CN113159105A - Unsupervised driving behavior pattern recognition method and data acquisition monitoring system - Google Patents

Unsupervised driving behavior pattern recognition method and data acquisition monitoring system Download PDF

Info

Publication number
CN113159105A
CN113159105A CN202110216457.6A CN202110216457A CN113159105A CN 113159105 A CN113159105 A CN 113159105A CN 202110216457 A CN202110216457 A CN 202110216457A CN 113159105 A CN113159105 A CN 113159105A
Authority
CN
China
Prior art keywords
driving behavior
driving
clustering
data
unsupervised
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.)
Granted
Application number
CN202110216457.6A
Other languages
Chinese (zh)
Other versions
CN113159105B (en
Inventor
王玲
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.)
University of Science and Technology Beijing USTB
Original Assignee
University of Science and Technology Beijing USTB
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 University of Science and Technology Beijing USTB filed Critical University of Science and Technology Beijing USTB
Priority to CN202110216457.6A priority Critical patent/CN113159105B/en
Publication of CN113159105A publication Critical patent/CN113159105A/en
Application granted granted Critical
Publication of CN113159105B publication Critical patent/CN113159105B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features

Abstract

The invention discloses a driving behavior unsupervised mode identification method and a data acquisition monitoring system, wherein the method comprises the following steps: collecting driving data segments of a driver in the driving process and extracting characteristics; clustering the extracted features by using an improved self-adaptive K-Means clustering algorithm based on Wasserstein distance to obtain a driving behavior mode; and establishing driving behavior topic description, and matching the driving behavior topic description with all the driving behavior patterns obtained by clustering by using an LDA model to obtain an unsupervised driving behavior recognition result. In addition, the invention also constructs a corresponding data acquisition monitoring system, and a complete driving behavior unsupervised mode monitoring system is formed by the mode recognition algorithm, the hardware architecture and the monitoring APP, so that the driving behavior can be more effectively recognized, the real-time monitoring on the driving behavior is realized, and the driving safety of a driver is improved.

