driving behaviors are identified by simultaneously collecting driver monitoring data and vehicle posture data and data mining. The research methods can be generally divided into two major categories, namely template matching based methods and threshold based discrimination methods. Chen, Fan, and Tien et al propose to use driving habit Diagrams (DHG) to simulate driving behavior, Chen et al convert dangerous driving events into an Attributive Relationship Map (ARM) of danger in their research, and then use a two-way fuzzy attribute mapping matching technique; HAN and the like collect speed, acceleration and yaw velocity data by using a vehicle black box, and identify 4 dangerous driving states of the vehicle, such as rapid acceleration, rapid deceleration, rapid turning and sudden lane change; JOHNSON et al studied the recognition threshold for aggressive dangerous driving behavior and found that the steering threshold for aggressive driving was 0.73g and the emergency steering threshold was 0.74 g.
As can be seen from the above review and analysis of the documents, in the process of studying the vehicle motion state and the dangerous driving behavior, the vehicle-mounted sensor unit (such as a GPS, an accelerometer, and the like) is basically used for collecting the vehicle motion state data, but there is no unified and mature method for processing the data and detecting the vehicle motion state. For the research on the dangerous driving behaviors of vehicles, at present, more emphasis is placed on drivers and the operation level of the drivers on the vehicles, and the research method mainly adopts a mathematical statistics method, a specific event template matching method and a machine learning method. Because drivers have very complicated influence factors, the individuation characteristics of the drivers are difficult to express, the privacy problem of people is involved when the drivers are directly monitored, and some monitoring devices can also interfere with the normal driving behaviors of the drivers, so that the management of the safe driving of the vehicles by directly monitoring the drivers is always in greater dispute and is also difficult in practical application. Due to different devices and different data acquisition frequencies, the output vehicle motion state data has strong randomness and dynamic ambiguity characteristics.
In fact, whether the vehicle is safely driven or not finally shows the motion state of the vehicle no matter how complicated the vehicle is influenced by during driving and no matter what driving behavior the driver takes during driving of the vehicle. In other words, the motion state of the vehicle necessarily reflects a certain driving behavior of the driver, and a dangerous vehicle driving state necessarily corresponds to a dangerous driving behavior. Based on the analysis, the invention adopts a research scheme of prejudging dangerous driving behaviors of the vehicle based on real-time monitoring vehicle motion state data.
The difficulty of automatic judgment of dangerous driving behaviors of vehicles and automatic prediction of dangerous driving behaviors of vehicles based on real-time monitored vehicle motion state data is as follows: on one hand, the driver has very complicated influence factors, so that the individuation characteristic of the driver is difficult to accurately express quantitatively, and on the other hand, the output vehicle motion state data has strong randomness and dynamic ambiguity characteristics due to different devices and different data acquisition frequencies. How to relate the vehicle motion state data with high uncertainty change with dangerous driving behaviors to construct a conversion model between quantitative data based on the vehicle motion state and a complex qualitative concept of dangerous driving behaviors, and designing a prediction algorithm of the dangerous driving behaviors based on the vehicle motion state by using the model is a difficult problem in the cross research field at present.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides an automatic early warning method and an automatic early warning system for dangerous driving behaviors of a vehicle.
The automatic early warning method for the dangerous driving behaviors of the vehicle comprises the following steps of:
s1: constructing an automatic discrimination cloud model of dangerous driving behaviors of the vehicle;
s2: based on the automatic discrimination cloud model, constructing an automatic prediction cloud model-Elman neural network for the dangerous driving behavior of the vehicle;
s3: collecting and storing vehicle motion attitude data in real time;
s4: training the automatic prediction cloud model-Elman neural network by using the measured data to obtain an automatic early warning cloud model-Elman neural network meeting the precision requirement;
s5: based on an automatic early warning cloud model-Elman neural network, the dangerous driving behavior of the vehicle is automatically early warned by utilizing the measured data.
The invention further improves, after step S3 is executed, the method further includes the vehicle motion attitude data processing step: preprocessing the acquired original data to remove noise in the data, and then repairing the removed data, wherein in the steps of S4 and S5, the actually measured data are data obtained by processing the vehicle motion attitude data.
The invention is further improved, in step S1, the automatic discrimination cloud model refers to a discrimination index of a corresponding relationship between a total weighted acceleration root mean square value and human subjective feeling, and an expert score and passenger feeling to construct an automatic discrimination index of dangerous driving behavior of the vehicle, which is issued by international standards organization and issued by mechanical vibration and impact evaluation standards of human body exposed to whole body vibration and detection standards of smooth driving of automobiles in China, and then extracts a cloud model digital characteristic parameter corresponding to each discrimination index based on experimental observation data by using a reverse cloud transformation algorithm of the cloud model, and the automatic discrimination cloud model is obtained through multiple experiments.
The invention is further improved, and the construction method of the discrimination index of the corresponding relation between the total weighted acceleration root mean square value and the subjective feeling of people comprises the following steps:
s101: for the vibration signal, a discrete fourier transform is adopted to convert the vibration signal into a frequency domain, and the conversion formula is as follows:
Figure BDA0002229866940000031
wherein, x (N) is a finite vibration signal with length N in time domain, and x (f) is a vibration signal in frequency domain;
s102: calculating the root mean square value of one third octave and the weight acceleration of the center of one third octave, wherein the calculation formula of the root mean square value of one third octave is as follows:
Figure BDA0002229866940000032
