An etc-based user behavior portrait generation management system
By identifying and classifying abnormal transaction events in the ETC system, a user behavior profile based on Euclidean distance is constructed, which solves the problem of non-real-time updates of user behavior profiles in existing technologies and achieves more efficient passage management and verification priority determination.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HUNAN EXPRESSWAY NETWORK TOLL MANAGEMENT CO LTD
- Filing Date
- 2026-04-28
- Publication Date
- 2026-05-29
AI Technical Summary
The existing ETC toll collection system lacks joint analysis of the delay, frequency and time distribution of supplementary payments when handling abnormal transaction events. This makes it difficult to accurately distinguish between occasional and recurring abnormal behaviors, and user behavior profiles cannot be updated in real time, affecting the efficiency of verification priority determination and traffic management.
The identification and differentiation module distinguishes between equipment malfunctions and reimbursement malfunctions. The event classification module extracts reimbursement delay, frequency, and time distribution entropy values to generate current risk feature points. The profile building module calculates Euclidean distance to generate user behavior profiles. The processing and scheduling module determines the verification priority and corrects the profiles through the update output module to achieve dynamic updates.
It improves the accuracy and stability of user behavior analysis, enhances the efficiency of access management, reduces the false blocking rate, and ensures the real-time performance and reliability of differentiated access management.
Smart Images
Figure CN122116501A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of ETC data processing technology, specifically a user behavior profile generation and management system based on ETC. Background Technology
[0002] With the continued popularization of electronic toll collection systems on highways, the scale of ETC transaction data, gantry identification data, on-board unit status data, and supplementary payment record data continues to grow, and the requirements for the identification, verification, and subsequent management of abnormal transaction events are becoming increasingly prominent.
[0003] In existing technologies, abnormal transaction events occurring during ETC toll collection are typically handled through methods such as transaction log verification, gantry identification and comparison, or payment status checks. In practical applications, existing solutions often employ static rules, manual verification, or simple counting methods for credit management and processing prioritization to address user arrears, missed payments, or subsequent payments.
[0004] However, the existing technology still has the following limitations: (1) The existing solutions usually only make judgments based on the results of a single payment, the number of payments, or fixed rules. They lack joint analysis of the payment delay, payment frequency, and time distribution patterns, making it difficult to accurately depict the continuity and repetition of the user's payment behavior, which makes it difficult to effectively distinguish between occasional negligence and repetitive abnormal behavior.
[0005] (2) After completing a verification or access management, the existing scheme usually does not dynamically backtrack and correct the historical profile judgment based on the subsequent supplementary payment results and the results of the second access, resulting in the user behavior profile remaining in an outdated state for a long time. It cannot be updated in real time as the user's supplementary payment obligation is fulfilled or new abnormal behavior occurs, which in turn affects the accuracy of subsequent verification priority determination and access management efficiency. Summary of the Invention
[0006] To overcome the shortcomings of the prior art, embodiments of the present invention provide a user behavior profile generation and management system based on ETC, which can effectively solve the problems involved in the prior art.
[0007] The objective of this invention can be achieved through the following technical solution: a user behavior profile generation and management system based on ETC, comprising: an identification and differentiation module, an event classification module, a profile construction module, a processing and scheduling module, and an update and output module.
[0008] The identification and differentiation module is connected to the event classification module, the event classification module is connected to the profile building module, the profile building module is connected to the processing and scheduling module, and the processing and scheduling module is connected to the update and output module.
[0009] The identification and differentiation module identifies abnormal transaction events based on vehicle transaction records, gantry identification results, OBU status information, and supplementary payment records, and classifies them into equipment abnormal events and supplementary payment abnormal events.
[0010] The event classification module extracts the payment delay, payment frequency, and time distribution entropy value of abnormal payment events, and performs normalization processing to generate the current risk feature points.
[0011] The profile building module extracts a set of historical risk feature points, calculates the Euclidean distance between the current risk feature point and each historical risk feature point, identifies historical risk feature points whose Euclidean distance is less than a preset distance threshold as associated feature points, and generates a user behavior profile based on the minimum Euclidean distance and the number of associated feature points.
[0012] The processing and scheduling module determines the verification priority of supplementary payment records based on user behavior profiles and selects the corresponding processing path based on the verification priority.
[0013] Update the output module to correct the user behavior profile based on the subsequent supplementary payment results and the results of the second passage, and output differentiated passage management results.
