Vehicle violation track AI monitoring and identification system based on multi-source data fusion

Abstract

The invention discloses a vehicle violation track AI monitoring identification system based on multi-source data fusion, which relates to the technical field of vehicle supervision and comprises an equipment encryption module, a vehicle space track module, an equipment flow information module, a data cleaning module, a data fusion module, an algorithm model module, a violation vehicle module, an intelligent early warning module and the like. The method has the advantages that the recognition accuracy of illegal vehicles is greatly improved and the potential safety hazards in the running process of the illegal vehicles are completely eradicated on the basis of ensuring the cooperative work of the modules and the efficient running of the system through an innovative framework and a real-time recognition algorithm. According to the invention, through multi-module cooperation and innovative design, problems caused by influence of human factors and equipment cloning are solved, violation abnormal vehicles can be accurately identified in combination with multiple modes, and the problems are solved without depending on local means.

Description

Technical Field

[0001] The present invention relates to the field of vehicle supervision technology, and in particular to an AI monitoring and identification system for vehicle violation trajectories based on multi-source data fusion. Background Art

[0002] With the acceleration of urban construction and the booming development of high-rise buildings and infrastructure, the amount of waste generated has increased dramatically, leading to a significant increase in the number of construction site transport vehicles. Against this backdrop, violations by waste trucks have become increasingly prominent, with frequent incidents of theft, illegal dumping, overloading, speeding, and driving violations, often resulting in fatal accidents and posing a serious threat to urban safety and management. While existing monitoring platforms utilize "department-standardized vehicles" and monitoring platforms to implement monitoring measures such as positioning, speeding alarms, and offline alerts, these methods face numerous challenges in practical application.

[0003] Existing vehicle violation monitoring technologies have significant limitations. Most monitoring systems employ simple device encryption. Once this encryption is cracked, widespread identification of violating vehicles becomes impossible, severely impacting regulatory effectiveness. Furthermore, these systems often rely on a single data source, such as trajectory or traffic data, lacking multi-dimensional information support. This results in high misjudgment rates and makes it difficult to accurately identify violating vehicles.

[0004] Furthermore, existing technologies fail to effectively integrate multi-source data. When faced with complex vehicle operation scenarios, they are unable to comprehensively analyze multiple data sources, including device codes, trajectories, traffic volume, and camera radar. This makes it difficult to accurately identify vehicle violations in real time, and thus fails to meet increasingly stringent vehicle regulatory requirements. A more efficient and accurate vehicle violation monitoring system is urgently needed. Summary of the Invention

[0005] The present invention proposes an AI monitoring and identification system for vehicle violation trajectories based on multi-source data fusion to solve the problems mentioned in the above-mentioned prior art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an AI monitoring and identification system for vehicle violation trajectories based on multi-source data fusion, comprising: The device encryption module uniformly encodes the "ministry-standard machine" before leaving the factory, writes it into the device firmware, and adjusts the device encoding through a private protocol. The process is encrypted and the device encoding is dynamically adjusted through a private protocol at regular intervals. The encoding information is stored on the service end to obtain the unique code corresponding to each vehicle; The vehicle trajectory module is used to upload vehicle trajectory information according to the JT808 protocol to form a vehicle trajectory library, classify and store data according to different trajectory types, and facilitate subsequent algorithm calls; Camera radar module: The camera radar module accurately grasps the vehicle's operating data and, as a supplement to multi-source data, provides real-time information on the vehicle's driving and surrounding environment.

[0007] The traffic information module records and categorizes the traffic data generated by the traffic card, obtains the traffic usage data of the vehicle in each period, and stores it according to different traffic types; The data cleaning module processes and cleans the generated spatial trajectory data, camera radar data, and traffic data. It cleans multiple GPS trajectory data of the vehicle to obtain initial trajectory data, filters and thins out all trajectories, processes and extracts trajectory interruptions, straight lines, and constant speed data, and records the no-flow segments of traffic data. The data fusion module performs multi-data fusion on trajectory data, traffic data, and device encryption data. By obtaining a multi-source data set and based on the data from steps 2, 3, and 4, the level of the violating vehicle is defined as normal, suspect, suspected, and confirmed abnormal. Based on the multi-source data set, the vehicle violation is graded; The algorithm model module calculates the fused data by combining the spatial trajectory data, traffic card flow data, camera radar data, and device encryption data in the system. After algorithm filtering, it forms a vehicle violation warning data pool and records detailed data such as the cause of the violation. The illegal vehicle module is a warning data pool calculated based on the algorithm model, forming an actual illegal vehicle information database for personnel inspection and verification and manual judgment; The intelligent early warning module is a millisecond-level warning module for illegal vehicles, which identifies illegal vehicles and related behaviors based on historically accumulated violation data.