Description

Unsupervised driving behavior pattern recognition method and data acquisition monitoring system
Technical Field
The invention relates to the technical field of driving behavior unsupervised pattern recognition and prediction monitoring, in particular to a driving behavior unsupervised pattern recognition method and a data acquisition monitoring system.
Background
In recent years, the automobile industry in China develops rapidly. In 2020, the domestic automobile keeping quantity reaches 2.7 hundred million, the drivers reach 4.1 hundred million, and the total quantity and increment of motor vehicles and drivers are the first in the world. However, with the increase in automobile keeping quantity, a deep level of contradiction in traffic management work is gradually revealed. China is one of the countries with the most road traffic accidents in the world. According to the data published by the national statistical bureau, the death rate of every ten thousand vehicles in China reaches 6.2 remarkably, and the data reaches 4 to 8 times of that of developed countries, so that the supervision task of the national poor driving behavior is far from the task. Therefore, the transportation enterprises and government-related departments should pay high attention to the road traffic safety problem.
Analysis of a large number of accident cases shows that although the causes of traffic accidents are various, human causes are always the main factors of accidents. Abnormal driving behavior of the driver often becomes a cause of a significant traffic accident. Therefore, if the abnormal driving behavior can be effectively identified and the alarm is given to the driver in time, the road traffic accident can be avoided to a certain extent. On the other hand, the traffic department and the operation enterprises can be helped to realize dynamic supervision and evaluation of the safety situation by effectively identifying the abnormal driving behaviors, and the safety management level is improved.
In order to save cost, monitoring and early warning by using information such as vehicle GPS, speed, acceleration and the like become a hot spot of current research. By analyzing relevant data at home and abroad, the method for identifying the abnormal driving behaviors is effectively applied to a series of safety supervision fields such as driver behavior management and the like, the supervision level is improved, and the accident occurrence probability is reduced. Therefore, in order to effectively improve the safety supervision of the traffic industry, deep analysis and application of vehicle monitoring data are imperative. However, the current abnormal driving behavior pattern recognition algorithm is still in a relatively early stage, and the following problems still exist: (1) the goal of the study is to detect isolated events at the traffic level, ignoring other factors that cause traffic pattern changes; (2) at present, supervision algorithms are mostly adopted for abnormal driving behavior recognition, however, marking data are expensive and rare, and related information is lost and generalization performance is reduced. Therefore, it is imperative to explore unsupervised abnormal driving behavior recognition algorithms.
Disclosure of Invention
The invention aims to provide a driving behavior unsupervised mode identification method and a data acquisition monitoring system.
To solve the above technical problem, an embodiment of the present invention provides the following solutions:
in one aspect, a driving behavior unsupervised pattern recognition method is provided, which includes the following steps:
s1, collecting driving data segments of a driver in the driving process and extracting characteristics;
s2, clustering the extracted features by using an improved self-adaptive K-Means clustering algorithm based on Wasserstein distance to obtain a driving behavior mode;
and S3, establishing driving behavior topic description, and matching the driving behavior topic description with all the driving behavior patterns obtained by clustering by using the LDA model to obtain an unsupervised driving behavior recognition result.
Preferably, in the step S1, the driving data segment includes four variables of trajectory curvature, acceleration, speed and steering angle; the extracted features are divided into two categories: statistical and temporal features; the statistical features include: mean, variance, ACF1, remainder autocorrelation, trend, curvature, and entropy; the time characteristics include: maximum mean difference, maximum variance difference, maximum shift of Kullback-Leibler divergence, and remainder variance.
Preferably, the step S2 specifically includes:
s200, inputting feature sample data set Q ═ xs|s＝1，...，N}；
S201, let k be from 1 to kmaxA loop of calculating cluster clusters when the number of clusters is k, wherein k ismaxIs the maximum number of clusters;
s202, randomly selecting k sample points in the feature sample data set Q, and respectively assigning the k sample points to initial clustering centers mu1，μ2，...，μb，...，μk；
S203, making S be a 1-N loop, wherein N represents the number of samples to be clustered, and calculating a sample xbTo each cluster center mubDistance d (x) ofs，μb)；
S204, finding xsAbout the cluster center μbMinimum distance of, will xsIs classified as to mubCluster C with minimum distancebPerforming the following steps;
s205, updating the clustering centers of all clusters, and if the new clustering center is not equal to the original clustering center, updating the clustering centers;
s206, repeating the step S203 to the step S205 until the clustering center is not updated any more;
s207, calculating the current clustering error square sum SSEk；
S208, calculating the difference value between the current clustering error square sum and the last clustering error square sum, namely the reduction amplitude delta SSEk＝SSEk-1-SSEk；
S209, calculating the difference value of the change of the descending amplitude, namely delta SSEk＝ΔSSEk-1-ΔSSEk；
S210, repeating the steps S201 to S209 until k is equal to kmax；
S211, the optimal clustering number is delta SSEkK is taken as the value when the maximum value is reached, and the clustering result is the optimal clustering result;
and S212, outputting the clustering result.
Preferably, the Wasserstein distance is defined as follows:
let θ and iota represent two distributions and the distance between them is calculated by the following formula:
Figure RE-GDA0003048002920000031
where pi (theta, iota) is the set of all possible combined distributions of the theta and iota distributions combined, where pi represents one of the two union distributions denoted by theta and iotaSum distribution, E(m，n)～Π[||m-n||]Calculating the expected distance values of the two distributions under the condition of the current combined distribution pi; after all the joint distributions are calculated, the joint distributions are utilized
Figure RE-GDA0003048002920000032
Obtaining all expected lower bounds, namely Wasserstein distance values;
let the subsequence set be described as X ═ X1，...，xs，...，xN) Wherein
Figure RE-GDA0003048002920000033
N is the total number of samples in the data set, l is the dimension of the subsequence feature, k is the number of clusters, and the cluster center vector is { mu1，μ2，...，μb，...，μk}; sample xsAnd cluster center mubThe distance of (a) is:
Figure RE-GDA0003048002920000041
Π(xs，μb) Is xsAnd mubSet of all possible joint distributions with distributions combined, Π representing xsAnd mubOne of the represented joint distributions; wherein xsf，μbfRespectively represent xsAnd mubThe f-th feature of (1).
Preferably, an elbow method is introduced to calculate the objective function to obtain the optimal clustering number:
Figure RE-GDA0003048002920000042
the best cluster number is obtained when the SSE value reaches a minimum.
Preferably, in the step S3, establishing the driving behavior topic description means performing semantic expression on the driving behavior by extracting driving features; wherein the driving behavior themes comprise 5 normal driving behavior themes and 6 abnormal driving behavior themes; the 5 normal driving behavior themes include: the method comprises the following steps of uniform-speed straight-line driving, mild starting, curve smooth driving, normal lane changing and mild stopping; the 6 abnormal driving behavior themes include: fast lane change, emergency braking, fast overtaking, curve fast driving, emergency acceleration and curve overtaking.
Preferably, in the step S3, the process of identifying the unsupervised driving behavior based on the LDA model includes the following steps:
the LDA model is essentially a three-layer Bayesian network, and unsupervised driving behavior matching is carried out by calculating the feature generation probability; set Ω to { Ω1，...，Ωd，...，ΩkRepresenting driving behavior patterns, wherein the number of the driving behavior patterns is the number k of the extracted clusters, and O is the number of driving behavior themes; the characteristic parameters of the local driving behavior observed in the driving behavior pattern omega are expressed as
Figure RE-GDA0003048002920000043
Wherein U isdThe number of local features is 8, and the 8 features are respectively a track curvature mean value, a track curvature variance, a speed mean value, a speed variance, an acceleration mean value, an acceleration variance, a steering angle mean value and a steering angle variance;
LDA assumes the u-th feature w observed in the d-th driving behavior patternd，uIs based on the potential driving behaviour zd，uGenerated, i.e. assuming the characteristic wd，uIs based on the actual value w
Figure RE-GDA0003048002920000044
Generated, then the distribution of LDA is thetao|dAnd a Dirichlet parameter α, where φoIs a feature distribution with a potential driving topic with Dirichlet parameter β; thus, the generative model of LDA is assumed to be:
θd～Dir(θ；α)
φo～Dir(φ；β)
zd，u～Mult(z；θd)
Figure RE-GDA0003048002920000051
when the feature w is observed ═ wd，uThe probability of generation of the feature is expressed as:
Figure RE-GDA0003048002920000052
the driving behavior theme with the maximum generation probability is the behavior matching result of the driving data segment.
In one aspect, a data acquisition and monitoring system based on the driving behavior unsupervised pattern recognition method is provided, and the data acquisition and monitoring system includes:
the data acquisition subsystem is used for acquiring driving data segments of a driver in the driving process and extracting characteristics, so that local data monitoring, historical data storage and key real-time data uploading are realized;
the cloud platform centralized monitoring subsystem is used for clustering the extracted features by utilizing an improved self-adaptive K-Means clustering algorithm based on Wasserstein distance to obtain a driving behavior mode; and establishing driving behavior topic description, and matching the driving behavior topic description with all the driving behavior patterns obtained by clustering by using an LDA model to obtain an unsupervised driving behavior recognition result.
Preferably, the data acquisition subsystem comprises a data acquisition front end, a serial server, a computer, a display and data acquisition software, and realizes interactive control through a touch screen;
the data acquisition subsystem provides two modes of automatic operation and manual operation, and in the automatic operation mode, the system is started or stopped by automatically receiving a trigger signal; in the manual operation mode, an operator starts or stops the system by pressing an operation button;
the cloud platform centralized monitoring subsystem comprises a cloud data center and is used for monitoring, storing, retrieving, analyzing and releasing data.
Preferably, the data acquisition monitoring system further comprises a vehicle abnormal driving behavior monitoring APP, wherein the vehicle abnormal driving behavior monitoring APP comprises a user registration and login module, an online query module, a region display module and a modification module, and is used for realizing query, display, online update and modification of vehicle driving information so as to be monitored by managers in real time.