[0014] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: (1) The present invention constructs the current risk feature points by combining the supplementary payment delay, supplementary payment frequency and time distribution entropy value, which can characterize the abnormal supplementary payment behavior from three dimensions: the delay feature, the repetition feature and the time distribution feature of the supplementary payment behavior, thereby improving the accuracy and stability of user behavior analysis.
[0015] (2) This invention constructs a user behavior profile that reflects the degree of historical similarity and the trend of behavior evolution by associating historical risk feature points, extracting minimum Euclidean distance, and generating behavior offset and behavior reproduction identifiers, so as to determine the verification priority and corresponding processing path, thereby improving the efficiency of passage management and reducing the false blocking rate.
[0016] (3) The present invention corrects and updates the user behavior profile by supplementing payment results and passing results again, which can continuously correct historical judgment deviations and thus improve the real-time performance and reliability of differentiated passage management results. Attached Figure Description
[0017] 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.
[0018] Figure 1 This is a module connection diagram of the present invention;
[0019] Figure 2Here is a flowchart of the method for determining associated feature points according to the present invention;
[0020] Figure 3 This is a flowchart for determining the verification priority of supplementary payment records. Detailed Implementation
[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0022] Reference Figure 1 As shown, the present invention provides a user behavior profile generation and management system based on ETC, including: an identification and differentiation module, an event classification module, a profile construction module, a processing and scheduling module, and an update and output module.
[0023] All modules are uniformly deployed on the access management server, which is connected to the ETC toll database, gantry identification database, OBU status database, and supplementary payment processing database, and completes data reading, data writing, and result output through internal service interfaces.
[0024] The identification and differentiation module is connected to the event classification module, the event classification module is connected to the profile building module, the profile building module is connected to the processing and scheduling module, and the processing and scheduling module is connected to the update and output module.
[0025] The identification and differentiation module identifies abnormal transaction events based on vehicle transaction records, gantry identification results, OBU status information, and supplementary payment records, and classifies them into equipment abnormal events and supplementary payment abnormal events.
[0026] The vehicle transaction records specifically refer to the vehicle identification, entrance station number, entrance time, exit station number, exit time, gantry sequence, amount to be deducted, actual amount deducted, and transaction status corresponding to the entrance records, exit records, and deduction records generated by the ETC toll collection system during the passage of the target vehicle.
[0027] The gantry identification result specifically refers to the vehicle identifier, gantry number, and identification time generated by the gantry equipment when the target vehicle passes through.
[0028] The OBU status information specifically refers to the communication status identifier and deduction status identifier of the OBU corresponding to the target vehicle during the transaction process.
[0029] The supplementary payment record specifically refers to the supplementary payment related transaction identifier, supplementary payment initiation time, supplementary payment completion time, supplementary payment completion status, and supplementary payment amount generated for abnormal transaction events.
[0030] Considering that equipment malfunctions originate from objective technical reasons and are unrelated to users' subjective behavior, they contribute little to the behavioral profile; while payment malfunctions directly reflect the characteristics of users' payment behavior and are the core data source for the profile. In order to reduce noise and improve the targeting of processing, the two need to be processed separately.
[0031] The specific process of identifying abnormal transaction events and distinguishing them into equipment abnormal events and reimbursement abnormal events is as follows: First, obtain the vehicle transaction records, gantry identification results, OBU status information and reimbursement records corresponding to the target vehicle, and associate them according to the vehicle identifier and timestamp.
[0032] Secondly, obtain the communication status identifier and the deduction status identifier from the OBU status information, and perform consistency verification between the vehicle identifier in the gantry identification result and the vehicle identifier in the vehicle transaction record; if the vehicle identifier is inconsistent, or the communication status identifier indicates a communication abnormality, or the deduction status identifier indicates a deduction failure, then the corresponding event is judged as a device identification abnormality event.
[0033] When the event is not determined to be an abnormal event of equipment identification, the gantry identification result is compared with the sequence of gantry that should be passed through the vehicle transaction record item by item. If there is a gantry that should be passed through but there is no corresponding transaction record, or if the entrance information and exit information do not belong to the same passage path, and there is a supplementary payment record at the same time, the corresponding event is determined to be a supplementary payment abnormal event and written into the supplementary payment abnormal event set.
