Abnormal charge detection method based on GNSS track credibility

Abstract

An abnormal charging detection method based on GNSS track credibility comprises the following steps: acquiring GNSS positioning data in a vehicle driving process through a vehicle-mounted unit, uploading the GNSS positioning data to a charging platform server in real time in a track form, acquiring vehicle identity information, a timestamp and vehicle motion information through a road side unit when a vehicle passes, and uploading the vehicle identity information, the timestamp and the vehicle motion information to the charging platform server; after the data center of the charging platform server preprocesses the received trajectory data, a candidate path set through which the vehicle may pass is generated through a map matching algorithm, multi-source information fusion consistency check is performed on the GNSS trajectory of the vehicle-mounted unit, the candidate paths of the electronic map and the vehicle passing records of the road side unit, and the reliability of the trajectory is calculated; and abnormal charging detection is carried out by screening the credibility of the track. According to the invention, high-credibility, all-weather and full-scene free flow metering and charging service in a complex traffic environment is realized.

Description

Technical Field

[0001] This invention relates to a technology in the field of satellite navigation applications, specifically an abnormal charging detection method based on GNSS trajectory reliability. Background Technology

[0002] Existing free-flow tolling systems primarily rely on vehicle-mounted GNSS positioning for mileage-based billing, with some schemes combining roadside unit records for statistical analysis. However, these methods only use electronic maps for trajectory correction and lack a consistency evaluation mechanism based on prior-observation-posterior, leading to insufficient reliability of billing results in environments with signal interference or multipath interference, and making them prone to mischarge or missed charges. Summary of the Invention

[0003] This invention addresses the shortcomings of existing technologies, which lack a systematic multi-source information consistency verification mechanism, are susceptible to multipath effects, interference, and pseudo-satellite deception, and cannot ensure the reliability of measurement results in all-weather, complex traffic environments. It proposes an anomaly toll detection method based on GNSS trajectory reliability. By fusing GNSS positioning results from vehicle-mounted units, roadside unit identification records, and electronic map matching results from the toll platform server's data center, the reliability of toll measurement is improved. Simultaneously, in the event of anomalies, an audit process is triggered to ensure the fairness and traceability of the free-flow tolling process. Thus, highly reliable, all-weather, and all-scenario free-flow metering and tolling services are achieved in complex traffic environments.

[0004] This invention is achieved through the following technical solution:

[0005] This invention proposes an abnormal toll detection method based on GNSS trajectory credibility. The method involves: ① collecting GNSS positioning data via an onboard unit during vehicle operation and uploading it in real-time to the toll platform server in trajectory form; ② acquiring vehicle identity information, timestamps, and vehicle movement information via a roadside unit as the vehicle passes by and uploading this information to the toll platform server. The toll platform server's data center then preprocesses the received trajectory data and generates a set of candidate paths the vehicle might take using a map matching algorithm. The method performs a multi-source information fusion consistency check on the onboard unit's GNSS trajectory, electronic map candidate paths, and roadside unit vehicle passage records, calculates trajectory credibility, and then detects abnormal tolls by filtering the trajectory credibility.

[0006] The reliability of the trajectory ,in: Reliability of satellite navigation observations , Positioning error covariance matrix, This is the observation bias vector; To assess the confidence level of roadside unit (RSU) observations, if a vehicle is observed within the RSU detection area, then... ,otherwise . .

[0007] The aforementioned abnormal charging detection method specifically includes:

[0008] 1) The on-board unit collects vehicle GNSS positioning information at a frequency of 1Hz and calculates the covariance. The data is reported to the toll platform in real time. At the same time, roadside units are deployed at key sections of the free-flow channel to collect vehicle IDs, timestamps and movement information to form vehicle passage records.

[0009] 2) After receiving the above two types of data, the charging platform performs time synchronization and outlier removal, and generates a candidate path set through map matching algorithms. ;

[0010] 3) Calculate the observation probability for each candidate path using a transition model. With transition probability And multiply them sequentially: .

