A multi-source data fusion positioning method for driving test vehicles

Abstract

The present invention provides a multi-source data fusion positioning method applied to driving test vehicles, belonging to the field of intelligent transportation technology; the technical problem to be solved is: to provide an improvement of the multi-source data fusion positioning method applied to driving test vehicles; to obtain the position information of the test vehicle to be tested during the test and the original sensor data related to the vehicle jitter, and input the position information and the original sensor data into a fuzzy algorithm model for preprocessing; to build a fuzzy algorithm model and train the model: to fuse the preprocessed vehicle sensor information of various types, and to use the basic rules for vehicle jitter determination to identify and calculate the time, driving section and initial range of the vehicle jitter; to locate the vehicle positioning point on the corresponding digital map road, to automatically identify the distance between the driving test vehicle and the road edge line reference object, and to determine the violation of the driving test vehicle; the present invention is applied to the judgment of driving test vehicle violation.

Description

Technical Field

[0001] The present invention provides a multi-source data fusion positioning method applied to driving test vehicles, belonging to the field of intelligent transportation technology. Background Art

[0002] Driving test vehicle positioning and trajectory monitoring are prerequisites for driving test behavior analysis. During the driving test vehicle positioning and trajectory monitoring process, test vehicles, especially large vehicles, may shake and present an unstable driving posture due to factors such as their age, body conditions, and road conditions, which may lead to deviations in vehicle positioning and affect the accuracy of vehicle trajectory monitoring.

[0003] Data fusion is an integrated intelligent information processing technology that combines, processes and analyzes information collected by sensors. It can perform a higher-level fusion of sensor data that has undergone preliminary analysis and processing, thereby achieving a unified and standardized representation of multi-source data, making it easier to conduct a more comprehensive and in-depth comprehensive analysis of monitoring information.

[0004] Applying China's BeiDou Navigation Satellite System (BDS) to the positioning of driving test vehicles and combining it with map matching technology to achieve real-time trajectory characterization of monitored vehicles is a reliable method currently used in intelligent driving test systems. The BeiDou positioning system integrates the vehicle's sensors and vehicle attribute data for information fusion analysis, and combines various related data to correct vehicle positioning, thereby reducing positioning errors during vehicle monitoring and more accurately depicting the vehicle's driving trajectory. This has very important practical significance for determining the driving posture and violation judgment of driving test vehicles.

[0005] However, in the process of real-time monitoring of driving test vehicles and determination of violations, the Beidou system positioning technology may be affected by various factors to produce different degrees of vehicle positioning deviation. In order to solve the vehicle positioning error caused by various factors, reduce the omission of violation monitoring that may occur during vehicle shaking, and achieve more accurate trajectory characterization, posture analysis and violation determination during vehicle shaking, the present invention provides a multi-source data fusion positioning method applied to driving test vehicles. The centimeter-level positioning of driving test vehicles is achieved by adopting Beidou technology combined with Kalman filtering and real-time differential positioning technology RTK (Real-time kinematic), and comprehensively considers various factors affecting vehicle shaking, analyzes the posture and details of the driving test vehicle shaking process, and increases the comprehensiveness and accuracy of vehicle driving behavior analysis. Summary of the invention

[0006] In order to overcome the deficiencies in the prior art, the present invention aims to solve the technical problem of providing an improved multi-source data fusion positioning method for driving test vehicles.

[0007] In order to solve the above technical problems, the technical solution adopted by the present invention is: a multi-source data fusion positioning method applied to driving test vehicles, comprising the following steps:

[0008] S1: Obtain the position information of the vehicle to be tested during the test and the original sensor data related to the vehicle vibration, and input the position information and the original sensor data into the fuzzy algorithm model for preprocessing;

[0009] S2: construct a fuzzy algorithm model and train the model: the input of the fuzzy algorithm model is data related to vehicle shaking, and the output is corrected vehicle positioning information;

[0010] The training of the model includes: simulating the shaking degree of different vehicles under different conditions, collecting sensor data on the vehicle, performing the same-direction level difference analysis on a certain frame of collected data, determining the difference range of the data under different conditions and normal conditions, obtaining the corresponding relationship between the shaking degree of the vehicle and the vehicle condition, and inputting the corresponding relationship into the model in the form of a multi-dimensional array as the basic rule for judging the vehicle positioning and shaking;