[0034] The required gantry sequence refers to the sequential list of gantry numbers that a target vehicle should pass through when traveling from the entrance station to the exit station, based on the road network topology and the shortest path algorithm. The generation method is as follows: The system pre-stores all gantry nodes in the road network and their connectivity relationships. Taking the entrance station number and the exit station number as input, the Dijkstra algorithm is used to calculate the shortest travel path. The gantry numbers passed through by the path are extracted and sorted according to the travel direction to form the required gantry sequence. When there are multiple shortest paths, the path with the highest historical travel frequency is selected first.
[0035] It should be noted that if an event is not identified as a device identification anomaly or a reimbursement anomaly, it will be classified as a normal transaction event and written into the normal transaction event set. It will not be included in the subsequent reimbursement anomaly event classification, user behavior profile construction, and differentiated access management process.
[0036] It should also be noted that events identified as device identification anomalies are recorded in the anomaly log and do not proceed to subsequent modules.
[0037] The event classification module extracts the payment delay, payment frequency, and time distribution entropy value of abnormal payment events, and performs normalization processing to generate the current risk feature points.
[0038] Reference Figure 2 As shown, the process of generating the current risk feature point is as follows: obtain the set of supplementary payment abnormal events of the vehicle to be processed within the statistical period; for each supplementary payment record in the set of supplementary payment abnormal events, calculate the time difference between the supplementary payment time and the time of the abnormality, and use it as the supplementary payment delay of the corresponding record; in order to reduce the impact of extreme values on the profiling results, take the median of all supplementary payment delays as the supplementary payment delay feature value of the current supplementary payment abnormal event.
[0039] Extract the number of supplementary payments for abnormal events that occurred within the statistical period, and use the cumulative number of supplementary payments as the supplementary payment frequency feature value.
[0040] Extract the completion time of each supplementary payment record from the set of supplementary payment exception events, and form a supplementary payment time interval sequence according to the chronological order.
[0041] When the number of supplementary payment records is greater than or equal to 3, the following steps are performed: First, based on the supplementary payment completion time series, the difference between the completion time of the next supplementary payment and the completion time of the previous supplementary payment is calculated in turn to obtain a series of time intervals and construct an adjacent supplementary payment time interval series.
[0042] Secondly, the embedding dimension of the permutation entropy algorithm is set to 3 and the time delay to 1. Starting from the first time interval, three consecutive time intervals are taken as a reconstruction vector.
[0043] Then, for each of the three time interval values in the reconstructed vector, sort them in ascending order of value, and record the position order of each element in the original reconstructed vector before sorting. This position order is used as the symbol arrangement pattern of the corresponding reconstructed vector. If there are the same values in the reconstructed vector, the arrangement order is determined according to the order in which the same values appear in the original reconstructed vector.
[0044] Since the embedding dimension is 3, there are a total of 6 possible symbol permutations. The number of times each symbol permutation appears in all reconstructed vectors is counted, and the probability distribution of each permutation is obtained by dividing the number of times each permutation appears by the total number of reconstructed vectors.
[0045] Finally, multiply the probability distribution of each pattern by the natural logarithm of that probability, sum all the products, and take the negative value. This yields the time distribution entropy.
[0046] The feature values of payment delay, payment frequency, and time distribution entropy are all subjected to extreme value normalization, and the results are mapped to the interval [0, 1]. Specifically, for any feature value to be normalized, the minimum and maximum values of the feature are determined among all values of the feature within the same statistical batch. Then, the minimum value is subtracted from the feature value, and the difference is divided by the difference between the maximum and minimum values. The result is the normalized feature value. If the maximum and minimum values of the feature within the batch are equal, the normalization result of the feature is directly recorded as 0.
[0047] The normalized feature values of payment delay, payment frequency, and time distribution entropy are combined in a fixed order to form a three-dimensional vector, which is the generated current risk feature point.
[0048] It should be noted that when the number of supplementary payment records is less than 3, the time distribution entropy value is marked as a special value, and in the subsequent normalization and Euclidean distance calculation, feature points containing this special value are directly judged as dissimilar and are not included in the construction of the current risk feature points.
[0049] This invention constructs current risk feature points by combining the delay, frequency, and time distribution entropy of supplementary payment, which can characterize abnormal supplementary payment behavior from three dimensions: delay characteristics, repetition characteristics, and time distribution characteristics, thereby improving the accuracy and stability of user behavior analysis.
