Vehicle data monitoring method based on ETC data

Abstract

The invention belongs to the technical field of ETC (electronic toll collection) vehicle monitoring, and particularly discloses a vehicle data monitoring method based on ETC data, which comprises the following steps of: constructing a vehicle driving track data set comprising a topological node sequence, a road section distance and OBU (on board unit) state change based on ETC transaction records, and comparing a current path with a historical high-frequency path to carry out deviation degree analysis, a detouring behavior is identified in combination with passing duration, and abnormal operation is detected by using equipment state switching frequency, so that path topology and multi-dimensional cooperative inspection of equipment state behaviors are fused, the illegal passing identification accuracy is remarkably improved, and the detection capability of hidden illegal passing behaviors is enhanced. And meanwhile, path cost re-checking is carried out on a specific abnormal road section when the illegal traffic risk is judged to exist, so that the refinement of risk positioning and the targeting of inspection operation can be realized, the re-checking efficiency is remarkably improved, the calculation and manual checking cost is reduced, the recognition precision of local illegal traffic behaviors is enhanced, and accurate recovery is supported.

Description

Technical Field

[0001] This invention belongs to the field of ETC vehicle monitoring technology, and specifically discloses a vehicle data monitoring method based on ETC data. Background Technology

[0002] With the acceleration of urbanization and the continuous growth of car ownership, the traffic pressure on highways is increasing day by day. Traditional manual toll collection is prone to congestion, time-consuming and energy-intensive. ETC (Electronic Toll Collection) achieves automatic vehicle identification and automatic toll deduction through wireless communication between on-board electronic tags and roadside equipment, which greatly reduces queuing time and effectively alleviates traffic congestion. However, in actual application, some vehicles have violated traffic rules, resulting in toll collection errors and seriously disrupting the toll collection order.

[0003] To effectively address the aforementioned issues, intelligent monitoring and enforcement technologies for ETC-enabled vehicles violating traffic rules have been continuously developing in recent years. Existing technical solutions include data-driven dynamic prediction enforcement methods. For example, the Chinese invention patent with publication number CN113053114A proposes a dynamic prediction enforcement method for highway exit stations and exit times for vehicles violating traffic rules. This method predicts the vehicle's possible exit and exit time based on the vehicle's historical travel habits and current travel route recorded by the highway's ETC toll gates. After a vehicle enters the highway, the toll station lane software system and the ETC gantry monitoring system observe the gantries the vehicle has passed through. A Bayesian method is used to predict the exit, ensuring that the exit toll station with the highest conditional probability is the most likely exit toll station, and providing a predicted arrival time at that exit toll station. This allows law enforcement personnel to proactively conduct enforcement and control at the target exit toll station, intercepting violating vehicles and effectively improving the success rate of intercepting illegal vehicles.

[0004] The above-mentioned scheme judges violations based on exit time deviation as the core criterion, lacking fine-grained topological analysis of vehicle travel paths. It relies solely on the time-series passage data of ETC gantries, which is easily affected by passage delays caused by objective factors such as traffic congestion, severe weather, or emergencies. This may lead to misjudgments of time deviations, resulting in false alarms and greatly weakening the detection accuracy of the inspection system.

[0005] Furthermore, the aforementioned scheme primarily relies on modeling the spatiotemporal trajectory sequence of vehicles between ETC gantries, failing to fully integrate the operational status information of the on-board electronic tags. Given that the ETC toll collection mechanism essentially depends on stable communication between the on-board electronic tags and roadside units, illegal passage is often accompanied by abnormal operation of the on-board electronic tags, such as frequent plugging and unplugging, signal jamming, etc. Such abnormal device status is key auxiliary evidence for identifying illegal passage; however, the lack of analysis of the on-board electronic tag's operational status easily leads to a single basis for judgment, resulting in an increased false positive rate.

[0006] Furthermore, the above-mentioned scheme focuses on end-to-end path prediction of the vehicle's overall journey, lacks the ability to identify local abnormal road sections during the journey, and cannot accurately locate the specific geographical area where the violation occurred, which is not conducive to subsequent accurate tracking and recovery. Summary of the Invention

[0007] Therefore, one objective of this application is to provide a vehicle data monitoring method based on ETC data. By constructing a multi-dimensional collaborative inspection system that integrates path topology analysis, vehicle electronic tag device behavior modeling, and local correlation mining, the method aims to optimize the process from macro-level assessment to precise source tracing, comprehensively improve the detection capability of ETC violation behavior, and effectively solve the problems mentioned in the background art.

[0008] The objective of this invention can be achieved through the following technical solution: A vehicle data monitoring method based on ETC data, comprising the following steps: Step 1: Real-time acquisition of vehicle toll station location, transaction timestamp and vehicle electronic tag device status data from the transaction records of the ETC system, and construction of a vehicle driving trajectory dataset containing topological node sequence, adjacent node travel distance and device status change records.

