CN-122020277-A - A abnormal monitoring of charge and recognition system for highway

CN122020277ACN 122020277 ACN122020277 ACN 122020277ACN-122020277-A

Abstract

The invention relates to the technical field of expressway intelligent traffic management, in particular to a charging abnormality monitoring and identifying system for an expressway. The system comprises a data acquisition module, a first processing module, a second processing module, a game decision module and a strategy execution module, wherein the data acquisition module is used for acquiring vehicle track running water data and road network topology basic data, the first processing module is used for calculating topology inconsistent energy and generating a purification subgraph according to the topology inconsistent energy, the second processing module is used for calculating an antagonistic characteristic drift rate and solving a posterior probability that a target vehicle is a real evasion fee, the game decision module is used for determining total expected utility by combining preset interception benefits and social congestion calculation cost and solving an optimal interception execution probability which maximizes the total expected utility, and the strategy execution module is used for responding to the optimal interception execution probability. The invention obviously reduces the interference of environmental noise and equipment faults on the recognition system, improves the purity and the credibility of input data, and lays a solid foundation for the follow-up accurate recognition.

Inventors

WU SHANSHAN
HOU JINQUAN
Bao Zelong
ZHANG NINGPING
WANG YANG
WEI PENGFEI
ZHAO DONG
LI WENQIANG

Assignees

甘肃新陆港科技有限公司

Dates

Publication Date: 20260512
Application Date: 20251223

Claims (8)

1. A tolling abnormality monitoring and recognition system for a highway, comprising: the data acquisition module is used for acquiring vehicle track running water data and road network topology basic data; the first processing module is used for constructing an anti-noise graph model based on vehicle track flow data and road network topology basic data, calculating topology inconsistent energy and generating a purified subgraph according to the topology inconsistent energy; the second processing module is used for inputting the purified subgraph into a graph neural network model, calculating the resistance characteristic drift rate and calculating the posterior probability of the target vehicle as a real evasion person; The game decision module is used for determining total expected utility by combining preset interception benefits and social congestion calculation cost based on the resistance characteristic drift rate and posterior probability, and solving the optimal interception execution probability for maximizing the total expected utility; And the strategy execution module is used for responding to the optimal interception execution probability, executing physical interception when the optimal interception execution probability indicates that the interception gain is greater than the information acquisition gain, executing release operation when the optimal interception execution probability indicates that the information acquisition gain is greater than or equal to the interception gain, and updating the graph neural network model by taking the current sample as an countermeasure sample.
2. The tolling anomaly monitoring and recognition system for highways of claim 1, wherein said first processing module calculates topologically inconsistent energies, comprising: invoking a node observation time stamp and a node physical position in vehicle track stream data; acquiring historical average flow velocity of a road section and a clock jitter statistic value of equipment; Calculating the time deviation between the actual observation time difference between the node and the neighborhood node and the theoretical physical road network distance running time; And determining the topological inconsistency energy based on the time deviation, the equipment clock jitter statistic value and the output result of the model characteristic mutual exclusion function.
3. The system for monitoring and identifying tolling anomalies for highways as set forth in claim 1, wherein said first processing module generates a purification subgraph from topologically inconsistent energy, comprising: calling a preset energy constraint threshold; comparing the topologically inconsistent energy with an energy constraint threshold; If the topological inconsistent energy is larger than the energy constraint threshold, judging the nodes as ghost nodes and eliminating the ghost nodes; If the topologically inconsistent energy is less than or equal to the energy constraint threshold, the nodes are preserved to construct a refined subgraph.
4. The tolling anomaly monitoring and recognition system for highways of claim 1, wherein said second processing module calculates an antagonistic characteristic drift rate comprising: collecting characteristic centroids of high-risk sample clusters in a current time window and a previous time window; calculating the centroid distance of the two feature centroids in Euclidean space; Calculating the gradient modular length of the model loss function relative to the input characteristic; Calling a preset gradient-distance mapping coefficient; an antagonistic feature drift rate is determined based on centroid distance, gradient modulo length, and gradient-to-distance mapping coefficients.
5. The system of claim 4, wherein the gradient-to-distance mapping coefficients are used to map the magnitude of the gradient to a characteristic distance space, wherein the gradient-to-distance mapping coefficients are determined based on normalization experiments against the generation network under simulated attacks.
6. A tolling anomaly monitoring and recognition system for highways as claimed in claim 1, wherein said gaming decision module determines total expected utility comprising: calling optimal interception execution probability, posterior probability, antagonistic characteristic drift rate, interception benefits and social congestion calculation cost; calculating the difference between the positive gain expectation caused by the interception execution and the negative risk expectation caused by the interception execution, and defining the difference as direct intervention effect; Calling an information value monetization coefficient, and calculating the utility of discarding the information gain acquired by interception based on the information value monetization coefficient and the antagonistic characteristic drift rate; the total expected utility is determined based on a weighted sum of the direct intervention utility and the information gain utility.
7. The system of claim 6, wherein the informative value monetization factor is used to convert observed characteristic drift into an equivalent informative economic value, wherein the informative economic value characterization system is configured to obtain short term toll benefits that the unit characteristic drift observation is willing to sacrifice.
8. The system for monitoring and identifying tolling anomalies for highways as set forth in claim 1, wherein said policy enforcement module performs physical interception or enforcement release operations, comprising: Calculating the product of the antagonistic characteristic drift rate and the information value monetization coefficient, and defining the product as information acquisition benefits; Calculating the product of interception benefits and posterior probability, and subtracting the product of social congestion calculation cost and non-evasion probability to define the interception benefits; if the interception gain is greater than the information acquisition gain, determining that the optimal interception execution probability approaches to 1, and executing physical interception; if the information acquisition benefit is greater than or equal to the interception benefit, determining that the optimal interception execution probability approaches 0, and executing release operation.