The technical scheme provided by the embodiment of the invention has the beneficial effects that at least:
the driving behavior unsupervised mode recognition method provided by the invention does not need to set a threshold value for abnormal driving behaviors, and realizes unsupervised recognition of the driving behaviors by the following two steps aiming at the characteristics of the abnormal driving behaviors: firstly, extracting statistical characteristics and space-time related characteristics as characteristic parameters for driving behavior identification, introducing Wassertein distance as a clustering measurement method for improving clustering accuracy and fully considering structural data characteristics, providing a self-adaptive K-Means algorithm based on the Wassertein distance, introducing a characteristic distribution concept, aggregating driving data segments, and generating a driving behavior pattern; and secondly, setting driving behavior theme description, matching the behavior pattern reflected by the driving characteristics of the actual driving behavior by using the LDA model, and matching the driving behavior by generating probability, thereby realizing the unsupervised marking of the driving behavior. The invention also constructs a corresponding data acquisition monitoring system, and a complete driving behavior unsupervised mode monitoring system is formed by the mode recognition algorithm, the hardware architecture and the monitoring APP, so that the driving behavior can be more effectively recognized, the real-time monitoring on the driving behavior is realized, and the driving safety of a driver is improved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a flow chart of a method for unsupervised pattern recognition of driving behavior according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a feature extraction process provided by an embodiment of the present invention;
fig. 3 is a flowchart of an AKCWD algorithm provided by an embodiment of the present invention;
FIG. 4 is a schematic diagram of an LDA model provided by an embodiment of the present invention;
FIG. 5 is a flow chart of driving behavior recognition provided by an embodiment of the present invention;
FIG. 6 is a schematic diagram of an LDA model driving theme visualization effect provided by the embodiment of the invention;
FIG. 7 is a visual effect diagram of a subject of rapid acceleration driving provided by an embodiment of the invention;
FIG. 8 is a schematic diagram of a data acquisition subsystem provided by an embodiment of the present invention;
fig. 9 is a diagram of a data link and access network structure of a cloud platform centralized monitoring subsystem according to an embodiment of the present invention;
FIG. 10 is a schematic illustration of vehicle monitoring data provided by an embodiment of the present invention;
FIG. 11 is a semantic graph of track segment features provided by embodiments of the present invention;
FIG. 12 is a schematic view of a topic identification of normal lane-change driving behavior provided by an embodiment of the invention;
FIG. 13 is a schematic diagram of the LDA-based driving behavior topic identification effect provided by the embodiment of the present invention;
14 a-14 d are schematic diagrams of speed changes in four driving behavior themes provided by embodiments of the invention;
15 a-15 d are schematic diagrams of acceleration changes in four driving behavior themes provided by embodiments of the invention;
FIGS. 16 a-16 d are schematic diagrams illustrating changes in steering angle in four driving behavior themes provided by an embodiment of the invention;
fig. 17 is an abnormal driving behavior feature analysis diagram provided in the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
An embodiment of the present invention first provides a driving behavior unsupervised pattern recognition method, as shown in fig. 1, the method including the steps of:
s1, collecting driving data segments of a driver in the driving process and extracting characteristics;
s2, clustering the extracted features by using an improved self-adaptive K-Means clustering algorithm based on Wasserstein distance to obtain a driving behavior mode;
and S3, establishing driving behavior topic description, and matching the driving behavior topic description with all the driving behavior patterns obtained by clustering by using the LDA model to obtain an unsupervised driving behavior recognition result.
Specifically, in the step S1, the driving data segment includes four variables of trajectory curvature, acceleration, speed, and steering angle.
A large amount of actually measured driving data need to be referred to during driving behavior recognition, and the independence degree between the data is high. According to the invention, through the analysis of the existing driving behavior recognition method, the heuristic characteristic parameter selection standard is obtained. For example, australian RTA has found that an increase in vehicle speed is often accompanied by an increase in the risk of a traffic accident by studying abnormal driving behaviour. Therefore, the vehicle speed is to be an important consideration for the identification of abnormal driving behavior; in addition, the driver's severe stepping on the accelerator pedal or the brake pedal is often accompanied by abnormal driving behavior, which has a great influence on driving safety.
Therefore, the invention can effectively identify the 'four-urgency' phenomenon (urgent acceleration, urgent deceleration, urgent braking and urgent turning) by analyzing the external expression of the driving behavior and combining the actual requirements of the driving behavior analysis, wherein the selected driving data segment mainly comprises four variables of track curvature, acceleration, speed and steering angle, thereby establishing the abnormal driving behavior identification model.
The feature extraction process converts the raw vehicle driving data segment into a set of features (F1) that capture either contextual information or temporal information around the point in time. These features can be divided into two categories: statistical features and temporal features. The feature extraction flow chart is shown in fig. 2.
The invention utilizes 11 feature extractors to perform feature conversion section by section aiming at four types of variable data of track curvature, speed, acceleration and steering angle. The statistical features describe some basic features around each data point in the time series, where mean, variance, ACF1, remainder autocorrelation, trend, curvature, and entropy are extracted, as shown in table 1; the time characteristic describes the change of the sequence data with time, and the time characteristic is determined by comparing the data in two continuous windows, wherein the maximum displacement and remainder variance of the maximum mean difference, the maximum variance difference, and the Kullback-Leibler divergence are extracted as shown in Table 2.
TABLE 1 statistical characteristics Table
Figure RE-GDA0003048002920000081
TABLE 2 time profiles
Figure RE-GDA0003048002920000091
In summary, each driving data segment comprises four variables of track curvature, acceleration, speed and steering angle, and 44 behavior characteristics can be captured. All these features are normalized to be comparable in different sequences.
Furthermore, in order to research the occurrence situation of abnormal driving behaviors, the segmented data needs to be clustered, driving behavior patterns are mined from a large number of label-free tracks, and a driving behavior scene is provided for unsupervised pattern recognition. In order to fully reflect the characteristics of multi-feature clustering data and measure the distribution difference of each index feature of different clustering clusters, the invention adopts Wassertein distance measurement to replace the Euclidean distance in the traditional K-Means algorithm, and provides a self-adaptive K-Means clustering Algorithm (AKCWD) based on Wassertein distance so as to improve the degree of feature differentiation.
Firstly, the Wasserstein distance is defined as follows:
let θ and iota represent two distributions and the distance between them is calculated by the following formula:
Figure RE-GDA0003048002920000092
where ii (theta, iota) is the set of all possible joint distributions of the combination of the theta and iota distributions, where pi represents one of the joint distributions represented by theta and iota, and E represents one of the joint distributions represented by theta and iota(m，n)～Π[||m-n||]Calculating the expected distance values of the two distributions under the condition of the current combined distribution pi; after all the joint distributions are calculated, the joint distributions are utilized
Figure DEST_PATH_FDA0003088277280000022
Obtaining all expected lower bounds, namely Wasserstein distance values;
let the subsequence set be described as X ═ X1，...，xs，...，xN) Wherein
Figure RE-GDA0003048002920000094
N is the total number of samples in the data set, l is the dimension of the subsequence feature, k is the number of clusters, and the cluster center vector is { mu1，μ2，...，μb，...，μk}; sample xsAnd cluster center mubThe distance of (a) is:
Figure RE-GDA0003048002920000101
Π(xs，μb) Is xsAnd mubSet of all possible joint distributions with distributions combined, Π representing xsAnd mubOne of the represented joint distributions; wherein xsf，μbfRespectively represent xsAnd mubThe f-th feature of (1).
To obtain the best clustering number, an elbow method is introduced to calculate an objective function so as to obtain the best clustering number:
Figure RE-GDA0003048002920000102
the best cluster number is obtained when the SSE value reaches a minimum.
Specifically, the AKCWD algorithm in step S2 includes the following steps:
s200, inputting feature sample data set Q ═ xs|s＝1，...，N}；
S201, let k be from 1 to kmaxA loop of calculating cluster clusters when the number of clusters is k, wherein k ismaxIs the maximum number of clusters;
s202, randomly selecting k sample points in the feature sample data set Q, and respectively assigning the k sample points to initial clustering centers mu1，μ2，...，μb，...，μk；
S203, making S be a 1-N loop, wherein N represents the number of samples to be clustered, and calculating a sample xbTo each cluster center mubDistance d (x) ofs，μb)；
S204, finding xsAbout the cluster center μbMinimum distance of, will xsIs classified as to mubCluster C with minimum distancebPerforming the following steps;
s205, updating the clustering centers of all clusters, and if the new clustering center is not equal to the original clustering center, updating the clustering centers;
s206, repeating the step S203 to the step S205 until the clustering center is not updated any more;
s207, calculating the current clustering error square sum SSEk；
S208, calculating the difference value between the current clustering error square sum and the last clustering error square sum, namely the reduction amplitude delta SSEk＝SSEk-1-SSEk；
S209, calculating the difference value of the change of the descending amplitude, namely delta SSEk＝ΔSSEk-1-ΔSSEk；
S210, repeating the steps S201 to S209, until k is kmax；
S211, the optimal clustering number is delta SSEkK is taken as the value when the maximum value is reached, and the clustering result is the optimal clustering result;
and S212, outputting the clustering result.
The algorithm flow chart is shown in fig. 3.
Further, in step S3, a driving behavior theme is first set, that is, a behavior description of a driving behavior scene is given according to the characteristics of the trajectory curvature, the speed, the acceleration, and the steering angle. Then the driving behavior theme is matched with the driving behavior patterns obtained by all the clusters, and the most similar pattern is selected as a representative of each cluster. And finally, marking the cluster with the highest matching degree as the driving behavior theme as an identification result of the unsupervised driving behavior.