[0009] Step 2: Perform path compliance analysis based on vehicle driving trajectory dataset. Calculate the node overlap rate between the real-time path and the baseline path to generate a path deviation coefficient. Combine this with abnormal travel time detection to generate abnormal detour markers, forming a spatial behavior feature set.

[0010] Step 3: Statistically analyze the frequency of device state switching per unit mileage in the vehicle driving trajectory dataset, and assess the risk of illegal passage by integrating the spatial behavior feature set and the frequency of device state switching per unit mileage through overall correlation and local correlation.

[0011] Step 4: When a risk of illegal passage is assessed, identify the road segment to be reviewed and trigger the route cost review process for the road segment to be reviewed.

[0012] Combining all the above technical solutions, the positive effects of this invention are as follows: 1. This invention constructs a vehicle driving trajectory dataset based on ETC transaction records, which includes topological node sequences, road segment distances, and OBU status changes. By comparing the current path with historical high-frequency paths to analyze deviation, and combining travel duration to identify detour behavior, and using the frequency of device status switching to detect abnormal operations, this invention integrates multi-dimensional collaborative auditing of path topology and device status behavior, significantly improving the accuracy of identifying vehicle violation risks, enhancing the ability to detect concealed violations, and achieving efficient and accurate intelligent auditing.

[0013] 2. When determining the risk of illegal passage, this invention focuses on reviewing the route cost of specific abnormal road sections, which can achieve refined risk positioning and targeted inspection operations. Compared with whole vehicle or full route review, this method significantly improves review efficiency, reduces calculation and manual verification costs, enhances the accuracy of identifying local illegal passage, supports accurate collection, and effectively improves the response speed and execution efficiency of the inspection system. Attached Figure Description

[0014] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0015] Figure 1 This is a diagram illustrating the implementation steps of the method of the present invention.

[0016] Figure 2 This is an application operation diagram of the present invention, which integrates spatial behavior feature sets and unit mileage device state switching frequency through overall and local correlation.

[0017] Figure 3 This is a schematic diagram illustrating the determination of road sections to be reviewed when a risk of illegal passage is assessed in this invention. Detailed Implementation

[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. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0019] See Figure 1 As shown, this invention proposes a vehicle data monitoring method based on ETC data, including the following steps: Step 1: Real-time acquisition of vehicle toll station location, transaction timestamp, and vehicle electronic tag device status data from the transaction records of the ETC system, and construction of a vehicle driving trajectory dataset containing topological node sequence, adjacent node travel distance, and device status change records.

[0020] It's important to understand that the aforementioned vehicle-mounted electronic tag, as the core on-board terminal of the ETC system, stores vehicle registration information, including license plate number, vehicle type classification, and other key data. It communicates with the roadside unit at the toll station via microwave using dedicated short-range communication technology to achieve vehicle identification and automatic toll deduction. The vehicle-mounted electronic tag has multiple operating states, primarily including an active state, supporting normal transactions; and an inactive state, such as being removed, having its signal blocked, or being deactivated, resulting in the inability to complete effective communication and toll deduction operations. State switching behavior can serve as an important monitoring indicator for identifying abnormal driving.

[0021] Preferably, the specific implementation process of step 1 is as follows: extract the vehicle identifier, toll station code, geographical coordinates, transaction timestamp, and vehicle electronic tag operation status information from each transaction record from the ETC system.

[0022] Based on the spatial topology of toll stations, the sequence of toll stations that vehicles pass through is linearly sorted to construct an ordered chain of geographical nodes.

[0023] The geographical locations of adjacent toll stations are abstracted as topological nodes in the road network space, and the passage events of the same vehicle are arranged in time sequence according to the timestamp to form continuous spatial trajectory segments at the vehicle level. The trajectory segments are then integrated to generate a vehicle driving trajectory dataset.

[0024] Each data unit in the dataset consists of the topological node location, the actual travel distance between adjacent nodes, and the vehicle electronic tag status change record.

[0025] The vehicle trajectory dataset formed above serves as the basic input for subsequent path behavior modeling and anomaly detection. It possesses the characteristics of spatiotemporal continuity and device behavior traceability, supporting high-precision vehicle path reconstruction and behavior analysis.

[0026] Step 2: Perform path compliance analysis based on vehicle driving trajectory dataset. Calculate the node overlap rate between the real-time path and the baseline path to generate a path deviation coefficient. Combine this with abnormal passage time interval detection to generate abnormal detour markers, forming a spatial behavior feature set.