[0011] S3: The information collection node on the vehicle fuses the pre-processed information from various sensors of the vehicle, and identifies and calculates the time, driving section, and initial range of vehicle jitter based on the acquired vehicle data and basic vehicle jitter determination rules;

[0012] S4: Use a weight-based map matching algorithm to locate the vehicle location point on the corresponding digital map road, place the coordinates of the vehicle body marking points on the GIS map for search and matching, determine whether the monitored vehicle is in the project area, automatically identify the distance between the driving test vehicle and the road edge line reference object, and determine the violation of the driving test vehicle.

[0013] In step S1, the vehicle position information and sensor raw data are obtained through a positioning system, and the positioning system includes a fixed Beidou base station installed in a driving test site, a Beidou mobile base station installed on the vehicle, and relevant sensors on the vehicle. The vehicle data collected by the relevant sensors on the vehicle include vehicle engine status, vehicle speed, vehicle acceleration, vehicle steering angular velocity, and vehicle distance reference object data.

[0014] In the step 1, the data is preprocessed by the fuzzy algorithm model, including positioning data processing, vehicle feature data processing and cluster analysis processing.

[0015] The positioning data processing steps are as follows:

[0016] The pseudo-distance method is used to correct the data information, and the calculation formula is:

[0017] σ=t·C=(t b -t a )·C;

[0018] In the above formula: σ is the pseudo distance of signal transmission; t a The time when the signal satellite sends the signal; t b is the time it takes for a ground GPS receiver to receive satellite signals; C is the speed of light;

[0019] The propagation time is then corrected according to the error between the satellite time and the ground time:

[0020] t a +V a =T a ;

[0021] t b +V b =T b ;

[0022] In the above formula: V a is the satellite clock error; V b is the GPS receiver clock error; T a The time when the corrected signal is sent by the satellite; T b It is the time when the ground GPS receiver receives the satellite signal after correction;

[0023] According to the above formula, the corrected signal propagation time is obtained, and then substituted into the correction formula to obtain:

[0024] σ=(t b -t a )·C+(V a -V b )·C;

[0025] Where: t b -t a is the actual signal propagation time after time error correction; V a -V b is the clock error after time error correction;

[0026] Perform the impurity removal correction operation again:

[0027] σ=(t b -t a )·C+(V a -V b )·C+I(t)N(t);

[0028] In the above formula: I(t) is the delay time caused by ionization refraction when the signal propagates in the atmosphere, N(t) is the delay time caused by convection refraction when the signal propagates in the atmosphere;

[0029] According to the above formula, the actual signal propagation distance is obtained by pseudo-distance method, and the spatial coordinates of the observation point (x′, y′, z′) are substituted to obtain accurate positioning data and correct the distance error:

[0030]

[0031] The vehicle feature data processing steps are as follows:

[0032] Assume V = {v1, v2, v3, ..., v i , …, v m} is the set of vehicles, m is the number of vehicles, v i is the measured distance data of any vehicle, v i ={p1, p2, p3, ..., p i ,…,p s}, where s is the number of data collection points;

[0033] Vehicle data at different collection points will contain a variety of feature data, p i It is a tuple of all features of a vehicle at a collection point, defined as p i =(t, lon, lat, speed, rot, ac, σ, eng, shake), where t, lon, lat, speed, rot, ac, σ, eng, shake represent time, longitude, latitude, vehicle speed, vehicle steering angular velocity, vehicle acceleration, signal propagation distance data, vehicle engine status, and vehicle shake data, respectively. The calculation formulas for vehicle steering angular velocity and vehicle acceleration are:

[0034]

[0035]

[0036]

[0037] In the above formula, lon p ,lat p , t p ,speed p ,rot p , ac p They represent the longitude, latitude, timestamp, speed, steering angular velocity and acceleration of the vehicle respectively.