The essence of the unsupervised pattern recognition of driving behavior is to convert the continuous driving behavior data into a sequence of "driving characteristics" in natural language; for example, "accelerate quickly from low speed", "cruise at high speed", and "brake stop at high speed", thereby recognizing the driver's behavior. Therefore, in the pattern recognition process, a large number of driving features can be extracted to realize semantic expression of driving behaviors, and such a recognition manner is effective for long-term prediction of driving behaviors, which does not require specifying a specific detection threshold and can be analyzed by using behavior semantics.
The invention provides some common driving behavior theme descriptions, and the number of the common driving behavior theme descriptions is 11, wherein the 11 driving behavior themes comprise 5 normal driving behavior themes and 6 abnormal driving behavior themes. Wherein, 5 kinds of normal driving behavior themes include: the method comprises the following steps of uniform-speed straight-line driving, mild starting, curve smooth driving, normal lane changing and mild stopping; the 6 abnormal driving behavior themes include: fast lane change, emergency braking, fast overtaking, curve fast driving, emergency acceleration and curve overtaking. The theme basically covers frequent driving behaviors of a driver in the driving process, has high representativeness and can reflect the driving style of the driver in the driving process. The driving behavior theme is described in table 3.
TABLE 3 subject description of driving behavior
Figure RE-GDA0003048002920000111
Figure RE-GDA0003048002920000121
Further, the present invention uses a document topic Generation model (LDA) to complete an unsupervised driving behavior recognition algorithm. The LDA model is essentially a three-layer Bayesian network, and unsupervised driving behavior matching is carried out by calculating the feature generation probability; set Ω to { Ω1，...，Ωd，...，ΩkAnd represents driving behavior patterns, wherein the number of the driving behavior patterns is the number k of the extracted clusters, and O is the number of driving behavior topics, namely the behavior topics given in Table 3. The characteristic parameters of the local driving behavior observed in the driving behavior pattern omega are expressed as
Figure RE-GDA0003048002920000122
Wherein U isdThe number of local features is 8, and the number of the local features is the track curvature mean, the track curvature variance, the velocity mean, the velocity variance, the acceleration mean, the acceleration variance, the steering angle mean, and the steering angle variance.
Here, LDA assumes the u-th feature w observed in the d-th driving behavior patternd，uIs based on the potential driving behaviour zd，uGenerated, i.e. assuming the characteristic wd，uIs based on the actual value w
Figure RE-GDA0003048002920000123
Generated, then the distribution of LDA is thetao|dAnd a Dirichlet parameter α, where φoIs a feature distribution with a potential driving topic with Dirichlet parameter β; thus, the generative model of LDA is assumed to be:
θd～Dir(θ；α) (4)
φo～Dir(φ；β) (5)
zd，u～Mult(z；θd) (6)
Figure RE-GDA0003048002920000124
when the feature w is observed ═ wd，uThe probability of generation of the feature is expressed as:
Figure RE-GDA0003048002920000125
the schematic diagram of the algorithm model is shown in fig. 4, and the driving behavior theme with the maximum generated probability is the behavior matching result of the driving data segment.
The driving behavior recognition flowchart is shown in fig. 5. The driving behavior scene comprises physical characteristics such as an acceleration mean value, an acceleration variance, a speed mean value, a speed variance, a steering angle mean value, a steering angle variance, a track curvature mean value and a track curvature variance, the characteristics are sequentially matched with the characteristics of the driving behavior theme, and the generation probability is calculated. Finally, the semantic meaning of the driving behavior theme with the maximum matching probability with the actual driving behavior scene is found and is given to the driving behavior scene to serve as the label of the driving behavior scene.
In order to obtain a more refined matching effect, each feature of the driving behavior scene is normalized to a numerical value between [0, 1] and is discretized into 20 intervals, the width of each discrete interval is set to be 0.05, and the theme label of the driving behavior scene is determined by estimating the salient features of the driving behavior scene. For example, the 20 discrete intervals of the speed mean are divided into five groups of "stationary" (0 interval), "very slow running" (1 st to 5 th intervals), "low speed running" (6 th to 10 th intervals), "medium speed running" (11 th to 15 th intervals), and "high speed running" (16 th to 20 th intervals).
The semantic labels corresponding to different discrete intervals are obtained by translating the 8 driving behavior physical characteristics of the driving behavior scene. And matching the semantic labels with the semantics of the driving behavior topic description in the table 3 to complete the conversion from the unknown behaviors to the known behaviors. The feature semantic translations are shown in table 4.
TABLE 4 characteristic semantic translation Table
Figure RE-GDA0003048002920000131
By integrating the behavior descriptions of the salient features, the behavior description of each driving theme is generated, and the visualization effect of the behavior description in the parallel coordinate system is shown in fig. 6. The vertical axis represents eight coordinate systems, namely a speed mean, a speed variance, an acceleration mean, an acceleration variance, a track corner mean, a track corner variance, a track curvature mean, a track curvature variance and a coordinate system value range of [0, 20], and represents 20 discrete intervals. As can be seen from the figure, no specific threshold setting is needed, and the invention realizes the unsupervised identification of the driving behavior by utilizing the LDA model and combining the characteristic semantic translation table.
To better represent a particular driving behavior, fig. 7 shows the visualization effect of a sudden acceleration driving behavior in a parallel coordinate system. It can be seen from the figure that the speed mean value of the rapid acceleration driving behavior is distributed in the interval [5, 15], the speed variance is distributed in the interval [10, 15], the acceleration mean value is distributed in the interval [15, 20], the acceleration variance is distributed in the interval [15, 20], the track corner mean value is distributed in the interval [1, 0], the track corner variance is distributed in the interval [1, 10], the track curvature mean value is distributed in the interval [1, 20], the track curvature variance is distributed in the interval [1, 20], and the identification of the rapid acceleration behavior is completed according to the combined action of 8 driving behavior characteristics.
Correspondingly, an embodiment of the present invention further provides a data acquisition monitoring system based on the above driving behavior unsupervised pattern recognition method, and the system includes:
the data acquisition subsystem is used for acquiring driving data segments of a driver in the driving process and extracting characteristics, so that local data monitoring, historical data storage and key real-time data uploading are realized;
the cloud platform centralized monitoring subsystem is used for clustering the extracted features by utilizing an improved self-adaptive K-Means clustering algorithm based on Wasserstein distance to obtain a driving behavior mode; and establishing driving behavior topic description, and matching the driving behavior topic description with all the driving behavior patterns obtained by clustering by using an LDA model to obtain an unsupervised driving behavior recognition result.
The data acquisition subsystem is responsible for acquiring data of the abnormal driving behaviors of the multivariate time sequence, and embeds a plurality of IEC 60870-5 data communication protocols of 101, 102, 103, 104, Modbus, CDT, DISA and the like by taking C + + as a development language; the modeling meets the requirements of an interface reference model, a Common Information Model (CIM) and a Component Interface Specification (CIS) in IEC 61970, meets the international standard, and can be used as a middleware to be seamlessly integrated with each system.
The cloud platform centralized monitoring subsystem acquires real-time monitoring data from the data acquisition subsystem, a communication protocol between the two subsystems can adopt a standard IEC104 protocol or other protocols, the real-time data acquisition frequency supports the second level according to the protocol requirement, and modes such as variable quantity uploading, circular uploading and calling can be supported.
Fig. 8 is a schematic diagram of a data acquisition subsystem provided in an embodiment of the present invention, and fig. 9 is a structure diagram of a data link and an access network of a cloud platform centralized monitoring subsystem provided in an embodiment of the present invention. The data acquisition subsystem comprises a data acquisition front end, a serial server, a computer, a display and data acquisition software, and realizes interactive control through a touch screen; the data acquisition subsystem provides two modes of automatic operation and manual operation, and in the automatic operation mode, the system is started or stopped by automatically receiving a trigger signal; in the manual operation mode, the operator starts or stops the system operation by pressing the operation button. The cloud platform centralized monitoring subsystem comprises a cloud data center and is used for monitoring, storing, retrieving, analyzing and releasing data.
The cloud data center stores real-time data of a monitoring process by using a real-time database and provides retrieval service, stores static data of a business process by using a business SQL database (Oracle or MYSQL) and provides retrieval service, and uses an application mode of a big data analysis platform for analyzing data in an off-line manner, so that the real-time monitoring requirement of the running state of vehicle running monitoring data can be supported, and various application-oriented and theme-oriented analysis requirements can be met. The database design organizes the management of the database according to an object-oriented mode which accords with a natural mode of human thinking, and the speed and the efficiency of data retrieval and searching are improved.
The real-time database system is novel database management system software, and is suitable for acquisition, storage, retrieval and release of massive real-time/historical data based on a high-speed database engine developed by a 64-bit system and an advanced distributed cluster architecture, has good horizontal expansion capability and high availability, and can process dynamic data which rapidly changes along with time.
The technical indexes of the real-time database system are as follows: 1) scale: the scale of more than 100 ten thousand labels is supported. 2) Speed: high-speed real-time, historical data retrieval capability; real-time data millisecond-level response; historical data for the monthly span retrieves second-level responses. 3) The storage type is as follows: flexible and diverse multiple data types are supported: a Boolean type; integer (8 bits/16 bits/32 bits/64 bits); floating point (32 bit/64 bit); date type data (time stamp); and others. 4) Efficient data compression: the method supports a plurality of lossless and lossy compression modes, greatly improves the storage efficiency, and simultaneously improves the analysis and retrieval speed in mass historical data; the system supports two-stage compression capacity, can effectively improve the utilization rate of network resources, reduces the requirement on hardware, provides multi-stage buffering and improves the high availability of the system; the configuration of point granularity is supported, and a compression algorithm can be flexibly selected and matched according to the characteristics of different data; compression ratios as high as tens of times; advanced distributed architecture: the cluster-based hot backup mechanism enables the system to have high availability of data; the distributed redundant storage architecture enables the system to have high elastic expansion capability; the disaster recovery mechanism fully ensures high safety of vehicle driving monitoring data. 5) Flexible data access interface: a C/C + +/JAVA/JSON interface is provided for third party calls and writing of data.