[0027] Optionally, step 2 is implemented as follows: retrieve all historical passage records of the vehicle within a preset time window from the ETC system. Each record contains a toll station topology sequence arranged in order of timestamps, that is, a chain of toll station nodes from the entrance to the exit. Then, the toll station sequence in each record is regarded as a complete historical driving path.

[0028] It should be noted that setting a preset time window aims to aggregate multiple historical travel records of vehicles in similar travel scenarios, avoiding the randomness and bias in path pattern recognition caused by relying on only a single path sample, and improving the statistical reliability and representativeness of high-frequency driving path extraction. Specifically, the selection of the preset time window can be dynamically configured according to vehicle travel behavior characteristics, such as based on travel cycle patterns, to ensure that the collected path data fully reflects the typical driving habits of vehicles.

[0029] Each historical travel route is measured using path similarity metrics and combined with cluster analysis to form several path clusters. The number of historical travel records contained in each cluster is counted, and the cluster with the most records is selected as the dominant travel mode cluster. The travel routes within this cluster are then extracted as the baseline routes for the vehicle.

[0030] In the specific implementation of the above scheme, a path similarity metric is used for each historical driving path to quantify the structural similarity between different historical driving paths. This provides a computable distance or similarity matrix for subsequent clustering analysis, thereby supporting the grouping of paths with similar topological structures into the same category. Specifically, the similarity between paths can be calculated using the longest common subsequence algorithm. Let's assume two historical driving paths... and Let and represent sequences consisting of ordered toll station nodes, respectively. The longest common subsequence algorithm is used to identify the longest common subsequence between the two, where the longest common subsequence does not need to be consecutive, but must maintain order. The path similarity is defined as . ,in This represents the number of nodes in the longest common subsequence of two historical travel paths. , These represent the number of tollbooth nodes in the two historical travel routes, respectively.

[0031] It is important to understand that multiplying the numerator by 2 in the path similarity calculation formula is to normalize the length difference between the two sequences, thereby ensuring that the value range of the similarity index is reasonably constrained within the interval [0, 1]. The larger the value, the higher the degree of overlap between the two paths in the node sequence.

[0032] Furthermore, selecting the driving path within the dominant travel mode cluster as the vehicle's baseline path aims to characterize its long-established high-frequency travel behavior patterns. Vehicle users typically exhibit significant behavioral inertia in their daily commutes, tending to choose familiar, efficient, and economical fixed routes. Identifying the most frequently traveled path clusters through clustering algorithms can effectively capture individual travel preferences and avoid misclassifying occasional paths as normal behavior. Although the paths within the dominant clusters formed through path similarity measurement and clustering analysis have high structural consistency, local differences in node sequences may still exist. Therefore, it is necessary to select the most representative path within the cluster as the baseline, typically the path with the highest average similarity to the remaining paths among all paths in the cluster.

[0033] The currently generated vehicle trajectory is spatially topologically aligned with the established baseline path. The overlap rate of nodes at the toll station level is calculated, and the path deviation coefficient is generated after normalization.

[0034] The node overlap rate is defined as the ratio of the number of toll station nodes coexisting in the two paths to the maximum value of the total number of nodes in both paths. Since the higher the node overlap rate, the higher the consistency between the vehicle's trajectory and the determined baseline path, while the path deviation reflects the inconsistency between the vehicle's trajectory and the determined baseline path, the path deviation coefficient is obtained by inversely normalizing the node overlap rate. Specifically, the path deviation coefficient can be obtained by subtracting the node overlap rate from the value 1, which realizes the quantitative characterization of the degree of anomaly. At the same time, the value is restricted to the interval [0, 1]. The larger the value, the more significant the deviation between the current path and the baseline path.

[0035] It should be noted that toll station nodes are used as the basic representation unit of the path topology in path deviation analysis because they possess significant structural and semantic characteristics in the highway network. As key control points in the road network, the completeness of the toll station node sequence directly reflects the authenticity of the vehicle's travel path. Therefore, this design is highly sensitive to illegal passage behaviors such as detours and skipping toll stations. Any additions, deletions, or changes in the order of nodes deviating from the baseline path can be effectively captured, thereby improving the accuracy and interpretability of path anomaly identification.

[0036] The current vehicle trajectory is segmented according to the adjacent topology nodes defined in the data unit, and the actual passage time interval between each pair of consecutive toll stations is extracted.

[0037] Extract the travel time sample set corresponding to each road segment from all historical travel records corresponding to the dominant travel mode cluster, statistically distribute the travel time sample set, and set the historical threshold interval using the quantile method.

[0038] It should be noted that the statistical distribution of the travel time sample set mentioned above refers to an empirical distribution modeling of multiple travel time observations corresponding to each segment of the baseline route. Specifically, a travel time histogram is constructed to visualize its distribution characteristics. Based on this, the quantile method is used to determine the normal travel time threshold range for each segment. Specifically, a portion of the travel time sample set is taken... Quantiles as the lower limit The quantile, as the upper limit, constitutes the threshold range for historical passage time.