[0011] 4) Take the highest confidence level The corresponding path is used as the actual driving path of the vehicle; when At that time, the system triggered an abnormal audit.

[0012] Technical effect

[0013] This invention proposes a trajectory reliability evaluation model based on three-source data fusion in a free-flow tolling system. This model simultaneously incorporates vehicle-mounted GNSS trajectory data, roadside RSU vehicle passage records, and candidate path information from the platform's electronic road network map. The model defines observation probabilities. With transition probability Multiply the two together to obtain the overall confidence level. The average value μ and standard deviation σ of the dynamic threshold are used for abnormal toll collection detection. Compared with the prior art, this invention can realize a unified quantitative credibility assessment of vehicle trajectory in free-flow channel environment, thereby improving the toll collection system's billing accuracy and risk identification capability in complex signal environments (such as obstruction, multipath, signal weakening). At the same time, it can determine the credibility / uncredibility of the trajectory in real time during vehicle passage and automatically trigger the manual audit process for low credibility trajectories, thereby significantly reducing mischarge, disputes and operational risks. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the invention;

[0015] Figure 2 This is a flowchart of the present invention;

[0016] Figure 3 This is a schematic diagram of the system in the embodiment;

[0017] Figure 4 This is a data range diagram for an example;

[0018] Figure 5 This is a probability diagram illustrating the credibility of the embodiments.

[0019] Figure 6 Position error diagram for an example;

[0020] Figure 7 The speed error result is shown in the example diagram. Detailed Implementation

[0021] like Figure 1 and Figure 3 As shown in this embodiment, an abnormal charging detection system based on GNSS trajectory reliability includes: an on-board unit, a roadside unit, a data center of the charging platform server, a data fusion system, a billing generation unit, a charging platform, and a manual auditing platform. Specifically: the on-board unit collects GNSS positioning data during vehicle operation in real time and uploads the trajectory information to the data center of the charging platform server via a communication module; the roadside unit is deployed at key road nodes susceptible to GNSS interference to collect vehicle passage information. The data center of the charging platform server receives the data uploaded by the on-board unit and the roadside unit, and performs trajectory cleaning and candidate path generation in conjunction with electronic map information, providing a basis for subsequent consistency verification; the data fusion system performs fusion processing and reliability calculation on multi-source information. The billing generation unit completes rate approval and bill calculation based on the reliable trajectory results, generating the corresponding charging bill; the charging platform distributes bills to users and provides complaint and traceability application channels. When users disagree with the billing results, they can submit an appeal request through this platform; the manual auditing platform processes trajectory data marked as abnormal by the data fusion system or appealed by users. The platform supports map information fusion, trajectory display and replay, which facilitates manual verification and avoids misreceipt or omission caused by GNSS errors or malicious interference.

[0022] The vehicle-mounted unit includes: a positioning and acquisition unit with a GNSS receiver, a data fusion unit, a trajectory reliability calculation unit, and a data communication and anomaly triggering unit. The positioning and acquisition unit collects positioning data based on BeiDou or GPS signals, calculates the positioning error covariance matrix Σ_GNSS, and outputs a time-synchronized positioning coordinate sequence. The data fusion unit performs a multi-source data consistency check based on the GNSS observation data output by the positioning and acquisition unit and the pass records provided by the RSU, obtaining the observation probability. The trajectory reliability calculation unit calculates the transition probability. and through probability multiplication The overall reliability of the trajectory is obtained. The data communication and anomaly triggering unit compares the reliability calculation result with historical statistical thresholds (μ, σ). When an anomaly occurs, it is automatically flagged and reported to the toll platform via the communication module, triggering a manual audit process.