[0008] Furthermore, the device encryption module includes the following steps: The "ministry-standard machine" is uniformly coded before leaving the factory, written into the device firmware, and the device coding is adjusted through a private protocol. The process is encrypted, and the device coding is dynamically adjusted through a private protocol at regular intervals. The coding information is stored on the service end to obtain a unique code corresponding to each vehicle; Encryption: First, the secret string generated by the device is initially encrypted, the parity sort is eliminated, and the digest value obtained is encrypted using the private key generated by the asymmetric national secret algorithm SM2 for the processed secret string; Signature Verification: Use the same initial encryption algorithm, remove the odd / even order, and then use the public key generated from the asymmetric national encryption algorithm SM2 to decrypt the digital signature to obtain the original digest value. Compare the two digest values.

[0009] Furthermore, the vehicle trajectory module includes the following operating steps: Based on the vehicle trajectory information uploaded using the JT808 protocol, a vehicle spatial trajectory library is formed. Combined with the AI spatial trajectory algorithm, the data is classified and stored according to different trajectory types. The AI spatial trajectory algorithm is used to process the trajectory data after the device is online. First, it is processed using the extraction algorithm and combined with verification coding. Unable to respond or match is recorded as a violation trajectory. The data that the vehicle cannot respond to is defined as the set S, S = {S1, S2, S3...SN}, where S1...SN corresponds to the number of unanswered tracks.

[0010] Furthermore, the flow information module is specifically: The traffic information module records and categorizes the traffic data generated by the traffic card, obtains the traffic usage data of the vehicle in each period, and stores it according to different traffic types; Combined with the traffic data of the IoT card of the "Ministry Standard Machine", the traffic consumption of the vehicle uploading trajectory is identified. The time period with no traffic record during the abnormal trajectory is defined as the K set, K={K1, K2...KN}, where K1...KN corresponds to the traffic data period and N is the number of trajectory uploads.

[0011] Furthermore, the camera radar module is specifically: The camera radar module records and stores the alarm data generated by the camera radar in a classified manner, and obtains the alarm status data of the vehicle at each time period; Combined with trajectory data, the violation of the vehicle is identified. The trajectory record during the vehicle violation period is defined as L set, L = {L1, L2...LN}, where L1...LN corresponds to the traffic data period and N is the number of vehicle violations.

[0012] Furthermore, the data fusion module steps are: Step 1: Verify the code of the trajectory data after the device is online. Record the illegal trajectory if the vehicle cannot respond or cannot be matched. Define the data that the vehicle cannot respond to as a set S, S = {S1, S2, S3...SN}, where S1...SN corresponds to the number of unanswered vehicles. Step 2: Clean multiple GPS trajectory data of the vehicle to obtain initial trajectory data. Filter and thin out all trajectories, extract trajectory interruption, straight line, and constant speed data, and add this data to the anomaly algorithm. Define the number of single transport interruptions, straight line times, and constant speed as the U set, U = {U1, U2…UN}, where U1…UN correspond to trajectory interruption, straight line, and constant speed types, and N is the number of related type anomalies. Step 3: Combined with the traffic data of the IoT card of the "Ministry Standard Machine", the traffic consumption of the vehicle uploading trajectory is identified. The K set is defined as the set with no traffic records during the abnormal trajectory period, K={K1, K2...KN}, where K1...KN corresponds to the traffic data period and N is the number of trajectory uploads; Step 4: Perform multivariate data fusion on the S set, U set, and K set. By obtaining the multi-source data set X, based on the data from steps 2, 3, and 4, the levels of illegal vehicles are defined as normal, suspect, suspected, and confirmed abnormal. Based on the multi-source data set X = {S, U, K}, the vehicle violations are graded. Step 5: Based on the illegal vehicle data evidence, the weight of S is set to 0.41, the weight of U is set to 0.38, and the weight of K is set to 0.11, and the modified evidence matrix is constructed; Step 6: If the result of m(A) is greater than 0.8, the relevant vehicle will be included in the illegal vehicles.