[0039] The actual travel time of each road segment in the current vehicle's travel trajectory is compared with the historical threshold range of the corresponding road segment. If the actual travel time of a certain road segment exceeds the upper limit of the historical threshold range, it is determined that there is abnormal driving behavior in that road segment, and an abnormal detour mark with a location label is generated.

[0040] It should be noted that the travel time of real vehicles on the same road segment usually follows a certain statistical distribution. Using historical records in high-frequency paths as a benchmark ensures that the time threshold reflects the vehicle's actual driving habits on that road segment. Time intervals that deviate significantly from this distribution may indicate abnormal detours.

[0041] The path deviation coefficient corresponding to the current vehicle's driving trajectory is integrated with the generated abnormal detour markers to form a spatial behavior feature set.

[0042] The aforementioned spatial behavior feature set combines high-confidence anomaly detection based on path deviation and time skewness, which is more effective than methods relying solely on time skewness in reducing false alarms caused by time delays due to external factors such as traffic congestion. By integrating spatial and temporal data analysis, a more comprehensive and reliable monitoring and evaluation of vehicle driving behavior is achieved.

[0043] Step 3: Statistically analyze the frequency of device state switching per unit mileage in the vehicle driving trajectory dataset, and assess the risk of illegal passage by integrating the spatial behavior feature set and the frequency of device state switching per unit mileage through overall correlation and local correlation.

[0044] In one feasible approach of the above scheme, statistical analysis of the device state switching frequency per unit mileage in the vehicle trajectory dataset includes the following: segment-by-segment analysis of road segments between adjacent topology nodes based on the vehicle electronic tag state change records of each data unit in the vehicle trajectory dataset.

[0045] For each road segment, the total number of on-board electronic tag status changes during the passage is extracted, and combined with the actual travel distance of that segment, the frequency of status change per unit mileage is calculated. A higher frequency of status change per unit mileage reflects more frequent abnormal operations of the on-board electronic tags within that segment. Such behavior may be associated with human intervention in the equipment to circumvent billing; therefore, this indicator can serve as an important behavioral characteristic for identifying illegal passage.

[0046] Applying the above operations, the frequency of status changes per unit mileage of a road segment can be calculated by dividing the total number of times the status of the on-board electronic tag of the road segment is switched by the actual travel distance of the road segment.

[0047] The unit mileage equipment state switching frequency is calculated by linearly weighting the unit mileage state change frequency of each road segment using the actual travel distance of each road segment as the weight.

[0048] As an explanation of the above operations, when aggregating and statistically analyzing the frequency of state changes per unit mileage for each road segment, a linear weighted average is used, with the actual travel distance of the road segment as the weight. This aims to reflect the differences in spatial contribution among different road segments. Longer road segments, due to their broader coverage of the driving process, have more statistically representative equipment state behavior and should be given higher weights. Conversely, high-frequency switching on shorter road segments may originate from local interference or transient anomalies; equal weighting could easily distort the overall index due to temporary disturbances. The distance-weighted mechanism effectively suppresses the excessive impact of isolated short-segment anomalies on the global index, improves the robustness and stability of the equipment state switching frequency per unit mileage, and ensures that the comprehensive index accurately reflects the overall pattern and true risk level of on-board electronic tag state changes throughout the vehicle's journey, possessing stronger anomaly identification credibility and statistical rationality.

[0049] Specifically, when using the actual travel distance of each road segment as a weight, the weight should be the proportion of the actual travel distance of that road segment to the total distance of the vehicle's entire travel trajectory.

[0050] In another possible implementation of the above scheme, the overall association is implemented as follows: the path deviation coefficient and the device state switching frequency per unit mileage in the spatial behavior feature set are compared with the corresponding allowable thresholds. When both the path deviation coefficient and the device state switching frequency per unit mileage exceed the corresponding allowable thresholds, an overall association is identified, which effectively improves the specificity of the judgment and avoids false triggering caused by fluctuations in a single indicator.

[0051] Local correlation identification is not triggered when no overall correlation exists, in order to control audit costs and false alarm rate.

[0052] In the example applied to the above operations, the path deviation coefficient and the allowable threshold for the frequency of device state switching per unit mileage respectively characterize the acceptable boundaries of abnormal vehicle driving paths and abnormal operation of on-board electronic tags, and are key criteria for distinguishing between normal behavior fluctuations and high-risk passage suspicions.

[0053] The permissible threshold for the path deviation coefficient can be determined by analyzing the path deviation distribution of normal vehicle groups and selecting the high quantile as the threshold to ensure that the vast majority of compliant passages are not misjudged. The permissible threshold for the frequency of device state switching per unit mileage can be determined by statistical methods based on the empirical distribution of the frequency of state switching of on-board electronic devices during historical driving of a sample set of vehicles with no violations. Specifically, the permissible threshold can be determined by fitting a probability distribution model and combining it with the significance level.