Further, data acquisition monitored control system still includes vehicle unusual driving behavior control APP, vehicle unusual driving behavior control APP includes user registration and login module, online query module, regional display module and modification module for realize vehicle driving information's inquiry, demonstration, online update, modification, for managers real time monitoring.
The APP uses HBuilderX as a development tool, is developed by using HTML5+ CSS + JavaScript language, and is built by an MUI front-end framework. The APP realizes the registration and login function, obtains real-time vehicle running monitoring data, has the functions of time series feature extraction, cluster analysis, pattern recognition, theme matching and the like, and displays the result in a visual interface.
The APP comprises the following components:
and the client performs development and design by using an MUI front-end framework and performs front-end development by using HTML5, CSS and JavaScript languages.
And the server side is developed by using a ThinkJS server side framework and is matched with a MySQL database, so that the functions of registration, login verification, data transmission, addition, modification and deletion can be realized.
And the system management background is developed by using HTML5, CSS and JavaScript languages and is used for managing the database.
The MUI front-end framework (based on HTML5, CSS and JavaScript) is used for designing and developing an Android client and is also used for developing a system management background. The ThinkJS server side framework (based on NodeJS) is used for designing a logic interface to provide service for the client side and the system management background, and corresponding functions are realized. The database MySQL is used for storing vehicle running monitoring data and user information.
Furthermore, the self-adaptive K-Means algorithm based on Wasserstein distance and the abnormal behavior recognition algorithm based on the LDA algorithm are verified by utilizing the driving data of part of vehicles in the vehicle monitoring data.
The vehicle monitoring data is from experimental data of a certain automobile company, and records the types of abnormal driving behaviors of drivers with different driving styles under different traffic conditions (including different road conditions and weather). As shown in table 5:
TABLE 5 Driving behavior monitoring watch
Figure RE-GDA0003048002920000161
Figure RE-GDA0003048002920000171
The raw vehicle monitoring data includes multiple dimensions such as GPS track points, acceleration, speed, accelerator pedal pressure, steering angle, battery voltage, etc. Because the invention aims to learn and identify the abnormal driving behavior of the driver through vehicle monitoring data, only data parameters with the control intention of the driver, such as GPS track points, speed, acceleration, steering angle and the like, are adopted. The data sample period is 0.1 seconds, with each set of samples containing 10000 to 20000 sample points. Fig. 10 shows a schematic view of vehicle monitoring data, which shows recorded data of a vehicle during driving, and the recorded data are respectively labeled with a data serial number, a vehicle number, a driver number, a positioning time, a vehicle speed, an acceleration, a position and corner information, wherein the vehicle speed, the acceleration, the position and the corner information are analyzed as a main basis for identifying abnormal driving behaviors.
The method adopts an SD value index (SD) index as an evaluation index of the final clustering effect. The SD evaluation algorithm relies on several metrics, respectively the variance of the data set and the variance of each cluster.
Let k denote the number of clusters, μbRepresenting the center of the cluster, b is more than 1 and less than k, N represents the number of cluster samples, and the set of all cluster samples is Q ═ xs1,., N }, each cluster set being Cb，NbIs CbThe number of elements in.
First, the sample mean is calculated:
Figure RE-GDA0003048002920000172
the data sets are divided according to dimensions, and the variance of the data sets of the f dimension is defined as follows:
Figure RE-GDA0003048002920000173
wherein
Figure RE-GDA0003048002920000174
Is the mean of the f-th dimension
Figure RE-GDA0003048002920000175
According to the definition above, the f-th dimension of the cluster CbVariance of (2)
Figure RE-GDA0003048002920000176
The definition is as follows:
Figure RE-GDA0003048002920000181
for all cluster clusters, the measure of cluster compactness is defined as follows:
Figure RE-GDA0003048002920000182
if the cluster compactness of the cluster is better, then the result should be smaller than the variance of the data set.
Further, the degree of dispersion between the respective clusters is defined as follows as a whole:
Figure RE-GDA0003048002920000183
wherein DmaxRepresents the maximum distance between the centers of any two clusters, DminRepresenting the minimum distance between the centers of any two clusters.
Finally, SD is defined as follows:
SD(k)＝αScat(k)+Dist(k) (14)
where α is a scaling factor.
The smaller the SD value result, the better the clustering effect.
Considering that drivers often show different driving behavior tendencies under different driving environments, the invention selects three driving data fragment sets with large driving environment difference for cluster analysis, wherein the three driving data fragment sets are respectively the driving environment of the first class weather and the expressway, the driving environment of the first class weather and the arterial road and the driving environment of the third class weather and the expressway. Wherein, the first kind of weather is good weather, and the third kind of weather is rainstorm, snowstorm and the like.
Table 6, table 7, and table 8 are the clustering experimental results of the first-class weather, expressway, the first-class weather, trunk road, and the third-class weather, expressway, respectively.
TABLE 6 weather and expressway clustering experiment
Figure RE-GDA0003048002920000184
Figure RE-GDA0003048002920000191
TABLE 7 clustering experiment of weather and trunk road
Figure RE-GDA0003048002920000192
TABLE 8 clustering experiment of three types of weather and expressway
Figure RE-GDA0003048002920000193
From tables 6 to 8, it can be found that the adaptive K-Means clustering algorithm based on Wasserstein distance provided by the invention obtains better clustering effect under the evaluation index of SD. In addition, under different weather and road environments, the optimal cluster number is often in different distributions, and is combined with actual information, so that it can be speculated that in the case of snowstorm weather, a driver may also present different driving styles due to the limitation of weather, and different driving behavior patterns are caused.
In order to mark an actual driving scene as a specific driving behavior theme, the driving behavior theme matching process based on the LDA model is analyzed by taking 11 driving behavior scenes obtained by clustering under driving environments of express roads and the like as an example. Here, 4 typical driving behavior scenes are intercepted for feature semantic translation, and as shown in fig. 11, eight behavior features are extracted from each driving behavior scene, and the behavior features are speed average values respectively
Figure RE-GDA0003048002920000194
Velocity variance δ v, mean acceleration
Figure RE-GDA0003048002920000195
Acceleration variance delta a and steering angle mean
Figure RE-GDA0003048002920000196
Steering angle variance delta theta, mean of track curvature
Figure RE-GDA0003048002920000197
The trajectory curvature variance δ c is discretized into 20 discrete values with a width of 0.05, and further divided into 5 groups of intervals 0, wherein the interval 1 is within [1, 5]]The interval 2 ∈ [6, 10]]The interval 3 ∈ [11, 15]]The interval 4 ∈ [16, 20]]。
And converting the behavior characteristic interval numerical value into behavior description according to the characteristic semantic translation table of the table 4, wherein the discrete value is the behavior characteristic interval numerical value. And translated into a behavioral description according to the size of its eigenvalues. Table 9 gives the translation results from the features to the behavioral description for 11 driving behavior scenarios.
TABLE 9 express way, driving behavior translation under one type of weather
Figure RE-GDA0003048002920000201
Fig. 12 shows a driving behavior semantic translation process of the driving scenario C1 representing the driving behavior sample 1. Firstly, calculating the behavior characteristic values of the samples, secondly, translating each behavior characteristic value into a driving behavior description, integrating 8 characteristics to obtain a final driving behavior description, matching the final driving behavior description with the driving behavior theme, wherein the driving behavior theme with the highest matching degree is a normal lane change, and the final behavior recognition result is obtained.
Fig. 13 shows the recognition effect of different driving behavior topics based on LDA more intuitively by using parallel coordinate systems, where the vertical axis has nine coordinate systems, the first eight coordinate systems are respectively the speed mean, the speed variance, the acceleration mean, the acceleration variance, the trajectory corner mean, the trajectory corner variance, the trajectory curvature mean, and the trajectory curvature variance, and each coordinate system takes the value of [0, 20] and represents 20 discrete intervals; the ninth coordinate system corresponds to recognition results of different driving behavior topics.
In order to better show different trends of different driving behaviors on different parameters to verify the recognition effect of the driving theme, the trend changes of three characteristics of speed, acceleration and steering angle in four driving behavior themes of sharp lane change, sharp acceleration, acceleration and deceleration and sharp left turn are respectively shown in fig. 14 a-14 d, fig. 15 a-15 d and fig. 16 a-16 d. Such as sharp lane change behavior, acceleration followed by deceleration is presented in the figure with a sharp change in steering angle midway through it, consistent with the behavior expectation, consistent with the driving behavior theme description of table 4.
The method is characterized in that 10093 driving behavior scene segments are synthesized, and after the driving behavior scene segments are matched with the driving behavior themes, 626 sudden braking/sudden acceleration behaviors, 516 rapid lane changing behaviors, 501 rapid speed overtaking behaviors, 313 overspeed behaviors and 278 curve rapid driving behaviors are shared. Fig. 17 shows the distribution of four driving behaviors of rapid acceleration and deceleration, rapid turning, rapid passing and rapid lane change, which are presented in a 3d scene by the characteristics of acceleration, track and steering angle. It can be seen that for the behavior of sudden lane change, the acceleration is low, the rotation angle value is high, and the track value is extremely high, which is very in line with the behavior topic description given by the prior knowledge. For the sharp turning behavior and the sharp acceleration/rapid deceleration behavior, the acceleration value is high, the steering angle value is low, the track characteristics are widely distributed, and the behavior theme description given by the prior knowledge is also met.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims (10)