[0023] The roadside unit includes: a perception and acquisition unit (containing an RFID identification module, a visual inspection module, or a millimeter-wave radar module), a time synchronization and coordinate correction unit, an edge computing unit, and a communication and caching unit. The perception and acquisition unit, depending on the traffic scenario, selects an RFID identification module, a visual inspection module, or a millimeter-wave radar module to acquire vehicle identification information, passage time, and lateral and longitudinal speeds; it automatically generates a passage record when a vehicle passes through the detection area. The time synchronization and coordinate correction unit, based on the GNSS timing signal and local clock synchronization mechanism, aligns the RSU-side timestamp with the vehicle positioning time, calculates the vehicle's actual coordinates at the passage section, and outputs the spatiotemporal matching result. The edge computing unit preprocesses and compresses the data collected by the perception and acquisition unit and the time synchronization and coordinate correction unit, and calculates the passage probability. Its definition is: Simultaneously, the detection time difference between adjacent RSU nodes is calculated to assist the toll collection platform in determining path continuity. The communication and caching unit integrates a cellular communication or C-V2X module for bidirectional data exchange with the toll collection platform; in cases of communication instability, a local caching mechanism is activated to store the most recent... The detection records are temporarily stored in the memory of the edge node and uploaded in batches after the network is restored to avoid data loss.

[0024] The toll collection platform server includes: a data receiving and preprocessing unit, a map matching and candidate path generation unit, a multi-source fusion and confidence calculation unit, and an anomaly detection and auditing unit. Specifically, the data receiving and preprocessing unit performs time synchronization, noise reduction, formatting, and deduplication operations on the raw data streams uploaded by the vehicle-mounted unit and roadside unit, outputting a sequence of trajectory points on a unified timeline. The map matching and candidate path generation unit generates candidate paths based on an electronic map database, using an improved HMM map matching algorithm to calculate the matching probability between each observation point and a road segment. and output a set of candidate paths. The multi-source fusion and confidence calculation unit calculates the probability of vehicle-mounted GNSS observations. RSU detection probability and path transition probability Calculate the overall confidence level of the trajectory: The anomaly detection and auditing unit dynamically determines the confidence threshold based on the historical statistical mean μ and standard deviation σ. When When this occurs, it is automatically marked as an abnormal trajectory, triggering a manual audit or system review process; when At that time, the system will automatically calculate and archive the results.

[0025] The billing and interaction module includes: a bill generation unit, a bill push and user interaction unit, a complaint and source tracing processing unit, and a manual audit and arbitration interface unit. The bill generation unit automatically completes path metering and cost calculation based on the data interface between the trusted fusion results of the charging platform and the rate approval system. When the fusion trustworthiness... When the system considers the trajectory valid, it generates a bill; if The bill generation process is interrupted and marked as pending audit. The bill push and user interaction unit, after signing and encrypting the bill generated by the bill generation unit, pushes it to the user terminal (such as an APP, mini-program, or vehicle terminal) via mobile network or Internet interface, while recording the push status. The complaint and traceability processing unit is used to receive user-submitted abnormal bill appeals. The system retrieves the original GNSS track, RSU record, and confidence calculation log based on the bill number to achieve end-to-end traceability and verification. When the bill generation unit determines that the bill is abnormal or the confidence level is below the threshold, the manual audit and arbitration interface unit automatically transfers the record to the manual audit platform, along with a complete track, time sequence, and confidence level report, supporting review by the arbitration department.

[0026] like Figure 2 As shown in the figure, this embodiment uses an abnormal toll detection method based on the credibility of the GNSS trajectory. During vehicle operation, the on-board unit collects GNSS positioning data and periodically uploads it to the toll platform server in the form of a trajectory. When the vehicle passes a roadside unit, the roadside unit collects the vehicle's identity information and corresponding timestamp through visual recognition, radio frequency identification, or vehicle-to-infrastructure communication, and uploads it to the toll platform. The toll platform cleans the received GNSS trajectory data, removes obvious error points, drift points, and non-compliant data, and performs map matching on the trajectory in conjunction with electronic map information to generate a set of candidate paths that the vehicle may take. The toll platform performs a multi-source information fusion consistency check on the on-board unit's GNSS trajectory, electronic map candidate paths, and roadside unit vehicle passing records, and uses a transition model to calculate the transition probability of the trajectory between each candidate path. After multiplying the probability to obtain the credibility of the vehicle from the starting position to the ending position, the overall confidence probability is compared with a preset threshold or historical statistical data. When it is higher than the credibility threshold, the trajectory is confirmed to be reliable and toll is completed accordingly. When it is lower than the threshold, the platform automatically marks the trajectory as abnormal and archives it to trigger manual audit.