[0054] It should be explained that the overall correlation analysis aims to identify the collaborative patterns between vehicle driving behavior and abnormal status of on-board electronic tag devices at a macro level, in order to determine whether there is a tendency for systemic illegal traffic behavior. Therefore, the path deviation coefficient and the device status switching frequency per unit mileage are chosen for joint determination, rather than the segment travel time deviation, which is also a spatial behavior feature. This is because: path deviation reflects the overall deviation of the vehicle from the baseline path in the entire travel topology, encompassing global anomalies such as detours and skipping stops; while segment travel time deviation only applies to a single adjacent node interval, representing a local operating state, and is easily affected by non-illegal traffic factors such as traffic flow fluctuations, lacking a globally consistent criterion.

[0055] Therefore, the collaborative analysis based on path deviation and equipment status can improve the overall accuracy and confidence of identifying high-risk vehicles, which is in line with the design goals of macro risk assessment.

[0056] In another possible implementation of the above scheme, the local association is described in the following implementation process: extract abnormal detour markers with geographical location labels from the spatial behavior feature set. These markers correspond to specific road segments between adjacent toll stations. The set of all road segments marked as abnormal detours is recorded as the abnormal driving road segment set, which represents that vehicles may have abnormal path selection or detour behavior in these road segments.

[0057] The frequency of state change per unit mileage of each road segment in the vehicle's driving trajectory is compared with the allowable threshold of this parameter. The set of road segments whose frequency of state change per unit mileage exceeds the allowable threshold is called the abnormal equipment road segment set, which reflects the behavior range of frequent switching of vehicle electronic tag status and possible human intervention.

[0058] The intersection and union of the abnormal driving segment set and the abnormal equipment segment set are calculated, and the ratio of the intersection to the union is taken as the segment overlap ratio, which reflects the co-occurrence ratio of the two types of anomalies at the segment level. This indicates that the vehicle simultaneously exhibits illegal route selection and evasive equipment operation on a specific segment. The ratio is then compared with the local association threshold. When the segment overlap ratio is higher than the local association threshold, local association is identified.

[0059] The aforementioned local association threshold is used to determine the spatial co-occurrence significance of abnormal paths and equipment behavior at the road segment level. Essentially, it is the critical value of the road segment overlap ratio. Specifically, the average and standard deviation of the road segment overlap ratio can be calculated by selecting a sample set of vehicles that have been confirmed as violating traffic regulations, and then the local association threshold can be determined by using the multiple of the average and standard deviation.

[0060] It is important to understand that, assuming an overall correlation exists, local correlation analysis aims to identify the geographical clustering and synergistic characteristics of path anomalies and equipment anomalies from a spatial dimension, verifying the degree of co-occurrence of the two on specific road segments. This analysis can further enhance the fine-grained verification and spatial interpretability of illegal passage behavior based on macro-risk assessment, significantly improving the accuracy, credibility, and completeness of the evidence chain of risk judgment, and enhancing the refinement of illegal passage behavior identification and the effectiveness of inspection and enforcement.

[0061] The above-mentioned application of integrating spatial behavior feature sets and unit mileage device state switching frequency through overall and local correlation is discussed in [reference needed]. Figure 2 As shown.

[0062] Furthermore, the risk assessment for illegal passage is conducted as follows: when the spatial behavior feature set and the device state switching frequency per unit mileage are correlated both overall and locally, a high-risk violation is assessed; conversely, when only an overall correlation exists, a medium-risk violation is assessed.

[0063] Step 4: When a risk of illegal passage is assessed, identify the road segment to be reviewed and trigger the route cost review process for the road segment to be reviewed.

[0064] In the preferred implementation of the above scheme, see [reference needed]. Figure 3 As shown, when a risk of illegal passage is assessed, the road sections to be reviewed include the following: when a high risk of illegal passage is assessed, the intersection of the abnormal driving road section set and the abnormal equipment road section set is taken as the road section to be reviewed.

[0065] It is important to understand that the intersecting road segment demonstrates the co-occurrence of path anomalies and equipment status anomalies in geospatial space. This indicates that vehicles not only deviate from the baseline driving path within a specific section but also exhibit suspected human intervention behaviors such as frequent switching of on-board electronic tags. The combined occurrence of these two types of anomalies significantly enhances the clarity of intent and the relevance of evidence for violations, possessing high-risk identification value. Using this intersection as the road segment to be reviewed, compared to indiscriminately reviewing all segments of the entire path, enables risk focus and targeted inspection, effectively reducing the scope of verification, improving review efficiency, and optimizing resource allocation.