1. An unsupervised pattern recognition method for driving behavior, characterized by comprising the steps of:
s1, collecting driving data segments of a driver in the driving process and extracting characteristics;
s2, clustering the extracted features by using an improved self-adaptive K-Means clustering algorithm based on Wasserstein distance to obtain a driving behavior mode;
and S3, establishing driving behavior topic description, and matching the driving behavior topic description with all the driving behavior patterns obtained by clustering by using the LDA model to obtain an unsupervised driving behavior recognition result.
2. The unsupervised pattern recognition method of driving behavior of claim 1, characterized in that in step S1, the driving data segments include four variables of trajectory curvature, acceleration, speed, steering angle; the extracted features are divided into two categories: statistical and temporal features; the statistical features include: mean, variance, ACF1, remainder autocorrelation, trend, curvature, and entropy; the time characteristics include: maximum mean difference, maximum variance difference, maximum shift of Kullback-Leibler divergence, and remainder variance.
3. The unsupervised pattern recognition method of driving behavior according to claim 1, characterized in that said step S2 specifically comprises:
s200, inputting feature sample data set Q ═ xs|s＝1,...,N}；
S201, let k be from 1 to kmaxA loop of calculating cluster clusters when the number of clusters is k, wherein k ismaxIs the maximum number of clusters;
S202、randomly selecting k sample points in the characteristic sample data set Q, and respectively assigning the k sample points to initial clustering centers mu1,μ2,...,μb,...,μk；
S203, making S be a 1-N loop, wherein N represents the number of samples to be clustered, and calculating a sample xbTo each cluster center mubDistance d (x) ofs,μb)；
S204, finding xsAbout the cluster center μbMinimum distance of, will xsIs classified as to mubCluster C with minimum distancebPerforming the following steps;
s205, updating the clustering centers of all clusters, and if the new clustering center is not equal to the original clustering center, updating the clustering centers;
s206, repeating the step S203 to the step S205 until the clustering center is not updated any more;
s207, calculating the current clustering error square sum SSEk；
S208, calculating the difference value between the current clustering error square sum and the last clustering error square sum, namely the reduction amplitude delta SSEk＝SSEk-1-SSEk；
S209, calculating the difference value of the change of the descending amplitude, namely delta SSEk＝ΔSSEk-1-ΔSSEk；
S210, repeating the steps S201 to S209 until k is equal to kmax；
S211, the optimal clustering number is delta SSEkK is taken as the value when the maximum value is reached, and the clustering result is the optimal clustering result;
and S212, outputting the clustering result.
4. The unsupervised pattern recognition method of driving behavior according to claim 3, characterized in that the Wasserstein distance is defined as follows:
suppose that
Figure RE-FDA0003048002910000021
And
Figure RE-FDA0003048002910000022
two distributions are represented, the distance between them being calculated by the following formula:
Figure RE-FDA0003048002910000023
wherein
Figure RE-FDA0003048002910000024
Is that
Figure RE-FDA0003048002910000025
And
Figure RE-FDA0003048002910000026
set of all possible joint distributions with the distributions combined, with pi representing
Figure RE-FDA0003048002910000027
And
Figure RE-FDA0003048002910000028
one of the represented joint distributions, E(m,n)～Π[||m-n||]Calculating the expected distance values of the two distributions under the condition of the current combined distribution pi; after all the joint distributions are calculated, the joint distributions are utilized
Figure 674053DEST_PATH_FDA0003088277280000022
Obtaining all expected lower bounds, namely Wasserstein distance values;
let the subsequence set be described as X ═ X1,...,xs,...,xN) Wherein x iss＝(xs1,xs2,...,xsf,...,xsl)TN is the total number of samples in the data set, l is the dimension of the subsequence feature, k is the number of clusters, and the cluster center vector is { mu }1,μ2,...,μb,...,μk}; sample xsAnd cluster center mubThe distance of (a) is:
Figure RE-FDA00030480029100000210
Π(xs,μb) Is xsAnd mubSet of all possible joint distributions with distributions combined, Π representing xsAnd mubOne of the represented joint distributions; wherein xsf,μbfRespectively represent xsAnd mubThe f-th feature of (1).
5. The unsupervised pattern recognition method of driving behavior of claim 4, characterized in that an elbow method is introduced to calculate the objective function to obtain the optimal cluster number:
Figure RE-FDA0003048002910000031
the best cluster number is obtained when the SSE value reaches a minimum.
6. The unsupervised pattern recognition method of the driving behavior according to claim 1, wherein in the step S3, establishing the driving behavior topic description means semantically expressing the driving behavior by extracting driving features; wherein the driving behavior themes comprise 5 normal driving behavior themes and 6 abnormal driving behavior themes; the 5 normal driving behavior themes include: the method comprises the following steps of uniform-speed straight-line driving, mild starting, curve smooth driving, normal lane changing and mild stopping; the 6 abnormal driving behavior themes include: fast lane change, emergency braking, fast overtaking, curve fast driving, emergency acceleration and curve overtaking.
7. The unsupervised pattern recognition method of driving behavior according to claim 1, wherein in step S3, the unsupervised driving behavior recognition process based on LDA model is as follows:
the LDA model is substantiallyA three-layer Bayesian network for unsupervised driving behavior matching by calculating feature generation probability; set Ω to { Ω1,...,Ωd,...,ΩkRepresenting driving behavior patterns, wherein the number of the driving behavior patterns is the number k of the extracted clusters, and O is the number of driving behavior themes; the characteristic parameters of the local driving behavior observed in the driving behavior pattern omega are expressed as
Figure RE-FDA0003048002910000032
Wherein U isdThe number of local features is 8, and the 8 features are respectively a track curvature mean value, a track curvature variance, a speed mean value, a speed variance, an acceleration mean value, an acceleration variance, a steering angle mean value and a steering angle variance;
LDA assumes the u-th feature w observed in the d-th driving behavior patternd,uIs based on the potential driving behaviour zd,uGenerated, i.e. assuming the characteristic wd,uIs based on the actual value w
Figure RE-FDA0003048002910000033
Generated, then the distribution of LDA is thetao|dAnd a Dirichlet parameter α, where φoIs a feature distribution with a potential driving topic with Dirichlet parameter β; thus, the generative model of LDA is assumed to be:
θd～Dir(θ；α)
φo～Dir(φ；β)
zd,u～Mult(z；θd)
Figure RE-FDA0003048002910000041
when the feature w is observed ═ wd,uThe probability of generation of the feature is expressed as:
Figure RE-FDA0003048002910000042
the driving behavior theme with the maximum generation probability is the behavior matching result of the driving data segment.
8. A data acquisition monitoring system based on the driving behavior unsupervised pattern recognition method according to any one of claims 1 to 7, characterized in that the data acquisition monitoring system comprises:
the data acquisition subsystem is used for acquiring driving data segments of a driver in the driving process and extracting characteristics, so that local data monitoring, historical data storage and key real-time data uploading are realized;
the cloud platform centralized monitoring subsystem is used for clustering the extracted features by utilizing an improved self-adaptive K-Means clustering algorithm based on Wasserstein distance to obtain a driving behavior mode; and establishing driving behavior topic description, and matching the driving behavior topic description with all the driving behavior patterns obtained by clustering by using an LDA model to obtain an unsupervised driving behavior recognition result.
9. The data acquisition monitoring system of claim 8, wherein the data acquisition subsystem comprises a data acquisition front end, a serial server, a computer, a display and data acquisition software, and realizes interactive control through a touch screen;
the data acquisition subsystem provides two modes of automatic operation and manual operation, and in the automatic operation mode, the system is started or stopped by automatically receiving a trigger signal; in the manual operation mode, an operator starts or stops the system by pressing an operation button;
the cloud platform centralized monitoring subsystem comprises a cloud data center and is used for monitoring, storing, retrieving, analyzing and releasing data.
10. The data acquisition and monitoring system of claim 8, further comprising a vehicle abnormal driving behavior monitoring APP, wherein the vehicle abnormal driving behavior monitoring APP comprises a user registration and login module, an online query module, an area display module and a modification module, and is used for realizing query, display, online update and modification of vehicle driving information so as to be monitored by a manager in real time.
CN202110216457.6A 2021-02-26 2021-02-26 Driving behavior unsupervised mode identification method and data acquisition monitoring system Active CN113159105B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110216457.6A CN113159105B (en) 2021-02-26 2021-02-26 Driving behavior unsupervised mode identification method and data acquisition monitoring system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110216457.6A CN113159105B (en) 2021-02-26 2021-02-26 Driving behavior unsupervised mode identification method and data acquisition monitoring system