wherein, aiIs the root mean square value of one third octave and has the unit of m/s2，fiuIs the upper cut-off frequency of the ith frequency band, filIs the lower cut-off frequency of the ith band, df represents the differential of the frequency f,
Figure BDA0002229866940000033
is to find a definite integral, the definite integral interval is fil,fiu]，
Because the human body has different responses to different vibration frequencies in different directions, a weighting factor is given to the center of the frequency to make real measurement data and react to the feeling of the human body, a table is looked up on a center frequency table of one third octave corresponding to the weighting factor, and the acceleration of each axis is calculated through a formula (3), wherein the calculation formula is as follows:
Figure BDA0002229866940000034
wherein, awjIs the weighted acceleration of the vibration signal of each axis, the unit of which is m/s2，j＝x，y，z，kiIs the weighting coefficient of the ith one-third octave band;
s103: and setting the weight value of the acceleration of each axis, weighting the total acceleration of each axis, and calculating the root mean square value of the total acceleration.
The invention further improves that in step S2, the method for constructing the automatic prediction cloud model-Elman neural network includes: the method comprises the steps of using an automatic discrimination cloud model as a target output vector of an Elman neural network, achieving mapping between a cloud model evaluation result and an output value of an MEMS sensor, wherein the Elman neural network comprises an input layer, an output layer and a hidden layer, the input layer, the output layer and the hidden layer are respectively provided with a plurality of neurons, training the Elman neural network, and sampling, identifying and outputting through adjustment of weights of all layers to enable a root mean square value error to be minimum.
The invention is further improved, and the vehicle motion attitude data comprises six-degree-of-freedom motion attitude data and vehicle motion speed parameters.
The invention also provides an automatic early warning system for realizing the automatic early warning method of the dangerous driving behaviors of the vehicle, which comprises the following steps:
a first building block: the automatic judgment cloud model is used for constructing the dangerous driving behavior of the vehicle;
a second building block: an automatic prediction cloud model-Elman neural network for constructing dangerous driving behaviors of the vehicle based on the automatic discrimination cloud model;
a data acquisition module: the system is used for acquiring vehicle motion attitude data in real time;
a data storage module: for storing data;
a training module: the system comprises a cloud model-Elman neural network, a cloud model-to-model and a cloud model-to-neural network;
an automatic early warning module: the real-time early warning method is used for carrying out real-time automatic early warning on dangerous driving behaviors of the vehicle by utilizing the measured data based on an automatic early warning cloud model-Elman neural network.
The invention is further improved, and the invention also comprises a vehicle motion attitude data processing module: the method is used for preprocessing the acquired original data to remove noise in the data and then repairing the removed data.
Compared with the prior art, the invention has the beneficial effects that: the method can accurately, quickly and reliably analyze and prejudge the dangerous driving behaviors of the vehicle, realizes active early warning of the dangerous driving behaviors of the vehicle, and effectively reduces road traffic accidents.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a flow chart of a method for building an automatic discrimination cloud model;
FIG. 3 is a flow chart of a cloud model-Elman neural network construction method for automatically predicting dangerous driving behaviors of a vehicle;
FIG. 4 is a schematic diagram of a hardware composition structure of the intelligent vehicle-mounted terminal.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples.
The invention discloses a cloud model-Elman neural network-based automatic early warning method for dangerous driving behaviors, which is designed by introducing a cloud model theory to combine vehicle driving posture data with subjective feelings of passengers and aiming at the safety supervision requirements of future unmanned vehicles.
As shown in FIG. 1, the invention designs a vehicle motion state data acquisition and processing system, and realizes real-time acquisition and processing of vehicle six-degree-of-freedom motion attitude data, vehicle motion speed and other parameters. And secondly, establishing a vehicle dangerous driving behavior automatic distinguishing cloud model corresponding to the vehicle motion state by combining the international standard and the domestic standard of the vehicle motion perception of the passengers and expert knowledge. And finally, designing an automatic prediction algorithm of the dangerous driving behavior of the vehicle based on a cloud model-Elman neural network, and taking the automatic judgment cloud model of the dangerous driving behavior of the vehicle as a target output vector of the Elman neural network, thereby carrying out real-time automatic early warning on the dangerous driving behavior of the vehicle. The respective steps are explained in detail below.
The method comprises the following steps: automatic judgment cloud model for dangerous driving behaviors of vehicle
Referring to the evaluation standards of mechanical vibration and impact of human body exposed to whole body vibration and the detection standards of automobile smoothness in China, which are published by the international standard organization, the invention relates to a judgment index of the corresponding relation between the total weighted acceleration root mean square value (RMS) and the subjective feeling of human, and an automatic judgment index of dangerous driving behaviors of vehicles, which is established by the scoring of experts and the feeling of passengers, then extracts the digital characteristic parameters of a cloud model corresponding to each judgment index based on experimental observation data by using the cloud transformation algorithm of the cloud model, and obtains a set of automatic judgment cloud models of dangerous driving behaviors of vehicles through a plurality of experiments.
Specifically, the method for constructing the discrimination index of the corresponding relationship between the root mean square value of the total weighted acceleration and the subjective feeling of the person comprises the following steps:
s101: for the vibration signal (triaxial acceleration), a discrete fourier transform is adopted to convert the vibration signal into a frequency domain, and the conversion formula is as follows:
Figure BDA0002229866940000051
wherein, x (N) is a finite vibration signal with length N in time domain, and x (f) is a vibration signal in frequency domain;
s102: calculating a root mean square value (RMS) of a third octave and a weight acceleration of a center of the third octave, the