[0013] Compared with the existing technology, the beneficial effects of the present invention are: The proposed AI-powered vehicle violation trajectory monitoring and identification system, based on multi-source data fusion, significantly improves the accuracy and reliability of vehicle violation identification through an innovative technical architecture and multi-dimensional monitoring approach. The system integrates multiple sources of data, including device encryption, vehicle trajectory, traffic information, and camera radar, overcoming the limitations of traditional single-source monitoring. Even if some identification points are compromised, the overall recognition effect remains unchanged, significantly reducing the risk of misjudgment.

[0014] The system boasts millisecond-level warning capabilities for vehicles violating regulations, enabling real-time and accurate identification of vehicles and their behavior, providing supervisors with timely warning information and enabling rapid action. The system also consolidates identified vehicle data into a database, providing comprehensive and accurate data support for inspection and verification, achieving a seamless integration of human judgment and intelligent identification.

[0015] Through collaborative analysis of multi-source data, the system not only accurately identifies vehicles violating traffic rules but also effectively eliminates issues caused by device cloning, ensuring effective regulation at multiple levels. This technical solution significantly improves the efficiency of vehicle violation monitoring, reduces safety hazards in urban traffic management, and provides strong technical support for vehicle regulation in urban development, with significant economic and social benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a schematic block diagram of the vehicle violation trajectory AI monitoring and identification system based on multi-source data fusion proposed by the present invention; Figure 2This is a bar chart comparing the recognition accuracy of the multi-source data fusion proposed by the present invention and the traditional method in different violation scenarios; Figure 3 A line chart comparing the early warning response times of different systems proposed in the present invention at various violation event time points; Figure 4 A pie chart showing the weight of multi-source data sets and other factors in violation ratings.

[0017] Figure 5 A bar chart comparing the number of device cloning events within a month under different encryption methods. DETAILED DESCRIPTION

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0019] The present invention will be described in further detail below with reference to the accompanying drawings.

[0020] Reference Figures 1 to 5 : An AI monitoring and recognition system for vehicle violation trajectories based on multi-source data fusion, including: The device encryption module uniformly encodes the "ministry-standard machine" before leaving the factory, writes it into the device firmware, and adjusts the device encoding through a private protocol. The process is encrypted and the device encoding is dynamically adjusted through a private protocol at regular intervals. The encoding information is stored on the service end to obtain the unique code corresponding to each vehicle; The vehicle trajectory module is used to upload vehicle trajectory information according to the JT808 protocol to form a vehicle trajectory library, classify and store data according to different trajectory types, and facilitate subsequent algorithm calls; Camera radar module: The camera radar module accurately grasps the vehicle's operating data and, as a supplement to multi-source data, provides real-time information on the vehicle's driving and surrounding environment.

[0021] The traffic information module records and categorizes the traffic data generated by the traffic card, obtains the traffic usage data of the vehicle in each period, and stores it according to different traffic types; The data cleaning module processes and cleans the generated spatial trajectory data, camera radar data, and traffic data. It cleans multiple GPS trajectory data of the vehicle to obtain initial trajectory data, filters and thins out all trajectories, processes and extracts trajectory interruptions, straight lines, and constant speed data, and records the no-flow segments of traffic data. The data fusion module performs multi-data fusion on trajectory data, traffic data, and device encryption data. By obtaining a multi-source data set and based on the data from steps 2, 3, and 4, the level of the violating vehicle is defined as normal, suspect, suspected, and confirmed abnormal. Based on the multi-source data set, the vehicle violation is graded; The algorithm model module calculates the fused data by combining the spatial trajectory data, traffic card flow data, camera radar data, and device encryption data in the system. After algorithm filtering, it forms a vehicle violation warning data pool and records detailed data such as the cause of the violation. The illegal vehicle module is a warning data pool calculated based on the algorithm model, forming an actual illegal vehicle information database for personnel inspection and verification and manual judgment; The intelligent early warning module is a millisecond-level warning module for illegal vehicles, which identifies illegal vehicles and related behaviors based on historically accumulated violation data.

[0022] In the present invention, the device encryption module includes the following steps: The "ministry-standard machine" is uniformly coded before leaving the factory, written into the device firmware, and the device coding is adjusted through a private protocol. The process is encrypted, and the device coding is dynamically adjusted through a private protocol at regular intervals. The coding information is stored on the service end to obtain a unique code corresponding to each vehicle; Encryption: First, the secret string generated by the device is initially encrypted, the parity sort is eliminated, and the digest value obtained is encrypted using the private key generated by the asymmetric national secret algorithm SM2 for the processed secret string; Signature Verification: Use the same initial encryption algorithm, remove the odd / even order, and then use the public key generated from the asymmetric national encryption algorithm SM2 to decrypt the digital signature to obtain the original digest value. Compare the two digest values.