[0038] The steps of cluster analysis processing are as follows:

[0039] Before clustering analysis, each vehicle feature data is normalized, and the normalized vehicle position features, vehicle speed features, vehicle direction features and vehicle basic information features are fused into a feature vector ε. Then, the feature vector ε is clustered and clustered using the Gaussian mixture model algorithm. Based on the clustering results, the vehicle jitter results are given.

[0040] In step S3, the pre-processed vehicle sensor information is fused, wherein the vehicle sensor information is collected by a positioning AT330 measurement antenna, a directional AT330 measurement antenna, a control switch sensor and a camera device arranged on the vehicle, and the collected information includes vehicle switch control information, vehicle engine status information, vehicle real-time positioning and directional information and vehicle driving information. The steps of data fusion are:

[0041] First, the spatiotemporal benchmarks of multi-source data are unified, data redundancy items are removed, and spatiotemporal labels are added to the multi-source data, which are then converted into final vehicle monitoring data in a unified format. Finally, the fused data is transmitted to the control and monitoring system terminal.

[0042] The steps of the weight-based map matching algorithm in step S4 are as follows:

[0043] The positioning point (t, lon, lat, speed, rot, ac, σ, eng, shake) is taken as the position of the vehicle on the current road, and different weights are assigned to different influencing factors. During the vehicle driving process, the vehicle steering angular velocity is an important factor in controlling the direction of the vehicle, and its weight is higher than other influencing factors. The vehicle speed and vehicle acceleration have the same weight.

[0044] The beneficial effects of the present invention compared to the prior art are as follows: the present invention mainly targets the real-time monitoring and violation assessment scenarios of driving test vehicles, and realizes centimeter-level positioning of mobile vehicles through BDS technology. The present invention focuses on the situation where the driving test vehicle shakes during driving, and explores the impact of vehicle shaking on BDS vehicle positioning through the relationship between vehicle-related attributes and vehicle shaking. The stage where the vehicle shakes is located by data clustering and grouping, so that a more comprehensive and detailed description and characterization of the vehicle driving process can be achieved.

[0045] The present invention uses data fusion technology to realize the centralized and unified processing of vehicle multi-sensor collected data, and obtains the spatiotemporal unified data of vehicle driving test process data, including vehicle switch control information, vehicle engine status information, vehicle real-time positioning and orientation information, and other vehicle driving information. The centralized and unified processing of vehicle collected information ensures the uniformity of sensor information, and also reduces the difficulty of data processing of the terminal control system.

[0046] The present invention combines the vehicle's own attribute data, engine status data, road condition information, etc. to perform positioning correction and trajectory range definition on the vehicle's positioning trajectory, explores the impact of different degrees of vehicle shaking on the illegal crossing of the vehicle body, and implements higher frequency detection and analysis within the time range of vehicle shaking, thereby enhancing the accuracy of vehicle violation judgment, which has practical guiding significance for the further promotion of intelligent evaluation of driving tests. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The present invention will be further described below in conjunction with the accompanying drawings:

[0048] Figure 1 It is a schematic diagram of the system structure of the present invention;

[0049] Figure 2 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0050] like Figure 1 to Figure 2 As shown, in order to solve the positioning error caused by various factors in the process of monitoring the driving test vehicle, the present invention provides a multi-source data fusion positioning method applied to the driving test vehicle. The method is used to fuse multi-source data to make a more reasonable correction to the positioning error caused by the vehicle shaking, which has very important guiding significance for reducing the omission of possible violations during the vehicle shaking process and achieving more accurate trajectory characterization, posture analysis and violation judgment during the vehicle shaking process. The centimeter-level positioning of the driving test vehicle is achieved by adopting Beidou technology combined with Kalman filtering and real-time differential positioning technology, and considering the various factors affecting the vehicle shaking, the posture and details of the driving test vehicle shaking process are analyzed. The specific technical scheme is as follows:

[0051] Step 1: Set up a Beidou fixed base station at the driving test site, and set up a Beidou mobile base station and related sensors on the test vehicle and the test vehicle. The Beidou fixed base station is used as a reference point and can be adjusted according to actual needs; the Beidou mobile base station is set on the vehicle to be tested, and the real-time positioning of the monitored vehicle is achieved through the RTK-based navigation and positioning system, such as Figure 1 As shown in Figure 2, the vehicle sensor data is collected through a positioning AT330 measurement antenna, a directional AT330 measurement antenna, a control switch sensor, and a camera device.