Publications (2)

Publication Number Publication Date
CN113159105A true CN113159105A (en) 2021-07-23
CN113159105B CN113159105B (en) 2023-08-08

Family

ID=76883511

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110216457.6A Active CN113159105B (en) 2021-02-26 2021-02-26 Driving behavior unsupervised mode identification method and data acquisition monitoring system

Country Status (1)

Country Link
CN (1) CN113159105B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113487936A (en) * 2021-07-26 2021-10-08 广州星锐传媒信息科技有限公司 Simulation driving judgment method based on multi-dimensional continuous signal analysis
CN113505852A (en) * 2021-07-28 2021-10-15 云南电网有限责任公司电力科学研究院 Low-voltage user phase identification method based on principal component analysis and EM distance
CN115204323A (en) * 2022-09-16 2022-10-18 华智生物技术有限公司 Seed multi-feature based clustering and synthesis method, system, device and medium
CN116975731A (en) * 2023-08-08 2023-10-31 山东大学 Cross-domain cutter damage monitoring method and system based on transfer learning

Citations (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2015217707A (en) * 2014-05-14 2015-12-07 株式会社豊田中央研究所 Vehicle control device and program
WO2016009266A1 (en) * 2014-07-16 2016-01-21 Schlumberger Canada Limited Systems and methods for estimating time of flight for an acoustic wave
CN107284442A (en) * 2017-05-15 2017-10-24 北京理工大学 A kind of longitudinally controlled method of negotiation of bends for automatic driving vehicle
CN108985488A (en) * 2018-06-06 2018-12-11 福建工程学院 The method predicted to individual trip purpose
CN109086794A (en) * 2018-06-27 2018-12-25 武汉理工大学 A kind of driving behavior mode knowledge method based on T-LDA topic model
CN109558482A (en) * 2018-07-27 2019-04-02 中山大学 A kind of parallel method of the text cluster model PW-LDA based on Spark frame
US20190100216A1 (en) * 2017-09-29 2019-04-04 Denso International America, Inc. Risk Assessment System For Assessing Current Driver Behavior Relative to Past Behavior and Behaviors of Other Drivers
US20190109810A1 (en) * 2011-05-12 2019-04-11 Jeffrey Alan Rapaport Social-topical adaptive networking (stan) system allowing for group based contextual transaction offers and acceptances and hot topic watchdogging
CN109849926A (en) * 2019-01-29 2019-06-07 北京市交通信息中心 A kind of method and system for distinguishing taxi and whether giving other people drive
CN109918476A (en) * 2019-01-26 2019-06-21 北京工业大学 A kind of subject retrieval method based on topic model
CN110171426A (en) * 2019-05-23 2019-08-27 重庆大学 A kind of recognition methods of driver's dangerous driving behavior and system based on smart machine
CN110175557A (en) * 2019-05-24 2019-08-27 北京交通大学 Driver's driving behavior method for identifying and classifying and system based on driving characteristics group
CN110471411A (en) * 2019-07-26 2019-11-19 华为技术有限公司 Automatic Pilot method and servomechanism
US20200133269A1 (en) * 2018-10-30 2020-04-30 The Regents Of The University Of Michigan Unsurpervised classification of encountering scenarios using connected vehicle datasets
CN111126438A (en) * 2019-11-22 2020-05-08 北京理工大学 Driving behavior recognition method and system
CN111461185A (en) * 2020-03-19 2020-07-28 哈尔滨工程大学 Driving behavior analysis method based on improved K-means
CN111724423A (en) * 2020-06-03 2020-09-29 西安交通大学 Non-rigid body registration method of differential homoembryo based on fluid divergence loss
CN111752254A (en) * 2019-03-28 2020-10-09 深圳市商汤科技有限公司 Test method, device, medium and equipment for trajectory tracking controller
CN112339747A (en) * 2020-10-30 2021-02-09 上海欧菲智能车联科技有限公司 Automatic parking track generation method and device, electronic equipment and storage medium