[0023] In the present invention, the vehicle trajectory module includes the following operating steps: Based on the vehicle trajectory information uploaded using the JT808 protocol, a vehicle spatial trajectory library is formed. Combined with the AI spatial trajectory algorithm, the data is classified and stored according to different trajectory types. The AI spatial trajectory algorithm is used to process the trajectory data after the device is online. First, it is processed using the extraction algorithm and combined with verification coding. Unable to respond or match is recorded as a violation trajectory. The data that the vehicle cannot respond to is defined as the set S, S = {S1, S2, S3...SN}, where S1...SN corresponds to the number of unanswered tracks.

[0024] In the present invention, the flow information module is specifically: The traffic information module records and categorizes the traffic data generated by the traffic card, obtains the traffic usage data of the vehicle in each period, and stores it according to different traffic types; Combined with the traffic data of the IoT card of the "Ministry Standard Machine", the traffic consumption of the vehicle uploading trajectory is identified. The time period with no traffic record during the abnormal trajectory is defined as the K set, K={K1, K2...KN}, where K1...KN corresponds to the traffic data period and N is the number of trajectory uploads.

[0025] In the present invention, the camera radar module is specifically: The camera radar module records and stores the alarm data generated by the camera radar in a classified manner, and obtains the alarm status data of the vehicle at each time period; Combined with trajectory data, the violation of the vehicle is identified. The trajectory record during the vehicle violation period is defined as L set, L = {L1, L2...LN}, where L1...LN corresponds to the traffic data period and N is the number of vehicle violations.

[0026] In the present invention, the steps of the data fusion module are: Step 1: Verify the code of the trajectory data after the device is online. Record the illegal trajectory if the vehicle cannot respond or cannot be matched. Define the data that the vehicle cannot respond to as a set S, S = {S1, S2, S3...SN}, where S1...SN corresponds to the number of unanswered vehicles. Step 2: Clean multiple GPS trajectory data of the vehicle to obtain initial trajectory data. Filter and thin out all trajectories, extract trajectory interruption, straight line, and constant speed data, and add this data to the anomaly algorithm. Define the number of single transport interruptions, straight line times, and constant speed as the U set, U = {U1, U2…UN}, where U1…UN correspond to trajectory interruption, straight line, and constant speed types, and N is the number of related type anomalies. Step 3: Combined with the traffic data of the IoT card of the "Ministry Standard Machine", the traffic consumption of the vehicle uploading trajectory is identified. The K set is defined as the set with no traffic records during the abnormal trajectory period, K={K1, K2...KN}, where K1...KN corresponds to the traffic data period and N is the number of trajectory uploads; Step 4: Perform multivariate data fusion on the S set, U set, and K set. By obtaining the multi-source data set X, based on the data from steps 2, 3, and 4, the levels of illegal vehicles are defined as normal, suspect, suspected, and confirmed abnormal. Based on the multi-source data set X = {S, U, K}, the vehicle violations are graded. Step 5: Based on the illegal vehicle data evidence, the weight of S is set to 0.41, the weight of U is set to 0.38, and the weight of K is set to 0.11, and the modified evidence matrix is constructed; Step 6: If the result of m(A) is greater than 0.8, the relevant vehicle will be included in the illegal vehicles.

[0027] Taking the monitoring scenario of muck trucks in a certain city as an example, the city dispatches an average of 2,000 muck trucks per day, covering 50 construction sites. The system of the present invention is used to achieve real-time monitoring of vehicle violations, as follows: Device encryption module implementation details: 1. Encoding and dynamic adjustment: Each dump truck is assigned a factory code (e.g., "BM-20230518-001") and written into the device firmware via a proprietary protocol. The code is dynamically adjusted every 15 minutes using the SM2 asymmetric algorithm. The adjustment rule is: the original code is split based on the parity bit, reassembled, and then encrypted with the private key to generate a new code. The server stores a code mapping table. For example, vehicle A's current code is "DYN-20230601-123," which corresponds to the original code "BM-20230518-001."

[0028] 2. Encryption and signature verification process During initial encryption, the device-generated secret string (e.g., "abc123") is sorted by odd and even, removing the even-digit character "b2" (retaining the even-digit character "b2"). The ciphertext "0x34ab56cd" is then encrypted using the SM2 private key. During signature verification, the digital signature is decrypted using the SM2 public key and compared to the original digest value. If the decrypted digest "b2" matches the original digest, verification is successful.