[0052] Step 2: Survey the entire test site and draw a GIS electronic map of the site. The main information on the electronic map is the area coordinates and area number of each item in Subject 2 and Subject 3.

[0053] Step 3: Conduct a simulation experiment to simulate different degrees of vehicle shaking and collect its sensor data. Specifically, first collect vehicle data when the vehicle is stationary, then collect the vehicle speed when shaking occurs and the speed when the shaking is maximum at the same position, classify according to the degree of shaking, and complete the calibration of the vehicle shaking.

[0054] Perform the same direction range analysis on the collected frame data, that is, first calculate the mean value of the direction data Then calculate The maximum value of the distance difference between data points within the direction range, where μ is an adjustable custom parameter, and the result is compared with the extreme value of the data collected when the vehicle is driving smoothly to determine the difference range between the data in different situations and normal situations. At the same time, define and identify vehicle-related attributes, correspond different degrees of vehicle jitter to the vehicle engine status, vehicle speed, vehicle acceleration, vehicle steering angular velocity and other attribute collection data related to vehicle jitter, obtain their corresponding relationship, and input the results into the system in the form of a multidimensional array as the basic rules for subsequent system analysis of vehicle positioning and jitter behavior.

[0055] When judging whether a vehicle has crossed the line, you can divide the vehicle body into the front and rear parts, and determine whether the vehicle has violated the rules based on the position of the vehicle body that has crossed the line and its shaking.

[0056] Based on the simulation experiment, the basic framework of the fuzzy algorithm model was obtained. The data of vehicle-related attributes was taken as input, and the judged vehicle jitter results were taken as output. In the fuzzy algorithm model, RTK data processing, feature processing, and cluster analysis were used to process the input data.

[0057] Step 4: After obtaining the raw data from the sensor, use the RTK-based navigation and positioning technology to obtain its positioning data, and then process the data. Cluster the vehicle positioning data and group the data according to the direction data information in the data message. This includes the following steps:

[0058] 1. RTK data processing: The signals collected by the data acquisition equipment have certain errors, and are easily interfered by many factors such as time difference and atmospheric refraction during signal transmission. The pseudo-range method is used to correct the data information, and the time difference or other information errors in the receiving process are calculated according to the distance between the receiver and the signal satellite, so that the final positioning measurement results are more accurate. The operation process of the pseudo-range method:

[0059] σ=t·C=(t b -t a )·C (1);

[0060] Where σ is the pseudo distance of signal transmission; t a The time when the signal satellite sends the signal; tb It is the time required for the ground GPS receiver to receive the satellite signal. The pseudo-range is equal to the time difference of signal propagation multiplied by the speed of light C.

[0061] The propagation time is then corrected according to the error between the satellite time and the ground time:

[0062] t a +V a =T a (2);

[0063] t b +V b =T b (3);

[0064] According to the above formula, the corrected signal propagation time can be obtained, and then substituted into formula (1) to obtain:

[0065] σ=(t b -t a )·C+(V a -V b )·C (4);

[0066] Where: V a -V b is the clock error after time error correction, t b -t a is the actual signal propagation time after time error correction. However, the measured signal will still be interfered by factors such as atmospheric ions during this process, so it is necessary to perform the impurity removal and correction operation again:

[0067] σ=(t b -t a )·C+(V a -V b )·C+I(t)N(t) (5);

[0068] Where I(t) and N(t) are the delay times caused by ionization refraction and convection refraction when the signal propagates in the atmosphere:

[0069] According to the above formula, the actual signal propagation distance is obtained by pseudo-distance method, and the spatial coordinates of the observation point (x′, y′, z′) are substituted to obtain accurate positioning data and correct the distance error:

[0070]

[0071] Through the above calculation process, positioning information and distance measurement data with small error and high accuracy are obtained.