Description

A abnormal monitoring of charge and recognition system for highway Technical Field The invention relates to the technical field of expressway intelligent traffic management, in particular to a charging abnormality monitoring and identifying system for an expressway. Background At present, expressway charging monitoring mainly relies on a road side unit and a portal system to acquire vehicle track running water and road network topology data, conventional abnormality is identified by calculating physical distance and time matching degree, and interception or release is executed based on established rules; However, in the related technology, with the emergence of high-contrast fee evasion means such as ghost data injection, a monitoring system based on static physical rules exposes obvious weaknesses, on one hand, a maliciously generated ghost node logically exists but is not physically feasible, so that the judgment accuracy of a traditional model is easily interfered, on the other hand, the characteristic drift rate of fee evasion behaviors is accelerated, a fixed model is difficult to capture a novel variety, and in addition, the existing strategy lacks game thinking, cannot find an optimal solution between interception benefits, social congestion cost and acquisition of unknown attack information value, so that the system responds passively in a congestion period and cannot convert the unknown threat into training data to realize self iteration, so that a scheme is needed to solve the problems existing in the prior art. The above information disclosed in the above background section is only for enhancement of understanding of the background of the disclosure and therefore it may include information that does not form the prior art that is already known to those of ordinary skill in the art. Disclosure of Invention In order to solve the technical problems, the invention discloses a charging abnormity monitoring and identifying system for a highway, and specifically, the technical scheme of the invention comprises the following steps: the data acquisition module is used for acquiring vehicle track running water data and road network topology basic data; the first processing module is used for constructing an anti-noise graph model based on vehicle track flow data and road network topology basic data, calculating topology inconsistent energy and generating a purified subgraph according to the topology inconsistent energy; the second processing module is used for inputting the purified subgraph into a graph neural network model, calculating the resistance characteristic drift rate and calculating the posterior probability of the target vehicle as a real evasion person; The game decision module is used for determining total expected utility by combining preset interception benefits and social congestion calculation cost based on the resistance characteristic drift rate and posterior probability, and solving the optimal interception execution probability for maximizing the total expected utility; And the strategy execution module is used for responding to the optimal interception execution probability, executing physical interception when the optimal interception execution probability indicates that the interception gain is greater than the information acquisition gain, executing release operation when the optimal interception execution probability indicates that the information acquisition gain is greater than or equal to the interception gain, and updating the graph neural network model by taking the current sample as an countermeasure sample. Optionally, the first processing module calculates topologically inconsistent energy, including: invoking a node observation time stamp and a node physical position in vehicle track stream data; acquiring historical average flow velocity of a road section and a clock jitter statistic value of equipment; Calculating the time deviation between the actual observation time difference between the node and the neighborhood node and the theoretical physical road network distance running time; And determining the topological inconsistency energy based on the time deviation, the equipment clock jitter statistic value and the output result of the model characteristic mutual exclusion function. Optionally, the first processing module generates a purification subgraph according to the topologically inconsistent energy, including: calling a preset energy constraint threshold; comparing the topologically inconsistent energy with an energy constraint threshold; If the topological inconsistent energy is larger than the energy constraint threshold, judging the nodes as ghost nodes and eliminating the ghost nodes; If the topologically inconsistent energy is less than or equal to the energy constraint threshold, the nodes are preserved to construct a refined subgraph. Optionally, the second processing module calculates an antagonistic characteristic drift rate, comprising: collecting char