[0029] Multi-source data collection and processing Trajectory data collection: Muck trucks upload their trajectories using the JT808 protocol, with a sampling frequency of 10Hz. For example, the coordinates of vehicle B at 10:00:00 are (116.481, 39.921), and the speed is 20km / h. A rarefaction algorithm is used to filter out redundant points, retaining key turning points. Abnormal data such as trajectory interruptions (e.g., no data from 10:05:00 to 10:06:00), straightening (trajectory deviation from the road centerline by more than 50m), and constant speed (consistent speed for 10 minutes) are classified into the U set. Traffic and radar data processing The traffic module records IoT card data usage. For example, if vehicle C consumes 5MB of data between 10:00 and 10:30, and there is no traffic record during the abnormal trajectory period (10:15-10:20), it is included in the K set. The camera radar module captures violations in real time. For example, if vehicle D drives in a prohibited area, the radar triggers an alarm and generates L set data.

[0030] Data fusion and violation classification: 1. Evidence Matrix Construction: Fusion of vehicle E’s monitoring data: S set: failed to respond 3 times (10:00, 10:05, 10:10); U set: the trajectory is interrupted once, the straight line is drawn twice, and the vehicle speed is constant once, U={1,2,1}; K set: There are two abnormal periods with no traffic records (10:05-10:06, 10:15-10:16). The weights are set to w_S=0.41, w_U=0.38, and w_K=0.11. Construct the evidence matrix: 2. Distance matrix calculation The distance between evidences is calculated using the formula: The matrix elements For example, if we calculate the distance between m(S) and m(U), we get d=0.25 and construct the DMM matrix: in, Represents: basic probability assignment functions (BPAs) modified based on evidence weights under the same identification framework; Represents: the modified basic probability assignment vector (converting BPAs into vector form); Indicates: one The matrix elements are: , where A and B are recognition frames A subset of represents the number of elements in the intersection of sets A and B, Indicates the number of elements in the union.

[0031] 3. Illegal classification The weighted average mass function formula is: ; represents the basic probability distribution of the fused evidence to the proposition A, and n represents the number of evidence sources (e.g. n=3 corresponds to the three sets S, U, and K). Indicates: the weight of the i-th evidence source (such as . Represents: the basic probability distribution of the i-th evidence source to proposition A, such as ; For example: the weighted average mass function is ; If m(A)>0.8 (such as the calculated m(A)=0.85), it is judged as "definitely abnormal".

[0032] Comparison of implementation effects: Table 1: Comparison of key vehicle violation monitoring indicators between traditional systems and this patented system Effectiveness Analysis: This patented system, through multi-source data fusion and AI algorithms, has increased violation identification accuracy by 41.5%, reduced false positives by 75%, and achieved millisecond-level early warning responses, effectively shortening resolution time. For example, traditional systems, relying on single-track data, often misidentify normal parking as violations. This system, combining traffic flow, radar data, and device encryption status, accurately identifies cloned devices and actual violations. For example, within a month, the number of false positives around a construction site was reduced by 12, with zero incidents of device cloning, significantly improving regulatory efficiency.

[0033] The above are only preferred specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solutions and inventive concepts of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. The AI monitoring and recognition system for vehicle violation trajectories based on multi-source data fusion is characterized by: include: The device encryption module writes the encryption code into the device, dynamically adjusts the device code at regular intervals, and stores the code on the service end for vehicle identification; The vehicle trajectory module uploads trajectories to form a trajectory library, which is classified and stored according to different types for subsequent algorithm calls; The camera radar module collects vehicle operating data, supplementing multi-source data to provide real-time information on the vehicle's movement and surrounding environment. Traffic information module, which stores traffic card traffic data and stores it according to traffic type; The data cleaning module filters and rarefies spatial trajectory, radar, and flow data, and records interruptions, straight lines, constant speeds, and sections without flow. The data fusion module encrypts and fuses trajectories, traffic flows, and devices, and classifies violations into normal, suspicious, suspected, and confirmed abnormal levels based on multi-source data sets. The algorithm model module calculates spatial trajectories, traffic card traffic, camera radar, and device encryption, and uses algorithm filtering to form a violation warning data pool; The illegal vehicle module forms an illegal information database based on the algorithm's early warning data pool for personnel inspection and verification and manual judgment; The intelligent warning module provides millisecond-level violation warnings and identifies violating vehicles and related behaviors based on historical violation data.