[0072] 2. Feature processing, assuming V = {v1, v2, v3, ..., v i , …, vm} is the set of vehicles, m is the number of vehicles, v i is the measured distance data of any vehicle. i ={p1, p2, p3, ..., p i ,…,p s}, where s is the number of data collection points. The vehicle data at different collection points will contain a variety of feature data, p i It is a tuple of all features of a vehicle at a collection point, including time, longitude, latitude, vehicle speed, vehicle steering angular velocity, vehicle acceleration, distance data, vehicle engine status, and vehicle vibration data. It is defined as p i =(t, lon, lat, speed, rot, ac, σ, eng, shake), but the original vehicle data lacks the vehicle steering angular velocity and vehicle acceleration, and the features need to be calculated to further analyze the vehicle shake. Therefore, formula (7)-formula (9) are used to calculate the angular velocity and acceleration.

[0073]

[0074]

[0075]

[0076] Among them, lon p ,lat p , t p ,speed p ,rot p , ac p They respectively represent the longitude, latitude, timestamp, speed, steering angular velocity and acceleration of the vehicle, and the timestamp information involved needs to be converted into numerical values ​​for calculation.

[0077] Since the collected data values ​​vary greatly, data normalization is required to eliminate the dimensionality impact between data indicators. Min-max normalization is used so that the results can fall within the [0,1] interval.

[0078]

[0079] Among them, max and min represent the maximum and minimum values ​​in a set of vehicle data, respectively.

[0080] 3. Cluster analysis: The data after normalization is processed by the autoencoder to extract the vehicle position features, vehicle speed features, vehicle direction features and vehicle basic information features, and the extracted features are fused into a feature vector ε, and then the feature vector ε is clustered. The Gaussian mixture model (GMM) algorithm is used for clustering. The Gaussian mixture model refers to a linear combination of multiple Gaussian distribution functions, with N categories, each category corresponds to a data point, and any given data can be perfectly fitted. The vehicle jitter result is given through the clustering result.

[0081] Step 5: Set up information collection nodes to process various types of vehicle sensor information that have been pre-processed, so that various sensor data are concentrated on a data main board, which is convenient for unified reception and processing with the control system. Including vehicle switch control information, vehicle engine status information, vehicle real-time positioning and orientation information, and other vehicle driving information. To fuse the above data, first unify the spatiotemporal benchmarks of multi-source data, remove data redundancy, add spatiotemporal labels to multi-source data, and standardize and convert them into final vehicle monitoring data with a unified format. Finally, transmit the fused data to the control and monitoring system terminal.

[0082] Step 6: Vehicle jitter identification and scope definition: Based on the acquired vehicle data, the basic rules for determining vehicle jitter obtained from the vehicle jitter simulation experiment conducted in advance are used to identify and calculate the time when the vehicle jitter occurs, the driving section, the approximate range of the jitter, etc.

[0083] Step 7: Map matching technology is to correct the position information received by the vehicle positioning information system, eliminate the influence of various errors, so that the vehicle positioning point is closer to the real map positioning point, and locate this positioning point on the corresponding digital map road, providing necessary support for vehicle monitoring and analysis.

[0084] The specific algorithm used is a weight-based map matching algorithm, which takes the positioning point (t, lon, lat, speed, rot, ac, σ, eng, sh ake) as the position of the vehicle on the current road, and comprehensively considers factors such as the vehicle speed and vehicle steering angular velocity during vehicle operation. Since different influencing factors have different effects on the matching map, different weights are assigned to them. During vehicle driving, the vehicle steering angular velocity (rot) is an important factor in controlling the direction of the vehicle and should have a greater influence than other weights. The weight of the vehicle speed (speed) is similar to the weight of the vehicle acceleration (ac), and they are assigned the same weight.

[0085] W=W rot +W spe +W ac (11);

[0086] Among them, Wrot , W spe , W ac are the weights of the vehicle steering angular velocity, vehicle speed and vehicle acceleration respectively. The total weight W determines the map section that the positioning data matches.