[0027] The multi-source information fusion consistency test is specifically as follows: the initial confidence level is calculated based on the geometric distance between the GNSS trajectory and the candidate path; the confidence level of the corresponding candidate path is corrected and weighted based on the vehicle observation results of the roadside unit; and finally, the path confidence level distribution after multi-source information cross-validation is obtained.

[0028] The aforementioned transition model is as follows: based on the road topology of the electronic map, the road is divided into several discrete state nodes. Each node represents a possible road segment where the vehicle may be located; an observation sequence is established based on the vehicle's GNSS trajectory points and timestamps. This establishes a state-observation correspondence, providing input for subsequent probabilistic reasoning.

[0029] The core optimization objective function used in the training phase of the aforementioned transfer model is: ,in: Indicates GNSS signal at the state node The likelihood probability is calculated from the localization covariance matrix: .

[0030] The experiment was conducted in a real-world application scenario, verifying the method under the free-flow tolling verification test platform in Haikou City. The test environment included: vehicles equipped with a BeiDou / GPS dual-mode receiver (1Hz update rate), RSU units using a C-V2X communication module, and a server running a Linux system with a 32-core CPU and 128GB of memory. The experiment used real road networks and measured vehicle trajectories as data input, and three sets of control experiments were run using the GNSS trajectory reliability-based abnormal tolling detection method of this invention. A total of 36,501 trajectories were used in the experiment, with the test area being Haikou City, Hainan Province. The spatiotemporal distribution of the trajectories is as follows: Figure 4 As shown, the calculated reliability probability, position error, and velocity error statistics of the embodiment results are as follows: Figures 5-7 As shown in the figure, the average billing accuracy rate is 99.999998%; the false collection rate decreased from 0.30% of the traditional GNSS single-source algorithm to 0.039%; the missed billing rate decreased from 0.25% to 0.020%; the anomaly detection accuracy rate increased from 84.2% to 98.6%; and the average trajectory reliability stability increased from 0.91 to 0.98. It is evident that this invention maintains high reliability output even in complex traffic scenarios (such as tunnels and obstructed areas), verifying the robustness and real-time performance of the described multi-source fusion model.

[0031] Compared with existing technologies, this invention employs [a specific technique] in the multi-source observation fusion process. The formal method replaces traditional single-source likelihood calculation, thereby enhancing the confidence information at the observation level; in the state transition modeling stage, Hidden Markov Model (HMM) transition probabilities are introduced. Instead of relying on empirical thresholds, it implements temporal consistency constraints; in the credibility assessment stage, it adopts... The cumulative multiplication model establishes a prior-observation-posterior closed loop; in the audit judgment stage, dynamic statistical thresholds are used. It replaces fixed standards to achieve adaptive anomaly recognition.

[0032] The overall results show that, under the same data conditions, the overall trajectory reliability of the present invention is improved by about 7.7%, and the audit triggering misjudgment rate is reduced by 73.5%, achieving a simultaneous improvement in the reliability and verifiability of the free-flow billing system.

[0033] The above-described specific implementations can be partially adjusted by those skilled in the art in different ways without departing from the principles and purpose of the present invention. The scope of protection of the present invention is defined by the claims and is not limited to the above-described specific implementations. All implementation schemes within the scope of the claims are bound by the present invention.

Claims

1. A method for detecting abnormal tolling based on GNSS trajectory credibility, characterized in that, The system works as follows: ① GNSS positioning data is collected by the vehicle-mounted unit during vehicle operation and uploaded to the toll platform server in real time in the form of a trajectory; ② Vehicle identity information, timestamps, and vehicle movement information are obtained by the roadside unit when a vehicle passes by and uploaded to the toll platform server. The data center of the toll platform server then preprocesses the received trajectory data and generates a set of candidate paths that the vehicle may take using a map matching algorithm. The system performs a multi-source information fusion consistency check on the GNSS trajectory of the vehicle-mounted unit, the candidate paths on the electronic map, and the vehicle passing records of the roadside unit, and calculates the trajectory credibility. Abnormal tolling is detected by filtering the trajectory credibility.