[0066] For each road segment to be reviewed, the actual travel distance between its adjacent nodes is extracted, and the road segments to be reviewed are sorted in descending order of actual travel distance to obtain the route cost review order for road segments with high risk of illegal passage.

[0067] It is also important to understand that, since long-distance road sections account for a high proportion of the overall toll revenue, the financial losses caused by violations are greater. Prioritizing the review of long road sections can maximize the recovery efficiency of inspection resources. At the same time, long road sections usually cross multiple geographical areas or toll management boundaries, with complex path topologies and richer gantry data and behavioral characteristics involved. This is conducive to building a complete and credible chain of evidence through cross-verification of multi-source data. Therefore, planning the review order after identifying the road sections to be reviewed is beneficial to achieving priority-driven and risk-oriented inspection processes.

[0068] When a medium-risk violation is identified, the abnormal driving section set and the abnormal equipment section set are merged into a section set to be reviewed.

[0069] It is important to understand that when a vehicle is determined to have a medium risk of violation, it means that the overall correlation is met but the local high-confidence co-occurrence is not triggered. In this case, in order to ensure the integrity of the inspection, the union of the abnormal driving segment set and the abnormal equipment segment set is selected to cover all segments that show significant abnormalities in route selection or equipment behavior, forming a broad-coverage and comprehensive inspection basis.

[0070] For road segments with concentrated areas of road to be reviewed, the ratio of the actual travel distance between adjacent nodes to the total travel distance in the vehicle's trajectory is used as the basic scale factor. This reflects the relative spatial contribution of the road segment in the entire vehicle's journey and characterizes the scale importance of the road segment in the overall travel mileage and potential toll composition. The higher the distance ratio, the greater its impact on the total cost.

[0071] Weights are assigned to driving anomalies and equipment anomalies. The anomaly intensity of road segments with concentrated anomalies is quantified according to their anomaly types, and anomaly scores are calculated by weighting the assigned weights.

[0072] It should be noted that assigning weights to driving anomalies and equipment anomalies essentially quantifies the relative importance of the sets of abnormal driving segments and abnormal equipment segments in terms of their risk contribution. This weight reflects the causal contribution of these two types of anomalies to triggering traffic violations. Specifically, this can be achieved by constructing a confirmed sample set of vehicles violating traffic regulations, combining route analysis and equipment status audits, and then determining the cause of each sample: identifying whether the cause of the violation stems from route detours or equipment intervention. Based on this classification, the proportion of violations caused by driving anomalies and equipment anomalies is statistically calculated, and this proportion is used as the basis for weight allocation.

[0073] Specifically, the anomaly intensity quantification for road segments with concentrated anomalies during the review process is implemented according to their respective anomaly types as follows: For abnormal travel segments, the determination is based on the degree to which the actual travel time exceeds the historical threshold range; therefore, its anomaly intensity can be defined as the normalized exceedance ratio, i.e. ,in Indicates the intensity of abnormality in abnormal road sections. This indicates the actual travel time for the abnormal traffic section. This indicates the upper limit of the historical threshold range for the passage time of abnormal driving sections.

[0074] For road sections with abnormal equipment, identification is based on the degree to which the frequency of equipment status changes per unit mileage exceeds a preset allowable threshold, and the abnormality intensity is quantified as follows: ,in Indicates the abnormal intensity of the road section with abnormal equipment. This indicates the frequency of equipment status changes per unit mileage in road sections with abnormal equipment. This represents the permissible threshold for the frequency of equipment status changes per unit mileage.

[0075] For each road segment in the review road segment set, a comprehensive risk score is obtained by establishing a comprehensive risk scoring function with the basic scale factor as the base and the anomaly score as the index.

[0076] It should be explained that the aforementioned comprehensive risk scoring function adopts a function form with the basic scale factor as the base and the anomaly score as the exponent. The base term reflects the relative mileage proportion of the road segment within the overall vehicle travel path, characterizing its potential influence weight in the toll composition and reflecting the structural importance of the space where anomalies occur. The exponent term of this function has a non-linear amplification effect: as the anomaly score increases, the risk score grows exponentially, effectively amplifying the risk representation of highly abnormal behavior, enhancing sensitivity to serious violations, and achieving a significant distinction between minor fluctuations and high-risk behaviors.

[0077] It should be noted that since the basic scale factor is between [0, 1], directly using it as the base would cause the power function value to decay as the exponent increases, contradicting the semantics of increasing risk. Therefore, the base needs to be shifted, i.e., using 1 + the basic scale factor as the new base, ensuring that the base is greater than 1. This makes the function value monotonically increase with the anomaly score, conforming to the intuitive logic of risk accumulation, while preserving the relative proportional relationship of the original scale factor. This achieves a positive synergistic amplification of anomaly intensity and road segment importance, improving the numerical stability and business interpretability of the scoring function.