[0087] Step 8: Vehicle violation determination and driving behavior analysis. By placing the coordinates of the vehicle body marking points in the GIS map for search and matching, it is determined whether the monitored vehicle is in the project area. At the same time, it can also automatically identify the distance between the vehicle and reference objects such as the road edge line, thereby determining violations such as vehicle wheels and body crossing the line or the vehicle driving route error. The control and monitoring system can observe and analyze the monitored vehicle in real time, and can intuitively display and record the vehicle's driving dynamics, providing data basis for the driving behavior analysis of driving test candidates.

[0088] The present invention uses the Beidou positioning system independently developed by my country to achieve accurate positioning of driving test vehicles, and uses electronic map matching technology to achieve vehicle driving trajectory characterization, realizing more intuitive and intelligent monitoring of the entire driving test vehicle process. The intrinsic connection between factors such as vehicle age, vehicle body conditions, and driving road conditions and vehicle positioning is explored, and the vehicle positioning is analyzed in conjunction with the vehicle jitter process, further enhancing the accuracy of the monitored vehicle's line-crossing detection during the driving test. The information acquisition node of the present invention can process different sensor signals, including vehicle switch control information, vehicle engine status information, vehicle real-time positioning and orientation information, and other vehicle driving information. The centralized and unified processing of vehicle information ensures the uniformity of sensor information, while also reducing the difficulty of data processing in the terminal control system.

[0089] Regarding the specific structure of the present invention, it should be noted that the connection relationship between the various component modules adopted in the present invention is definite and feasible. Except for special instructions in the embodiments, the specific connection relationship can bring about corresponding technical effects, and solve the technical problems raised by the present invention without relying on the execution of corresponding software programs. The components, modules, models of specific components, the connection methods between each other, and the conventional use methods and expected technical effects brought about by the above-mentioned technical features, except for specific instructions, all belong to the disclosed contents in patents, journal articles, technical manuals, technical dictionaries, and textbooks that can be obtained by technical personnel in this field before the application date, or belong to the existing technologies such as conventional technologies and common knowledge in this field, and there is no need to elaborate, so that the technical solution provided in this case is clear, complete, and feasible, and the corresponding physical products can be reproduced or obtained according to the technical means.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-source data fusion positioning method for driving test vehicles, characterized by: The steps include: S1: Obtain the position information of the vehicle to be tested during the test and the original sensor data related to the vehicle vibration, and input the position information and the original sensor data into the fuzzy algorithm model for preprocessing; In step S1, the vehicle position information and the sensor raw data are obtained through a positioning system, wherein the positioning system includes a fixed Beidou base station arranged at a driving test site, a Beidou mobile base station arranged on a vehicle, and relevant sensors on the vehicle, wherein the vehicle data collected by the relevant sensors on the vehicle include vehicle engine status, vehicle speed, vehicle acceleration, vehicle steering angular velocity, and data of vehicle distance to a reference object; In the step 1, the data is preprocessed by the fuzzy algorithm model, including positioning data processing, vehicle feature data processing and cluster analysis processing; The positioning data processing steps are as follows: The pseudo-distance method is used to correct the data information, and the calculation formula is: σ=t·C=(t b -t a )·C; In the above formula: σ is the pseudo distance of signal transmission; t a The time when the signal is sent by the signal satellite; t b is the time it takes for a ground GPS receiver to receive satellite signals; C is the speed of light; The propagation time is then corrected according to the error between the satellite time and the ground time: t a +V a =T a ; t b +V b =T b ; In the above formula: V a is the satellite clock error; V b is the GPS receiver clock error; T a The time when the satellite sends the corrected signal; T b It is the time when the ground GPS receiver receives the satellite signal after correction; According to the above formula, the corrected signal propagation time is obtained, and then substituted into the correction formula to obtain: σ=(t b -t a )·C+(V a -V b )·C; Where: t b -t a is the actual signal propagation time after time error correction; V a -V b is the clock error after time error correction; perform the impurity removal and correction operation again: σ=(t b -t a )·C+(V a -V b )·C+I(t)N(t); In the above formula: I(t) is the delay time caused by ionization refraction when the signal propagates in the atmosphere, N(t) is the delay time caused by convection refraction when the signal propagates in the atmosphere; According to the above formula, the actual signal propagation distance is obtained by pseudo-distance method, and the spatial coordinates of the observation point (x′, y′, z′) are substituted to obtain accurate positioning data and correct the distance error: S2: construct a fuzzy algorithm model and train the model: the input of the fuzzy algorithm model is data related to vehicle shaking, and the output is corrected vehicle positioning information; The training of the model includes: simulating the shaking degree of different vehicles under different conditions, collecting sensor data on the vehicle, performing the same-direction level difference analysis on a certain frame of collected data, determining the difference range of the data under different conditions and normal conditions, obtaining the corresponding relationship between the shaking degree of the vehicle and the vehicle condition, and inputting the corresponding relationship into the model in the form of a multi-dimensional array as the basic rule for judging the vehicle positioning and shaking; S3: The information collection node on the vehicle fuses the pre-processed information from various sensors of the vehicle, and identifies and calculates the time, driving section, and initial range of vehicle jitter based on the acquired vehicle data and basic vehicle jitter determination rules; S4: Use a weight-based map matching algorithm to locate the vehicle location point on the corresponding digital map road, place the coordinates of the vehicle body marking points on the GIS map for search and matching, determine whether the monitored vehicle is in the project area, automatically identify the distance between the driving test vehicle and the road edge line reference object, and determine the violation of the driving test vehicle.