The one-third octave root mean square value calculation formula is as follows:
Figure BDA0002229866940000052
wherein, aiIs the root mean square value of one third octave and has the unit of m/s2，fiuIs the upper cut-off frequency of the ith frequency band, filIs the lower cut-off frequency of the ith band, df represents the differential of the frequency f,
Figure BDA0002229866940000053
is to find a definite integral, the definite integral interval is fil,fiu]，
Since the human body reacts differently to different vibration frequencies in different directions, real measurement data can be made if a weighting factor is given to the center of the frequency, thereby reacting to the feeling of the human body. ISO2631-1 (1997)/Amd 1:2010 gives a table of the center frequency of the third octave corresponding to the weighting factor for each axis. Thus, the acceleration of each axis is calculated by looking up the table and by equation (3) as:
Figure BDA0002229866940000054
wherein, awjIs the weighted acceleration of the vibration signal of each axis, the unit of which is m/s2，j＝x，y，z，kiIs the weighting coefficient of the ith one-third octave band;
s103: and setting the weight value of the acceleration of each axis, weighting the total acceleration of each axis, and calculating the root mean square value of the total acceleration. In this example, the weight of the x-axis and the y-axis is 1.4, the weight of the z-axis is 1.0, and the root mean square value of the total acceleration is calculated as:
Figure BDA0002229866940000061
wherein, awIs the root mean square value of the total acceleration in m/s2And a is awx、awy、awzIs the root mean square value of each axis in equation (3).
The inverse cloud transformation algorithm adopted by the automatic discrimination cloud model digital feature calculation of the embodiment is as follows:
inputting: n cloud drops xi(i＝1，2，...，n)；
And (3) outputting: numerical feature estimate (expectation)
Figure BDA0002229866940000062
Entropy of the entropy
Figure BDA0002229866940000063
And entropy
Figure BDA0002229866940000064
)。
Step 1:
Figure BDA0002229866940000065
calculating an arithmetic mean value, and taking the arithmetic mean value as an estimated value of the expected parameter of the cloud model;
and 2, step 2: randomly sampled packets
For(i＝1，i≤m，i++)
For(j＝1，j≤r，j++)
Randomly sampling n cloud drop xi (i ═ 1, 2.., n) samples
END For
Xi＝{Xi1，Xi2，...Xir}，
Figure BDA0002229866940000066
Figure BDA0002229866940000067
END For
Figure BDA0002229866940000068
Figure BDA0002229866940000069
And 3, step 3:
Figure BDA00022298669400000610
the step 2 and the step 3 are entropy of the cloud model calculated by computer software programming
Figure BDA00022298669400000611
And entropy
Figure BDA00022298669400000612
In which two loops are involved, wherein m, j, r have the meaning only in this program generationThe code module is effective, and respectively represents the random sampling grouping number of cloud droplet samples, and is a variable set by a program when the program is used for calculation in the module.
Step two: method for constructing automatic prediction cloud model-Elman neural network of dangerous driving behaviors of vehicle
The construction process is shown in fig. 3, the automatic judgment cloud model of the dangerous driving behaviors of the vehicle is used as a target output vector of the Elman neural network, the mapping between the cloud model evaluation result and the output value of the MEMS (micro electro mechanical system) sensor is realized, and the automatic early warning of the dangerous driving behaviors is completed. The input layer of the Elman neural network is set to be 6 neurons, the output layer is set to be 1 neuron, and the number of the selected hidden layer neurons is 13. The network is trained by adopting a gradient descent method, and sampling output and recognition output can be performed by adjusting the weight of each layer of network, so that the Mean Square Error (MSE) is minimum.
The nonlinear space state expression of the automatic prediction cloud model-Elman neural network is as follows:
Figure BDA0002229866940000071
xc(k)＝x(k-1) (6)
Figure BDA0002229866940000072
wherein u (k-1) represents the network input, y (k) represents the network output, x (k) represents the hidden layer output,
Figure BDA0002229866940000073
and f and g respectively represent transfer functions of the hidden layer and the output layer.
Thirdly, collecting the vehicle motion attitude data in real time and storing the data
And opening a vehicle motion state data acquisition and processing system, acquiring vehicle motion attitude data in real time, and storing the data. As shown in fig. 4, the system mainly includes a vehicle-mounted GPS, Micro-Electro-Mechanical Systems (MEMS) sensors, a CAN bus, a camera, and the like, and the MEMS sensors include a three-axis acceleration sensor and a three-axis angle sensor, so as to realize real-time acquisition and processing of parameters such as vehicle six-degree-of-freedom motion attitude data and vehicle motion speed.
Fourthly, vehicle motion attitude data processing
The method comprises the steps of utilizing Kalman filtering to preprocess original data to remove noise (error data) in the data, and utilizing a quadratic exponential smoothing method to repair the removed data.
And fifthly, training the cloud model-Elman neural network by utilizing the measured data to obtain the automatic early warning cloud model-Elman neural network for the dangerous driving behavior of the vehicle, which meets the precision requirement.
And sixthly, based on an automatic early warning cloud model-Elman neural network, real-time automatic early warning is carried out on dangerous driving behaviors of the vehicle by utilizing the measured data.
The invention can accurately, quickly and reliably analyze and prejudge the dangerous driving behaviors of the vehicle. Research reports on NHTSA (1998) show that dangerous driving behavior is a major cause of road traffic safety problems. Expert scholars from different countries found that 93-94% of road traffic accidents are caused by human factors, 8-12% by vehicle factors, and 28-34% by road factors [1] (Edu Ardo AVascon cells, 1996). The dangerous driving behaviors of the vehicle directly reflect the operation problems of a vehicle driver, if the dangerous driving behaviors of the vehicle can be found in time, the driver can be reminded and warned in time, and the vehicle can be forcibly taken over even by the vehicle-mounted safety management and control device at a critical moment, so that traffic accidents can be effectively prevented. Taking a bus and a taxi as an example, if a driver frequently implements dangerous driving behaviors such as rapid acceleration, rapid deceleration or rapid turning in the driving process, the traffic order is easily disturbed, traffic accidents are caused, passengers feel uncomfortable, and the fear psychology is generated.
The above-described embodiments are intended to be illustrative, and not restrictive, of the invention, and all changes that come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