[0078] The road segments to be reviewed are grouped together and sorted in descending order of comprehensive risk score to obtain the route cost review order for road segments with medium-risk violations.

[0079] The above method of sorting the comprehensive risk scores in descending order and prioritizing the review of the road sections with the highest scores can maximize the recovery of revenue with limited audit costs and improve the efficiency of fee review.

[0080] In the further optimization and implementation of the above scheme, the route cost verification process is implemented as follows:

[0081] Retrieve the actual route costs incurred by the vehicle on the road segment to be reviewed from the ETC back-end billing system.

[0082] The route cost is predicted based on the adjacent topological nodes of the road segment to be reviewed, the actual travel distance, and the regional toll rate rules.

[0083] The actual route cost of the road segment to be reviewed is compared with the predicted route cost to obtain the cost deviation, which is then compared with the allowable deviation. The allowable deviation reflects the acceptable cost error boundary of the ETC system during normal billing. Its value is usually set according to the "ETC Network Toll Collection Technical Specification" or regional billing rules. When the cost deviation is higher than the allowable deviation, an equipment status alarm is sent to the ETC system of that road segment, activating the location locking mechanism of the last toll station the vehicle passed through, and adding it to the key monitoring list.

[0084] The above operations can prevent suspected vehicles from continuing to travel or leaving the regulatory scope before being dealt with, thereby enhancing the real-time nature and closed-loop capability of the inspection.

[0085] The parameters involved in the above formula are all dimensionless and calculated numerically. The formula is a formula obtained from the most recent real situation by collecting a large amount of data and simulating it with software. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0086] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0087] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0088] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0089] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0090] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A vehicle data monitoring method based on ETC data, characterized in that, Includes the following steps: Step 1: Obtain the vehicle toll station location, transaction timestamp, and on-board electronic tag device status data in real time from the transaction records of the ETC system, and construct a vehicle driving trajectory dataset that includes topological node sequence, distance between adjacent nodes, and device status change records; Step 2: Perform path compliance analysis based on vehicle driving trajectory dataset. Calculate the node overlap rate between the real-time path and the baseline path to generate a path deviation coefficient. Combine this with abnormal travel time detection to generate abnormal detour markers, forming a spatial behavior feature set. Step 3: Statistically analyze the frequency of device state switching per unit mileage in the vehicle driving trajectory dataset, and assess the risk of illegal passage by integrating the spatial behavior feature set and the frequency of device state switching per unit mileage through overall and local correlation. Step 4: When a risk of illegal passage is assessed, identify the road segment to be reviewed and trigger the route cost review process for the road segment to be reviewed.

2. The vehicle data monitoring method based on ETC data as described in claim 1, characterized in that: The implementation process of step 1 is as follows: Extract vehicle identification, toll station code, geographic location coordinates, transaction timestamp, and on-board electronic tag operation status information from each transaction record from the ETC system; Based on the spatial topology of toll stations, the sequence of toll stations passed by vehicles is linearly sorted to construct an ordered chain of geographical nodes. The geographical locations of adjacent toll stations are abstracted as topological nodes in the road network space, and the passage events of the same vehicle are arranged in time sequence according to the timestamp to form continuous spatial trajectory segments with vehicles as the granularity. The trajectory segments are then integrated to generate a vehicle driving trajectory dataset. Each data unit in the dataset consists of the topological node location, the actual travel distance between adjacent nodes, and the vehicle electronic tag status change record.

3. The vehicle data monitoring method based on ETC data as described in claim 1, characterized in that: Step 2 includes the following: The system retrieves all historical passage records of the vehicle within a preset time window from the ETC system. Each record contains a toll station topology sequence arranged in order of timestamps, that is, a chain of toll station nodes from the entrance to the exit. The toll station sequence in each record is then regarded as a complete historical driving path. Each historical travel route is measured using path similarity and combined with cluster analysis to form several path clusters. The number of historical travel records contained in each cluster is counted, and the cluster with the most records is selected as the dominant travel mode cluster. The travel routes within this cluster are then extracted as the baseline routes for the vehicle. The current real-time generated vehicle trajectory is spatially topologically aligned with the established baseline path. The node overlap rate of the two at the toll station level is calculated, and the path deviation coefficient is generated after normalization. The current vehicle trajectory is segmented according to the adjacent topology nodes defined in the data unit, and the actual travel time between each pair of consecutive toll stations is extracted. Extract the travel time sample set corresponding to each road segment from all historical travel records corresponding to the dominant travel mode cluster, statistically distribute the travel time sample set, and set the historical threshold interval using the quantile method; The actual travel time of each road segment in the current vehicle's travel trajectory is compared with the historical threshold range of the corresponding road segment. If the actual travel time of a certain road segment exceeds the upper limit of the historical threshold range, it is determined that there is abnormal driving behavior in that road segment, and an abnormal detour mark with a location label is generated. The path deviation coefficient corresponding to the current vehicle's driving trajectory is integrated with the generated abnormal detour markers to form a spatial behavior feature set.