2. The multi-source data fusion positioning method for driving test vehicles according to claim 1 is characterized by: The vehicle feature data processing steps are as follows: Assume V = {v1, v2, v3, ..., v i , …, v m } is the set of vehicles, m is the number of vehicles, v i is the measured distance data of any vehicle, V i ={p1, p2, p3, ..., p i ,…,p s }, where s is the number of data collection points; Vehicle data at different collection points will contain a variety of feature data, p i It is a tuple of all features of a vehicle at a collection point, defined as p i =(t, lon, lat, speed, rot, ac, σ', eng, shake), where t, lon, lat, speed, rot, ac, σ', eng, shake represent time, longitude, latitude, vehicle speed, vehicle steering angular velocity, vehicle acceleration, corrected signal transmission distance, vehicle engine status, and vehicle shake data, respectively. The calculation formulas for vehicle steering angular velocity and vehicle acceleration are: In the above formula, lon p ,lat p , t p ,speed p ,rot p , ac p They represent the longitude, latitude, timestamp, speed, steering angular velocity and acceleration of the vehicle respectively.

3. The multi-source data fusion positioning method for driving test vehicles according to claim 2 is characterized by: The steps of cluster analysis processing are as follows: Before clustering analysis, each vehicle feature data is normalized, and the normalized vehicle position features, vehicle speed features, vehicle direction features and vehicle basic information features are fused into a feature vector ε. Then, the feature vector ε is clustered and clustered using the Gaussian mixture model algorithm. Based on the clustering results, the vehicle jitter results are given.

4. The multi-source data fusion positioning method for driving test vehicles according to claim 1 is characterized by: In step S3, the pre-processed vehicle sensor information is fused, wherein the vehicle sensor information is collected by a positioning AT330 measurement antenna, a directional AT330 measurement antenna, a control switch sensor and a camera device arranged on the vehicle, and the collected information includes vehicle switch control information, vehicle engine status information, vehicle real-time positioning and directional information and vehicle driving information. The steps of data fusion are: First, unify the spatiotemporal benchmarks of multi-source data, remove data redundancy, add spatiotemporal labels to multi-source data, and standardize and transform them into final vehicle monitoring data with a unified format. Finally, transmit the fused data to the control and monitoring system terminal.

5. The multi-source data fusion positioning method for driving test vehicles according to claim 2 is characterized in that: The weight-based map matching algorithm steps in step S4 are as follows: Based on the positioning point (t, lon, lat, speed, rot, ac ， σ', eng, shake) is used as the position of the vehicle on the current road, and different weights are assigned to different influencing factors. During vehicle driving, the vehicle steering angular velocity, as an important factor in controlling the direction of the vehicle, has a higher weight than other influencing factors, and the vehicle speed has the same weight as the vehicle acceleration.

Images