Patent Citations (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190109810A1 (en) * 2011-05-12 2019-04-11 Jeffrey Alan Rapaport Social-topical adaptive networking (stan) system allowing for group based contextual transaction offers and acceptances and hot topic watchdogging
JP2015217707A (en) * 2014-05-14 2015-12-07 株式会社豊田中央研究所 Vehicle control device and program
WO2016009266A1 (en) * 2014-07-16 2016-01-21 Schlumberger Canada Limited Systems and methods for estimating time of flight for an acoustic wave
CN107284442A (en) * 2017-05-15 2017-10-24 北京理工大学 A kind of longitudinally controlled method of negotiation of bends for automatic driving vehicle
US20190100216A1 (en) * 2017-09-29 2019-04-04 Denso International America, Inc. Risk Assessment System For Assessing Current Driver Behavior Relative to Past Behavior and Behaviors of Other Drivers
CN108985488A (en) * 2018-06-06 2018-12-11 福建工程学院 The method predicted to individual trip purpose
CN109086794A (en) * 2018-06-27 2018-12-25 武汉理工大学 A kind of driving behavior mode knowledge method based on T-LDA topic model
CN109558482A (en) * 2018-07-27 2019-04-02 中山大学 A kind of parallel method of the text cluster model PW-LDA based on Spark frame
US20200133269A1 (en) * 2018-10-30 2020-04-30 The Regents Of The University Of Michigan Unsurpervised classification of encountering scenarios using connected vehicle datasets
CN109918476A (en) * 2019-01-26 2019-06-21 北京工业大学 A kind of subject retrieval method based on topic model
CN109849926A (en) * 2019-01-29 2019-06-07 北京市交通信息中心 A kind of method and system for distinguishing taxi and whether giving other people drive
CN111752254A (en) * 2019-03-28 2020-10-09 深圳市商汤科技有限公司 Test method, device, medium and equipment for trajectory tracking controller
CN110171426A (en) * 2019-05-23 2019-08-27 重庆大学 A kind of recognition methods of driver's dangerous driving behavior and system based on smart machine
CN110175557A (en) * 2019-05-24 2019-08-27 北京交通大学 Driver's driving behavior method for identifying and classifying and system based on driving characteristics group
CN110471411A (en) * 2019-07-26 2019-11-19 华为技术有限公司 Automatic Pilot method and servomechanism
CN111126438A (en) * 2019-11-22 2020-05-08 北京理工大学 Driving behavior recognition method and system
CN111461185A (en) * 2020-03-19 2020-07-28 哈尔滨工程大学 Driving behavior analysis method based on improved K-means
CN111724423A (en) * 2020-06-03 2020-09-29 西安交通大学 Non-rigid body registration method of differential homoembryo based on fluid divergence loss
CN112339747A (en) * 2020-10-30 2021-02-09 上海欧菲智能车联科技有限公司 Automatic parking track generation method and device, electronic equipment and storage medium

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
D.R.KUMAR RAJA等: ""Diversifying personalized mobile multimedia application recommendations through the Latent Dirichlet Allocation and clustering optimization"", 《MULTIMEDIA TOOLS AND APPLICATIONS》, pages 24047 - 24066 *
R.J.HYNDMAN等: ""Large-scale unusual time series detection"", 《2015 IEEE INTERNATIONAL CONFERENCE ON DATA MINING WORKSHOP》, pages 1616 - 1619 *
ZHONGQIU WANG等: ""Unsupervised learning trajectory anomaly detection algorithm based on deep representation"", 《DISTRIBUTED SENSOR NETWORKS》, vol. 16, no. 12, pages 1 - 21 *
张俊辉等: ""基于主题提取模型的交通违法行为文本数据的挖掘"", 《人工智能》, vol. 45, no. 6, pages 41 - 45 *
毕孝儒等: ""无监督纹理遥感图像分形分割仿真"", 《计算机仿真》, vol. 37, no. 1, pages 22 - 26 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113487936A (en) * 2021-07-26 2021-10-08 广州星锐传媒信息科技有限公司 Simulation driving judgment method based on multi-dimensional continuous signal analysis
CN113505852A (en) * 2021-07-28 2021-10-15 云南电网有限责任公司电力科学研究院 Low-voltage user phase identification method based on principal component analysis and EM distance
CN115204323A (en) * 2022-09-16 2022-10-18 华智生物技术有限公司 Seed multi-feature based clustering and synthesis method, system, device and medium
CN116975731A (en) * 2023-08-08 2023-10-31 山东大学 Cross-domain cutter damage monitoring method and system based on transfer learning
CN116975731B (en) * 2023-08-08 2024-02-20 山东大学 Cross-domain cutter damage monitoring method and system based on transfer learning

Also Published As

Publication number Publication date
CN113159105B (en) 2023-08-08

Similar Documents

Publication Publication Date Title
CN113159105B (en) Driving behavior unsupervised mode identification method and data acquisition monitoring system
JP6026959B2 (en) Driving scene label estimation device
CN112749266B (en) Industrial question and answer method, device, system, equipment and storage medium
JP2018502386A (en) Mapper element for neuro-language behavior recognition system
CN113221975B (en) Working condition construction method based on improved Markov analysis method and storage medium
JP2015056058A (en) Driving support device
CN110727706A (en) Method for rapidly extracting and grading risk driving scene for intelligent networking automobile test
JP2018503183A (en) Vocabulary analyzer for neuro-language behavior recognition system
CN111242387A (en) Talent departure prediction method and device, electronic equipment and storage medium
JP2017010206A (en) Driving operation determination device
US9940759B2 (en) Drive data collection system
Mou et al. Structured behaviour prediction of on‐road vehicles via deep forest
CN112084339B (en) Traffic knowledge graph construction method based on cross-media data
CN112685374B (en) Log classification method and device and electronic equipment
JP2011123552A (en) Device for retrieval of similar image in driving, and program
CN116108276A (en) Information recommendation method and device based on artificial intelligence and related equipment
CN113159326B (en) Intelligent business decision method based on artificial intelligence
CN113010727B (en) Live platform portrait construction method, device, equipment and storage medium
CN111914138A (en) Working condition analysis method, device and system based on Internet of vehicles big data
CN116226686B (en) Table similarity analysis method, apparatus, device and storage medium
CN109918367A (en) A kind of cleaning method of structural data, device, electronic equipment and storage medium
CN115309713B (en) Traffic data compression method and device, electronic equipment and storage medium
CN116129640B (en) Data management method and device for road data
Han et al. A pattern accumulated compression method for trajectories constrained by urban road networks
Noering Unsupervised Pattern Discovery in Automotive Time Series

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