Claims (8)

1.一种车辆危险行驶行为的自动预警方法，其特征在于，包括如下步骤：1. an automatic warning method of vehicle dangerous driving behavior, is characterized in that, comprises the steps: S1：构建车辆危险行驶行为的自动判别云模型；S1: Build a cloud model for automatic discrimination of dangerous driving behaviors of vehicles; S2：基于自动判别云模型，构建车辆危险行驶行为自动预测云模型-Elman神经网络；S2: Based on the automatic discriminant cloud model, construct a cloud model for automatic prediction of vehicle dangerous driving behavior-Elman neural network; S3：实时采集并存储车辆运动姿态数据；S3: Collect and store vehicle motion and attitude data in real time; S4：利用实测数据对自动预测云模型-Elman神经网络进行训练，得到满足精度要求的自动预警云模型-Elman神经网络；S4: Use the measured data to train the automatic prediction cloud model-Elman neural network, and obtain the automatic early warning cloud model-Elman neural network that meets the accuracy requirements; S5：基于自动预警云模型-Elman神经网络，利用实测数据对车辆的危险行驶行为进行实时自动预警，S5: Based on the automatic early warning cloud model-Elman neural network, real-time automatic early warning of dangerous driving behaviors of vehicles using measured data, 所述自动判别云模型数字特征计算所采用的逆向云变换算法如下：The inverse cloud transformation algorithm used in the automatic discrimination cloud model digital feature calculation is as follows: 输入：
Figure DEST_PATH_IMAGE001
个云滴
Figure 228532DEST_PATH_IMAGE002
；enter:
Figure DEST_PATH_IMAGE001
cloud drop
Figure 228532DEST_PATH_IMAGE002
; 输出：数值特征估计值（期望
Figure DEST_PATH_IMAGE003
、熵
Figure 314431DEST_PATH_IMAGE004
和超熵
Figure DEST_PATH_IMAGE005
），Output: Numerical feature estimates (expected
Figure DEST_PATH_IMAGE003
,entropy
Figure 314431DEST_PATH_IMAGE004
and hyperentropy
Figure DEST_PATH_IMAGE005
), 步1：
Figure 37493DEST_PATH_IMAGE006
；求算术平均值，将其作为云模型的期望参数的估计值；Step 1:
Figure 37493DEST_PATH_IMAGE006
; Calculate the arithmetic mean as an estimate of the expected parameters of the cloud model; 步2：随机抽样分组Step 2: Randomly sample groups
Figure 240941DEST_PATH_IMAGE008
Figure 240941DEST_PATH_IMAGE008
Figure DEST_PATH_IMAGE009
Figure DEST_PATH_IMAGE009
 对
Figure 489651DEST_PATH_IMAGE001
个云滴
Figure 887397DEST_PATH_IMAGE002
样本进行随机抽样right
Figure 489651DEST_PATH_IMAGE001
cloud drop
Figure 887397DEST_PATH_IMAGE002
random sampling
Figure 962669DEST_PATH_IMAGE010
Figure 962669DEST_PATH_IMAGE010
Figure DEST_PATH_IMAGE011
Figure DEST_PATH_IMAGE011
Figure 595907DEST_PATH_IMAGE012
Figure 595907DEST_PATH_IMAGE012
Figure 574489DEST_PATH_IMAGE010
Figure 574489DEST_PATH_IMAGE010
Figure DEST_PATH_IMAGE013
Figure DEST_PATH_IMAGE013
Figure DEST_PATH_IMAGE015
Figure DEST_PATH_IMAGE015
 步3：
Figure 205453DEST_PATH_IMAGE016
Step 3:
Figure 205453DEST_PATH_IMAGE016
 以上步2、步3，是计算机软件编程计算云模型的熵
Figure DEST_PATH_IMAGE018A
和超熵
Figure DEST_PATH_IMAGE020A
的程序代码，其中涉及两个循环，其中的m、j、r的含义只在本程序代码模块中有效，分别表示对云滴样本再进行的随机抽样分组数，是程序在这个模块内用于计算时所设置的变量，EY2为Y2的期望值，DY2为Y2的方差。The above steps 2 and 3 are computer software programming to calculate the entropy of the cloud model
Figure DEST_PATH_IMAGE018A
and hyperentropy
Figure DEST_PATH_IMAGE020A
The program code of , which involves two loops, in which the meanings of m, j, and r are only valid in this program code module, respectively indicating the number of random sampling groups for the cloud droplet sample, which are used by the program in this module. The variables set during calculation, EY 2 is the expected value of Y 2 , and DY 2 is the variance of Y 2 . 2.根据权利要求1所述的车辆危险行驶行为的自动预警方法，其特征在于：在步骤S3执行后，还包括所述车辆运动姿态数据处理步骤：对采集的原始车辆运动姿态数据进行预处理以剔除数据中的噪声,再对剔除后的数据进行修补,2. The automatic warning method for dangerous driving behavior of vehicles according to claim 1, characterized in that: after step S3 is executed, it also includes the step of processing the vehicle motion attitude data: preprocessing the collected original vehicle motion attitude data In order to remove the noise in the data, and then repair the removed data, 步骤S4和S5中，所述实测数据为对所述车辆运动姿态数据经过处理后的数据。In steps S4 and S5, the measured data is the data after processing the vehicle motion attitude data. 3.