2. The abnormal charging detection method based on GNSS trajectory reliability according to claim 1, characterized in that, The trajectory credibility wherein: , satellite navigation observation credibility , positioning error covariance matrix, is an observation bias vector; is a road side unit observation credibility, if the vehicle is observed within the RSU detection area, then , otherwise .

3. The method of claim 1, wherein the GNSS trajectory credibility based toll fraud detection method is characterized by, The multi-source information fusion consistency test specifically involves: calculating the initial confidence level based on the geometric distance between the GNSS trajectory and the candidate path; and correcting and weighting the confidence level of the corresponding candidate path based on the vehicle observation results of the roadside unit. Finally, the path confidence distribution, which has been cross-validated by multi-source information, is obtained.

4. The method of claim 1-3, wherein the method comprises: include: ​ 1) Vehicle unit collects vehicle GNSS positioning information at a frequency of 1 Hz and calculates the covariance , and reports to the toll platform in real time, while the roadside unit is arranged at the key section of the free flow channel to collect vehicle identity ID, timestamp and motion information to form a vehicle passing record; 2) The charging platform receives the above two types of data, performs time synchronization and outlier rejection, and generates a candidate path set through a map matching algorithm ; 3) Calculate observation probability for each candidate path using transition model with transition probability and multiply them sequentially: ; 4) take the maximum confidence The corresponding path as the actual driving path of the vehicle; When an exception audit is triggered by the system.

5. The abnormal charging detection method based on GNSS trajectory reliability according to claim 4, characterized in that, The transfer model divides roads into several discrete state nodes according to the road topology of the electronic map Each node represents a road section where the vehicle can be; an observation sequence is established according to the vehicle-mounted GNSS track point and time stamp , forming a state-observation correspondence relationship to provide input for subsequent probability inference; the core optimization objective function used in the training stage of the transfer model is: , wherein: represents the likelihood probability of the GNSS signal at the state node , which is calculated by the positioning covariance matrix: .

6. A GNSS trajectory plausibility based toll fraud detection system implementing the method of any of claims 1-5, characterized in that, include: The system comprises vehicle-mounted units, roadside units, a data center at the toll platform server, a data fusion system, a billing generation unit, a toll platform, and a manual auditing platform. Specifically: the vehicle-mounted units collect GNSS positioning data in real time during vehicle operation and upload trajectory information to the toll platform server's data center via a communication module; roadside units are deployed at key road nodes susceptible to GNSS interference to collect vehicle passage information; the toll platform server's data center receives data uploaded from the vehicle-mounted and roadside units and combines it with electronic map information to perform trajectory cleaning and candidate path generation, providing a basis for subsequent consistency verification; the data fusion system performs fusion processing and credibility calculation on multi-source information; the billing generation unit completes rate approval and bill calculation based on the credible trajectory results, generating the corresponding bill; the toll platform distributes bills to users and provides complaint and traceability application channels; when users disagree with the billing results, they can submit appeal requests through this platform; the manual auditing platform processes trajectory data marked as abnormal by the data fusion system or appealed by users; this platform supports map information fusion, trajectory display, and replay, facilitating manual verification and avoiding mischarges or omissions caused by GNSS errors or malicious interference.

7. The abnormal charging detection system according to claim 6, characterized in that, The vehicle-mounted unit comprises a positioning acquisition unit comprising a GNSS receiver, a data fusion unit, a trajectory credibility calculation unit, and a data communication and abnormal triggering unit, wherein: the positioning acquisition unit performs positioning data acquisition according to Beidou or GPS signals, calculates a positioning error covariance matrix Sigma_GNSS, and outputs a time-synchronized positioning coordinate sequence; the data fusion unit performs multi-source data consistency verification according to GNSS observation data output by the positioning acquisition unit and through records provided by the RSU, obtains observation probability ; the trajectory credibility calculation unit calculates transition probability , and obtains overall trajectory credibility through probability multiplication ; the data communication and abnormal triggering unit compares the credibility calculation result with historical statistical thresholds (mu, sigma), and when , automatically marks the abnormality and reports the toll platform through the communication module to trigger the manual audit process.