4. The vehicle data monitoring method based on ETC data as described in claim 2, characterized in that: The statistical analysis of device state switching frequency per unit mileage in the vehicle driving trajectory dataset includes the following: Based on the vehicle driving trajectory dataset, each data unit records the status change records of the vehicle electronic tag, and the road segments between adjacent topology nodes are analyzed segment by segment. For each road segment, extract the total number of times the vehicle electronic tag status changes during the passage, and calculate the frequency of status change per unit mileage of the road segment in combination with the actual passage distance of the road segment; The unit mileage equipment state switching frequency is calculated by linearly weighting the unit mileage state change frequency of each road segment using the actual travel distance of each road segment as the weight.

5. The vehicle data monitoring method based on ETC data as described in claim 1, characterized in that: The overall association is described in the following implementation process: The path deviation coefficient and the device state switching frequency per unit mileage in the spatial behavior feature set are compared with the corresponding allowable thresholds. When both the path deviation coefficient and the device state switching frequency per unit mileage exceed the corresponding allowable thresholds, an overall correlation is identified. Local association identification is not triggered when there is no overall association.

6. The vehicle data monitoring method based on ETC data as described in claim 5, characterized in that: The local association is described in the following implementation process: Extract abnormal detour markers with geographic location labels from the spatial behavior feature set, and denote the set of all road segments marked as abnormal detours as the abnormal driving road segment set; The frequency of state change per unit mileage of each road segment in the vehicle's driving trajectory is compared with the allowable threshold of this parameter. The set of road segments whose frequency of state change per unit mileage exceeds the allowable threshold is called the abnormal equipment road segment set. Calculate the intersection and union of the abnormal driving segment set and the abnormal equipment segment set, and then take the ratio of the intersection and union as the segment overlap ratio, and compare it with the local association threshold. When the segment overlap ratio is higher than the local association threshold, local association is identified.

7. The vehicle data monitoring method based on ETC data as described in claim 1, characterized in that: The risk assessment for illegal passage is conducted as follows: When the spatial behavior feature set and the device state switching frequency per unit mileage are correlated both overall and locally, a high-risk violation is assessed; conversely, when only an overall correlation exists, a medium-risk violation is assessed.

8. The vehicle data monitoring method based on ETC data as described in claim 6, characterized in that: When an assessment indicates a risk of traffic violations, the determination of road sections requiring review includes the following: When a high risk of violation is identified, the intersection of the abnormal driving section set and the abnormal equipment section set will be used as the section to be reviewed. For each road segment to be reviewed, the actual travel distance between its adjacent nodes is extracted, and the road segments to be reviewed are sorted in descending order of actual travel distance to obtain the route cost review order for road segments with high risk of violations.

9. A vehicle data monitoring method based on ETC data as described in claim 6, characterized in that: The determination of road sections to be reviewed when there is a risk of illegal passage also includes the following: When a medium-risk violation is identified, the abnormal driving segment set and the abnormal equipment segment set are merged into a segment set to be reviewed. For road sections with concentrated areas of road sections to be reviewed, the ratio of the actual travel distance between adjacent nodes to the total travel distance in the vehicle's trajectory is used as the basic scale factor. Weights are assigned to driving anomalies and equipment anomalies. The anomaly intensity of road sections with concentrated anomalies is quantified according to their anomaly types, and anomaly scores are calculated by weighting the assigned weights. For each road segment in the review road segment set, a comprehensive risk score is obtained by establishing a comprehensive risk scoring function with the basic scale factor as the base and the abnormal score as the index. The road segments to be reviewed are grouped together and sorted in descending order of comprehensive risk score to obtain the route cost review order for road segments with medium-risk violations.

10. The vehicle data monitoring method based on ETC data as described in claim 1, characterized in that: The route cost review process is implemented as follows: Retrieve the actual route costs incurred by the vehicle on the road segment to be reviewed from the ETC back-end billing system; The route cost to be repaid is predicted based on the adjacent topological nodes of the road segment to be reviewed and the actual travel distance, combined with the regional toll rate rules. The actual route cost of the road segment to be reviewed is compared with the predicted route cost to obtain the cost deviation, and then compared with the allowable deviation. When the cost deviation is higher than the allowable deviation, an equipment status alarm is sent to the ETC system of that road segment, activating the location locking mechanism of the last toll station the vehicle passed through, and the segment is included in the key monitoring list.

Images