根据权利要求1或2所述的车辆危险行驶行为的自动预警方法，其特征在于：在步骤S1中，所述自动判别云模型参考国际标准组织发布的人体暴露于全身振动的机械振动和冲击评估标准以及我国汽车平顺性行驶检测标准中，关于总加权加速度均方根值与人的主观感受对应关系的判别指标，以及专家打分及乘客感受构建车辆危险行驶行为自动判别指标，再利用云模型的逆向云变换算法提取基于实验观测数据的各判别指标对应的云模型数字特征参数，经过多次实验得到。3. the automatic warning method of vehicle dangerous driving behavior according to claim 1 and 2, is characterized in that: in step S1, described automatic discriminating cloud model is exposed to the mechanical vibration of whole body vibration and In the shock evaluation standard and my country's vehicle ride comfort detection standard, the discriminant index on the corresponding relationship between the root mean square value of the total weighted acceleration and the subjective feeling of the person, as well as the expert scoring and the passenger's feeling, construct the automatic discriminating index of the dangerous driving behavior of the vehicle, and then use the cloud The inverse cloud transformation algorithm of the model extracts the digital characteristic parameters of the cloud model corresponding to each discriminant index based on the experimental observation data, which are obtained after many experiments. 4.根据权利要求3所述的车辆危险行驶行为的自动预警方法，其特征在于：总加权加速度均方根值与人的主观感受对应关系的判别指标的构建方法为：4. the automatic warning method of vehicle dangerous driving behavior according to claim 3, is characterized in that: the construction method of the discriminant index of total weighted acceleration root mean square value and people's subjective feeling correspondence relation is: S101：对于振动信号，采用离散傅立叶变换将其转换为频域，其转换公式为：S101: For the vibration signal, the discrete Fourier transform is used to convert it into the frequency domain, and the conversion formula is:
Figure DEST_PATH_IMAGE022A
（1）
Figure DEST_PATH_IMAGE022A
(1) 其中，
Figure DEST_PATH_IMAGE024AA
为时域中长度为N的有限振动信号，
Figure DEST_PATH_IMAGE026A
为频域中的振动信号；in,
Figure DEST_PATH_IMAGE024AA
is a finite vibration signal of length N in the time domain,
Figure DEST_PATH_IMAGE026A
is the vibration signal in the frequency domain; S102：计算三分之一倍频程的均方根值和三分之一倍频程中心的重量加速度，所述S102: Calculate the root mean square value of one-third octave and the weight acceleration of the center of one-third octave, the 三分之一倍频程均方根值计算公式为：The formula for calculating the root mean square value of one-third octave band is:
Figure DEST_PATH_IMAGE028AA
Figure DEST_PATH_IMAGE030A
（2）
Figure DEST_PATH_IMAGE028AA
Figure DEST_PATH_IMAGE030A
(2) 其中，
Figure DEST_PATH_IMAGE032AA
为三分之一倍频程的均方根值，单位为
Figure DEST_PATH_IMAGE034AA
，
Figure DEST_PATH_IMAGE036AA
是第i频段的上截止频率，
Figure DEST_PATH_IMAGE038AA
是第i频段的下截止频率，
Figure DEST_PATH_IMAGE040AA
表示频率
Figure DEST_PATH_IMAGE042AA
的微分量，
Figure DEST_PATH_IMAGE044AAAA
是求定积分，定积分区间是[
Figure DEST_PATH_IMAGE046AA
]，in,
Figure DEST_PATH_IMAGE032AA
is the root mean square value of one-third octave, in units of
Figure DEST_PATH_IMAGE034AA
,
Figure DEST_PATH_IMAGE036AA
is the upper cutoff frequency of the i-th band,
Figure DEST_PATH_IMAGE038AA
is the lower cutoff frequency of the i-th band,
Figure DEST_PATH_IMAGE040AA
Indicate frequency
Figure DEST_PATH_IMAGE042AA
the derivative of ,
Figure DEST_PATH_IMAGE044AAAA
is the definite integral, and the definite integral interval is [
Figure DEST_PATH_IMAGE046AA
], 由于人体对不同方向的不同振动频率有不同的反应，在频率中心给定一个加权因子，做出真实的测量数据，对人体的感受做出反应，对加权因子相对应的三分之一倍频程的中心频率表查表，并通过公式（3）计算各轴的加速度，计算公式为：Since the human body has different responses to different vibration frequencies in different directions, a weighting factor is given at the center of the frequency to make real measurement data, to respond to the human body's feelings, and to the one-third octave corresponding to the weighting factor. Look up the table from the center frequency table of the process, and calculate the acceleration of each axis by formula (3). The calculation formula is:
Figure DEST_PATH_IMAGE048AAAA
（3）
Figure DEST_PATH_IMAGE048AAAA
(3) 其中，
Figure DEST_PATH_IMAGE050AAAA
是各轴振动信号的加权加速度，其单位是
Figure DEST_PATH_IMAGE034AAA
，