8. The abnormal charging detection system according to claim 6, characterized in that, The roadside unit includes: a perception and acquisition unit containing an RFID identification module, a visual inspection module, or a millimeter-wave radar module; a time synchronization and coordinate correction unit; an edge computing unit; and a communication and caching unit. Specifically: the perception and acquisition unit selects an RFID identification module, a visual inspection module, or a millimeter-wave radar module based on the traffic scenario to acquire vehicle identity information, passage time, and lateral and longitudinal speeds; it automatically generates a passage record when a vehicle passes through the detection area; the time synchronization and coordinate correction unit, based on the GNSS timing signal and local clock synchronization mechanism, aligns the RSU-side timestamp with the vehicle positioning time, calculates the vehicle's actual coordinates at the passage section, and outputs the spatiotemporal matching result; the edge computing unit preprocesses and compresses the data collected by the perception and acquisition unit and the time synchronization and coordinate correction unit, and calculates the passage probability. Simultaneously, the detection time difference between adjacent RSU nodes is calculated to assist the toll collection platform in determining path continuity; the communication and caching unit integrates a cellular communication or C-V2X module for bidirectional data exchange with the toll collection platform; when communication is unstable, a local caching mechanism is activated to store the most recent... The detection records are temporarily stored in the memory of the edge node and uploaded in batches after the network is restored to avoid data loss.

9. The abnormal charging detection system according to claim 6, characterized in that, The toll collection platform server includes: a data receiving and preprocessing unit, a map matching and candidate path generation unit, a multi-source fusion and confidence calculation unit, and an anomaly detection and auditing unit. Specifically: the data receiving and preprocessing unit performs time synchronization, noise reduction, formatting, and deduplication operations on the raw data streams uploaded by the vehicle-mounted unit and roadside unit; it outputs a sequence of trajectory points on a unified time axis; the map matching and candidate path generation unit generates candidate paths based on an electronic map database and uses an improved HMM map matching algorithm to calculate the matching probability between each observation point and a road segment. and output a set of candidate paths. The multi-source fusion and confidence calculation unit calculates the probability of vehicle-mounted GNSS observations. RSU detection probability and path transition probability Calculate the overall confidence level of the trajectory: The anomaly detection and auditing unit dynamically determines the credibility threshold based on the historical statistical mean μ and standard deviation σ; when When this occurs, it is automatically marked as an abnormal trajectory, triggering a manual audit or system review process; when At that time, the system will automatically calculate and archive the results.

10. The abnormal charging detection system according to claim 6, characterized in that, The billing and interaction module includes: a bill generation unit, a bill push and user interaction unit, a complaint and source tracing processing unit, and a manual audit and arbitration interface unit. Specifically, the bill generation unit automatically completes path metering and cost calculation based on the data interface between the credible fusion results of the charging platform and the rate approval system; when the fusion credibility... When the system considers the trajectory valid, it generates a bill; if The bill generation process is interrupted and marked as pending audit; the bill push and user interaction unit sends the bill generated by the bill generation unit to the user terminal after signing and encrypting it, through the mobile network or Internet interface, and records the push status at the same time; the complaint and traceability processing unit is used to receive abnormal bill appeals submitted by users. The system retrieves the original GNSS trajectory, RSU record and confidence calculation log according to the bill number to achieve full-link traceability verification; when the bill generation unit determines that the bill is abnormal or the confidence level is lower than the threshold, the manual audit and arbitration interface unit automatically transfers the record to the manual audit platform, with a complete trajectory, time sequence and confidence level report attached, supporting the arbitration department to review.

Images