[0050] The profile building module extracts a set of historical risk feature points, calculates the Euclidean distance between the current risk feature point and each historical risk feature point, identifies historical risk feature points whose Euclidean distance is less than a preset distance threshold as associated feature points, and generates a user behavior profile based on the minimum Euclidean distance and the number of associated feature points.
[0051] The specific process of determining historical risk feature points whose Euclidean distance is less than a preset distance threshold as associated feature points is as follows: read the historical risk feature points corresponding to all marked supplementary payment abnormal events of the corresponding vehicle within a preset historical time period from the database to form a set of historical risk feature points. Each historical risk feature point is represented by the same three-dimensional vector as the current risk feature point.
[0052] Calculate the Euclidean distance between the current risk feature point and each historical risk feature point in the historical risk feature point set.
[0053] Historical risk feature points with an Euclidean distance less than a preset distance threshold are selected as associated feature points, and the generation time and Euclidean distance of each associated feature point are recorded.
[0054] It should be noted that the Euclidean distance is used to characterize the degree of similarity between current abnormal supplementary payment behavior and historical abnormal supplementary payment behavior. The smaller the Euclidean distance, the closer the corresponding behaviors are in terms of supplementary payment delay, supplementary payment frequency and time distribution.
[0055] The selection of the preset historical duration should ensure that the set of historical risk feature points can contain a sufficient number of supplementary payment anomalies to support the effective coding of behavior reproduction identifiers. For example, it is set to 90 days.
[0056] The preset distance threshold is calibrated based on the actual data distribution. Specifically, by statistically analyzing the Euclidean distance distribution between the feature points of normal and abnormal payment behaviors of the same vehicle in historical data, the distance value corresponding to the valley between the two types of distance distributions is taken as the preset distance threshold.
[0057] It should also be noted that when the historical risk feature point set is empty, the Euclidean distance calculation is skipped, the current risk feature point is used as the initial profile, the number of associated feature points is 0, the minimum Euclidean distance is set to 1.0, and the subsequent processing scheduling module selects the tracking and verification path based on the condition that the number of associated feature points is 0.
[0058] Sort the Euclidean distances corresponding to each associated feature point from smallest to largest, count the number of associated feature points, and take the minimum value as the minimum Euclidean distance.
[0059] The historical risk feature points corresponding to the minimum Euclidean distance are used as reference feature points.
[0060] Calculate the differences between the current risk feature point and the reference feature point in terms of the feature values of payment delay, payment frequency, and time distribution entropy. Then, encode the location according to positive change, negative change, and no change. Form a behavioral offset identifier based on the order of payment delay, payment frequency, and time distribution entropy. The specific process is as follows: Let the normalized results of the current risk feature point in terms of payment delay, payment frequency, and time distribution entropy be respectively... , , Let the normalized results of the reference feature point on the feature values of payment delay, payment frequency, and time distribution entropy be as follows: , , .
[0061] Calculate the difference between the current risk feature point and the reference feature point on each feature dimension, where the difference in the due payment delay is denoted as... The difference in the frequency of supplementary payments is recorded as The difference in entropy values over time is denoted as .
[0062] For the difference corresponding to any feature dimension ,when When the difference exceeds a preset threshold, the feature dimension is judged to be a positive change and encoded as 1; when... When the difference is less than a preset threshold, the feature dimension is determined to be a reverse change and encoded as N1; when... When the negative value is greater than or equal to the preset difference judgment threshold but less than or equal to the preset difference judgment threshold, the feature dimension is determined to be unchanged and encoded as 0. .
[0063] Following a fixed order of the supplementary payment delay feature value, the supplementary payment frequency feature value, and the time distribution entropy value, the encoding results corresponding to each feature dimension are read sequentially and combined to form a three-dimensional behavior offset identifier.
[0064] The associated feature points are sorted according to their generation time, and the time interval sequence between adjacent associated feature points is extracted. The number of associated feature points and the time interval sequence are sequentially encoded as behavior reproduction identifiers. The specific process is as follows: First, the generation time corresponding to each associated feature point is obtained and sorted from earliest to latest according to the generation time to form an associated feature point sequence.
[0065] Secondly, the generation time difference between two adjacent associated feature points in the associated feature point sequence is extracted to form a time interval sequence.