Figure DEST_PATH_IMAGE052AAAA
，
Figure DEST_PATH_IMAGE054AA
是第i个三分之一倍频程带的加权系数；in,
Figure DEST_PATH_IMAGE050AAAA
is the weighted acceleration of the vibration signal of each axis, and its unit is
Figure DEST_PATH_IMAGE034AAA
,
Figure DEST_PATH_IMAGE052AAAA
,
Figure DEST_PATH_IMAGE054AA
is the weighting factor for the i-th one-third octave band; S103：设定各轴加速度的权重值，对各轴的总加速度进行加权，计算总加速度的均方根值。S103: Set the weight value of the acceleration of each axis, weight the total acceleration of each axis, and calculate the root mean square value of the total acceleration. 5.根据权利要求1或2所述的车辆危险行驶行为的自动预警方法，其特征在于：步骤S2中，所述自动预测云模型-Elman神经网络的构建方法为：将自动判别云模型作为Elman神经网络的目标输出向量，实现云模型评价结果和MEMS传感器输出值两者之间的映射，其中，Elman神经网络包括输入层、输出层和隐含层，输入层、输出层和隐含层分别设有若干个神经元，对Elman神经网络进行训练，通过对各层权值的调整，进行采样和识别输出，使均方根值误差最小。5. the automatic warning method of vehicle dangerous driving behavior according to claim 1 and 2, is characterized in that: in step S2, the construction method of described automatic prediction cloud model-Elman neural network is: will automatically discriminate cloud model as Elman The target output vector of the neural network realizes the mapping between the evaluation result of the cloud model and the output value of the MEMS sensor. Among them, the Elman neural network includes an input layer, an output layer and a hidden layer. The input layer, the output layer and the hidden layer are respectively There are several neurons, the Elman neural network is trained, and by adjusting the weights of each layer, sampling and identification output are performed to minimize the root mean square error. 6.根据权利要求1或2所述的车辆危险行驶行为的自动预警方法，其特征在于：所述车辆运动姿态数据包括六自由度运动姿态数据以及车辆运动速度参数。6 . The automatic warning method for dangerous driving behavior of a vehicle according to claim 1 or 2 , wherein the vehicle motion attitude data comprises six degrees of freedom motion attitude data and vehicle motion speed parameters. 7 . 7.一种实现权利要求1-6任一项所述的车辆危险行驶行为的自动预警方法的自动预警系统，其特征在于，包括：7. An automatic warning system for realizing the automatic warning method for the dangerous driving behavior of a vehicle according to any one of claims 1-6, characterized in that, comprising: 第一构建模块：用于构建车辆危险行驶行为的自动判别云模型；The first building block: a cloud model for automatically discriminating dangerous driving behaviors of vehicles; 第二构建模块：用于基于所述自动判别云模型，构建车辆危险行驶行为的自动预测云模型-Elman神经网络；The second building module: based on the automatic discrimination cloud model, to construct the automatic prediction cloud model of vehicle dangerous driving behavior-Elman neural network; 数据采集模块：用于实时采集车辆运动姿态数据；Data acquisition module: used to collect vehicle motion and attitude data in real time; 数据存储模块：用于存储数据；Data storage module: used to store data; 训练模块：用于用实测数据对自动预测云模型-Elman神经网络进行训练，得到满足精度要求的自动预警云模型-Elman神经网络；Training module: used to train the automatic prediction cloud model-Elman neural network with the measured data, and obtain the automatic early warning cloud model-Elman neural network that meets the accuracy requirements; 自动预警模块：用于基于自动预警云模型-Elman神经网络，利用实测数据对车辆的危险行驶行为进行自动预警。Automatic early warning module: Based on the automatic early warning cloud model-Elman neural network, it uses the measured data to automatically warn the dangerous driving behavior of the vehicle. 8.根据权利要求7所述的自动预警系统，其特征在于：还包括数据处理模块：用于对车辆运动姿态数据进行预处理以剔除数据中的噪声,再对剔除后的数据进行修补。8 . The automatic early warning system according to claim 7 , further comprising a data processing module for preprocessing the vehicle motion and attitude data to eliminate noise in the data, and then repairing the eliminated data. 9 .
CN201910963943.7A 2019-10-11 2019-10-11 An automatic warning method and automatic warning system for dangerous driving behavior of vehicles Expired - Fee Related CN110696835B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910963943.7A CN110696835B (en) 2019-10-11 2019-10-11 An automatic warning method and automatic warning system for dangerous driving behavior of vehicles