2. The vehicle violation trajectory AI monitoring and identification system based on multi-source data fusion according to claim 1 is characterized in that: The device encryption module includes the following steps: The "ministry-standard machine" is uniformly coded at the factory and written into the device firmware. The device coding is adjusted through a private protocol. The process is encrypted and the device coding is dynamically adjusted through a private protocol at regular intervals. The coding information is stored on the service end to obtain a unique code corresponding to each vehicle. Encryption: First, the secret string generated by the device is initially encrypted, the parity sort is eliminated, and the digest value obtained is encrypted using the private key generated by the asymmetric national secret algorithm SM2 for the processed secret string; Signature verification: Use the same initial encryption algorithm, eliminate the parity sort, and then use the public key generated from the asymmetric national secret algorithm SM2 to decrypt the digital signature to obtain the original digest value. Compare the two digest values.

3. The vehicle violation trajectory AI monitoring and identification system based on multi-source data fusion according to claim 1 is characterized in that: The vehicle trajectory module includes the following steps: Based on the vehicle trajectory information uploaded using the JT808 protocol, a vehicle spatial trajectory library is formed. Combined with the AI spatial trajectory algorithm, the data is classified and stored according to different trajectory types. The AI spatial trajectory algorithm is used to process the trajectory data after the device is online. First, it is processed using the extraction algorithm and combined with verification coding. Unable to respond or match is recorded as a violation trajectory. The data that the vehicle cannot respond to is defined as the set S, S = {S1, S2, S3...SN}, where S1...SN corresponds to the number of unanswered tracks.

4. The vehicle violation trajectory AI monitoring and identification system based on multi-source data fusion according to claim 2 is characterized in that: The flow information module is specifically: The traffic information module records and categorizes the traffic data generated by the traffic card, obtains the traffic usage data of the vehicle in each period, and stores it according to different traffic types; Combined with the traffic data of the IoT card of the "ministry standard machine", the traffic consumption of the vehicle uploading trajectory is identified. The K set is defined as the period of no traffic record during the abnormal trajectory period, K={K1, K2...KN}, where K1...KN corresponds to the traffic data period and N is the number of trajectory uploads.

5. The vehicle violation trajectory AI monitoring and identification system based on multi-source data fusion according to claim 2 is characterized in that: The camera radar module is specifically: The camera radar module records and stores the alarm data generated by the camera radar in a classified manner, and obtains the alarm status data of the vehicle at each time period; Combined with trajectory data, the violation of the vehicle is identified. The trajectory record during the vehicle violation period is defined as L set, L = {L1, L2...LN}, where L1...LN corresponds to the traffic data period and N is the number of vehicle violations.

6. The vehicle violation trajectory AI monitoring and identification system based on multi-source data fusion according to claim 1 is characterized in that: The steps of the data fusion module are: Step 1: Verify the code of the trajectory data after the device is online. Record the illegal trajectory if the vehicle cannot respond or cannot be matched. Define the data that the vehicle cannot respond to as a set S, S = {S1, S2, S3...SN}, where S1...SN corresponds to the number of unanswered vehicles. Step 2: Clean multiple GPS trajectory data of the vehicle to obtain initial trajectory data. Filter and thin out all trajectories, extract trajectory interruption, straight line, and constant speed data, and add this data to the anomaly algorithm. Define the number of single transport interruptions, straight line times, and constant speed as the U set, U = {U1, U2…UN}, where U1…UN correspond to trajectory interruption, straight line, and constant speed types, and N is the number of related type anomalies. Step 3: Combined with the traffic data of the IoT card of the "Ministry Standard Machine", the traffic consumption of the vehicle's uploaded trajectory is identified. The K set is defined as the set with no traffic records during the abnormal trajectory period, where K = {K1, K2...KN}, where K1...KN correspond to the traffic data period and N is the number of trajectory uploads; Step 4: Perform multivariate data fusion on the S set, U set, and K set. By obtaining the multi-source data set X, based on the data from steps 2, 3, and 4, the levels of illegal vehicles are defined as normal, suspect, suspected, and confirmed abnormal. Based on the multi-source data set X = {S, U, K}, the vehicle violations are graded. Step 5: Based on the illegal vehicle data evidence, the weight of S is set to 0.41, the weight of U is set to 0.38, and the weight of K is set to 0.11, and the modified evidence matrix is constructed; Step 6: If the result of m(A) is greater than 0.8, the relevant vehicle will be included in the illegal vehicles.

Images