[0066] Subsequently, the number of associated feature points is encoded. If the number of associated feature points is equal to 1, the encoding is recorded as follows: If the number of associated feature points is greater than or equal to 2 and less than or equal to 3, then the quantity encoding is denoted as... If the number of associated feature points is greater than or equal to 4, the quantity code is denoted as... .
[0067] The time intervals in the time interval sequence are encoded according to their order relationship. When the time interval sequence contains only one time interval, the interval code corresponding to that time interval is recorded as... When the time interval sequence contains at least two time intervals, compare the size relationship between two adjacent time intervals in turn.
[0068] If the subsequent time interval is greater than the previous time interval, the interval code at the corresponding position is denoted as... If the subsequent time interval is shorter than the previous time interval, the interval code at the corresponding position is denoted as... If the subsequent time interval is equal to the previous time interval, then the interval code at the corresponding position is denoted as... Arrange the interval codes corresponding to each position in chronological order of generation to form an interval code string; when the time interval sequence is empty, the interval code string is denoted as... .
[0069] Finally, the quantity code and the interval code string are concatenated in a fixed order to generate a behavior reproduction identifier. For example, when the number of associated feature points is 4, and the comparison results of adjacent time intervals in the time interval sequence are increasing, decreasing, and remaining the same, a behavior reproduction identifier is generated. When the number of associated feature points is 1, a behavior reproduction identifier is generated. .
[0070] The minimum Euclidean distance, behavior offset identifier, and behavior reproduction identifier are combined and encoded to generate a user behavior profile. The specific process is as follows: Obtain the minimum Euclidean distance, denoted as... The behavior offset identifier is denoted as The behavior reproduction identifier is recorded as .
[0071] Distance encoding Behavior offset identifier and behavior reproduction identifiers The distance encoding field, behavior offset field, and behavior reproduction field are concatenated in that order and combined for encoding, denoted as... Generate user behavior profiles, namely: .
[0072] For example, when It is 0.23. for , for At that time, the generated user profile is .
[0073] It should be noted that the behavior offset identifier represents the direction of change of the current risk feature point relative to historical similar behaviors; the behavior recurrence identifier represents the recurrence of similar abnormal behaviors over time. Through combined coding, user behavior profiles can reflect both the similarity characteristics of a single abnormal behavior and the persistence and recurrence of such behaviors.
[0074] It should also be noted that the preset difference judgment threshold is adjusted according to the actual needs of the system's sensitivity to behavioral changes. For example, the preset difference judgment threshold is set to 0.05 to improve sensitivity, and the implementer can adjust the specific value adaptively according to the actual situation.
[0075] The processing and scheduling module determines the verification priority of supplementary payment records based on user behavior profiles and selects the corresponding processing path based on the verification priority.
[0076] Reference Figure 3As shown, the process of determining the verification priority of supplementary payment records based on user behavior profiles is as follows: obtain the user behavior profiles corresponding to each supplementary payment record in the current verification batch, and parse them to obtain the minimum Euclidean distance, behavior offset identifier, and behavior reproduction identifier.
[0077] The supplementary payment records are sorted in ascending order of minimum Euclidean distance, and the supplementary payment record with the highest position in the first sort is given priority in subsequent priority determination.
[0078] When the minimum Euclidean distance is the same, the supplementary payment records with the same minimum Euclidean distance are sorted in the second order according to the number of associated feature points from largest to smallest, and the supplementary payment record with the highest position in the second sort takes priority over the other supplementary payment records.
[0079] When the number of associated feature points is the same, obtain the original time interval sequence corresponding to the behavior reproduction identifier.
[0080] If the time interval sequence is not empty, the arithmetic mean of all time intervals in the sequence is calculated as the average time interval, and then sorted in the third order according to the average time interval from smallest to largest.
[0081] If the time interval sequence is empty, the average time interval of the supplementary payment record will be placed at the end of the third sorting, and the supplementary payment record at the beginning of the third sorting will take priority over the other supplementary payment records.
[0082] When the average time intervals are the same, the corresponding position codes are compared in the order of the supplementary payment delay feature value, supplementary payment frequency feature value and time distribution entropy value in the behavior offset identifier. The order of arrangement is determined according to the positive change, no change and reverse change, and the supplementary payment record with the highest comparison result is determined as the priority supplementary payment record.
[0083] Finally, the ranking of all supplementary payment records is determined according to the above comparison order, and the supplementary payment record with the highest ranking position is determined as the highest verification priority.