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910963943.7A CN110696835B (en) 2019-10-11 2019-10-11 An automatic warning method and automatic warning system for dangerous driving behavior of vehicles

Publications (2)

Publication Number Publication Date
CN110696835A CN110696835A (en) 2020-01-17
CN110696835B true CN110696835B (en) 2021-11-02

Family

ID=69199280

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910963943.7A Expired - Fee Related CN110696835B (en) 2019-10-11 2019-10-11 An automatic warning method and automatic warning system for dangerous driving behavior of vehicles

Country Status (1)

Country Link
CN (1) CN110696835B (en)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2021178385A1 (en) * 2020-03-02 2021-09-10 Strong Force Intellectual Capital, Llc Intelligent transportation systems including digital twin interface for a passenger vehicle
US12380746B2 (en) 2018-09-30 2025-08-05 Strong Force Tp Portfolio 2022, Llc Digital twin systems and methods for transportation systems
US12169987B2 (en) 2018-09-30 2024-12-17 Strong Force Tp Portfolio 2022, Llc Intelligent transportation systems including digital twin interface for a passenger vehicle
CN111930117B (en) * 2020-07-31 2024-04-05 广州景骐科技有限公司 Steering-based lateral control method, device, equipment and storage medium
CN113380048B (en) * 2021-06-25 2022-09-02 中科路恒工程设计有限公司 Neural network-based high-risk road section vehicle driving behavior identification method
CN116572984B (en) * 2023-05-16 2025-09-19 重庆长安汽车股份有限公司 Dangerous driving management and control method and system based on multi-feature fusion
CN116363600B (en) * 2023-06-01 2023-08-01 深圳恒邦新创科技有限公司 Method and system for predicting maintenance operation risk of motor train unit

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103279671A (en) * 2013-06-03 2013-09-04 南京大学 Urban water disaster risk prediction method based on RBF (radial basis function) neural network-cloud model
CN106530717A (en) * 2016-12-26 2017-03-22 长安大学 Construction road section risk evaluating method based on cloud model
CN106740864A (en) * 2017-01-12 2017-05-31 北京交通大学 A kind of driving behavior is intended to judge and Forecasting Methodology
CN106781581A (en) * 2016-11-29 2017-05-31 深圳职业技术学院 Safe driving behavior monitoring early warning system and method based on the coupling of people's car
US9830823B1 (en) * 2016-08-25 2017-11-28 International Business Machines Corporation Detection of vehicle operation characteristics
KR20180072523A (en) * 2016-12-21 2018-06-29 삼성전자주식회사 Electronic apparatus and operating method for the same
CN109910904A (en) * 2019-03-22 2019-06-21 深圳市澳颂泰科技有限公司 A kind of driving behavior and vehicle drive gesture recognition system

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102360180B (en) * 2011-09-29 2013-06-05 华南理工大学 Method for identifying brake performance of safety monitoring system for motor vehicles
US20170293837A1 (en) * 2016-04-06 2017-10-12 Nec Laboratories America, Inc. Multi-Modal Driving Danger Prediction System for Automobiles
US10109198B2 (en) * 2017-03-08 2018-10-23 GM Global Technology Operations LLC Method and apparatus of networked scene rendering and augmentation in vehicular environments in autonomous driving systems
US10611371B2 (en) * 2017-09-14 2020-04-07 Toyota Motor Engineering & Manufacturing North America, Inc. System and method for vehicle lane change prediction using structural recurrent neural networks
CN108875902A (en) * 2017-12-04 2018-11-23 北京旷视科技有限公司 Neural network training method and device, vehicle detection estimation method and device, storage medium
US10578456B2 (en) * 2018-03-28 2020-03-03 Intel Corporation Safety enhanced computer assisted driving method and apparatus
CN110097055A (en) * 2019-04-29 2019-08-06 上海工程技术大学 A kind of vehicle attitude detection method and system based on grid convolutional neural networks

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103279671A (en) * 2013-06-03 2013-09-04 南京大学 Urban water disaster risk prediction method based on RBF (radial basis function) neural network-cloud model
US9830823B1 (en) * 2016-08-25 2017-11-28 International Business Machines Corporation Detection of vehicle operation characteristics
CN106781581A (en) * 2016-11-29 2017-05-31 深圳职业技术学院 Safe driving behavior monitoring early warning system and method based on the coupling of people's car
KR20180072523A (en) * 2016-12-21 2018-06-29 삼성전자주식회사 Electronic apparatus and operating method for the same
CN106530717A (en) * 2016-12-26 2017-03-22 长安大学 Construction road section risk evaluating method based on cloud model
CN106740864A (en) * 2017-01-12 2017-05-31 北京交通大学 A kind of driving behavior is intended to judge and Forecasting Methodology
CN109910904A (en) * 2019-03-22 2019-06-21 深圳市澳颂泰科技有限公司 A kind of driving behavior and vehicle drive gesture recognition system

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于危险理论-云决策的发动机状态检测技术研究;孟庆华等;《汽车工程》;20141125(第11期);全文 *

Also Published As

Publication number Publication date
CN110696835A (en) 2020-01-17

Similar Documents

Publication Publication Date Title
CN110696835B (en) An automatic warning method and automatic warning system for dangerous driving behavior of vehicles
Khodairy et al. Driving behavior classification based on oversampled signals of smartphone embedded sensors using an optimized stacked-LSTM neural networks
CN116013110B (en) Vehicle collision risk prediction device and method
Pentland et al. Modeling and prediction of human behavior
EP3498559B1 (en) Method for recognizing the driving style of a driver of a land vehicle, and corresponding apparatus
CN112528815A (en) Fatigue driving detection method based on multi-mode information fusion
CN109910904A (en) A kind of driving behavior and vehicle drive gesture recognition system
CN110321954A (en) The driving style classification and recognition methods of suitable domestic people and system
CN110143202A (en) A kind of dangerous driving identification and method for early warning and system
CN106184220B (en) Abnormal driving detection method in a kind of track based on vehicle location track
CN115179960A (en) Multi-source data acquisition man-vehicle state comprehensive monitoring system and method
CN110781873A (en) Driver fatigue grade identification method based on bimodal feature fusion
CN110901385B (en) Active speed limiting method based on fatigue state of driver
CN113406955B (en) Complex network-based automatic driving automobile complex environment model, cognitive system and cognitive method
CN110781872A (en) Driver fatigue grade recognition system with bimodal feature fusion
CN112308136B (en) Driving distraction detection method based on SVM-Adaboost
El Masri et al. Toward self-policing: Detecting drunk driving behaviors through sampling CAN bus data
CN111753586A (en) Fatigue driving monitoring device and method
EP3382570A1 (en) Method for characterizing driving events of a vehicle based on an accelerometer sensor
CN119152483B (en) A fatigue state identification method and device
Wu et al. Fuzzy logic based driving behavior monitoring using hidden markov models
CN112462759B (en) Evaluation methods, systems and computer storage media for regulatory algorithms
CN120412290A (en) A 5G+V2X module and risk warning terminal for smart transportation and its application method
CN120048127A (en) Data model processing method and system
Sun et al. Recognition method of drinking-driving behaviors based on PCA and RBF neural network

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20211102