[0084] It should be noted that the first, second, and third sorting are hierarchical sortings triggered step by step. The results of the previous level sorting take precedence over the results of the next level sorting, and the next level sorting is triggered only when the previous level sorting cannot distinguish the order.
[0085] It should also be noted that the number of associated feature points can be obtained from the original count value corresponding to the quantity code in the behavior recurrence identifier. It reflects the number of times the abnormal behavior is repeated and is the primary indicator for risk assessment. Therefore, its priority in the hierarchical ranking is higher than the time interval reflecting the recurrence frequency. The time interval sequence can be obtained from the original time interval data corresponding to the interval code string in the behavior recurrence identifier.
[0086] In a preferred embodiment of the present invention, the process of selecting the corresponding processing path according to the verification priority is as follows: First, read the sorting position of the target supplementary payment record, the number of associated feature points, and the user behavior profile generation result.
[0087] When the target supplementary payment record is ranked first, the processing path of the target supplementary payment record is determined as the key verification path, and the gantry identification result review, OBU status backtracking and passage path restoration are performed.
[0088] Among them, the verification of gantry identification results involves comparing the vehicle identification, gantry number, and identification time in the gantry identification results with the vehicle identification, gantry sequence, and transaction time in the vehicle transaction records item by item.
[0089] OBU status tracing involves retrieving the OBU status log corresponding to the time period of the anomaly and reading the communication status identifier and the deduction status identifier. The passage path restoration involves determining the actual passage path of the target vehicle based on the entrance station number, exit station number, and the sequence of identified gantry numbers.
[0090] When the target supplementary payment record is not ranked first and a user behavior profile has been generated, the processing path for the target supplementary payment record is determined to be the regular verification path. The process of verifying the consistency between the gantry recognition result and the vehicle transaction record, as well as the validity verification of the supplementary payment completion status and the supplementary payment completion time, is performed. The validity verification process verifies whether the supplementary payment completion status is completed and whether the supplementary payment completion time is later than the time when the anomaly occurred.
[0091] When the associated feature point of the target supplementary payment record is not determined, the processing path of the target supplementary payment record is determined to be the tracking and verification path, and field integrity verification processing and subsequent passage tracking processing are performed. The field integrity verification processing verifies whether the vehicle identifier, supplementary payment associated transaction identifier, supplementary payment completion status and supplementary payment completion time are complete, and writes the corresponding supplementary payment record into the subsequent passage tracking queue; the determined processing path is written into the processing path field of the target supplementary payment record.
[0092] The processing path identifier obtained from the above retrieval will be used as the processing path for the corresponding supplementary payment record.
[0093] This invention constructs a user behavior profile that reflects the degree of historical similarity and the trend of behavioral evolution by associating historical risk feature points, extracting minimum Euclidean distance, and generating behavioral offset and behavioral reproduction identifiers. This profile is then used to determine verification priorities and corresponding processing paths, thereby improving traffic management efficiency and reducing false blocking rates.
[0094] Update the output module to correct the user behavior profile based on the subsequent supplementary payment results and the results of the second passage, and output differentiated passage management results.
[0095] Given that user payment behavior is not static, as time progresses and activities continue, users may fulfill their payment obligations, thus reducing their risk level, or new payment anomalies may occur, accumulating higher risks. If user behavior profiles are generated solely based on a set of payment anomalies at a specific historical moment and are not updated, the profile results will gradually deviate from the user's actual behavior over time, leading to inaccuracies in subsequent verification priority determination and processing path selection.
[0096] The process of correcting the user behavior profile based on the subsequent payment results and the re-pass results is as follows: obtain the subsequent payment results corresponding to the target payment record and the re-pass results of the target vehicle after the current verification is completed. The subsequent payment results include the payment completion status and payment completion time, and the re-pass results include the transaction records corresponding to the re-pass and the determination results of whether there are new payment anomalies.
[0097] When the subsequent supplementary payment result indicates that the supplementary payment has been completed, the completion time of the supplementary payment will be written into the corresponding supplementary payment exception event data.
[0098] When the result of the passage indicates the existence of a new supplementary payment exception event, the new supplementary payment exception event is written into the supplementary payment exception event set.
[0099] Based on the updated set of abnormal events for supplementary payments, the feature values of supplementary payment delay, frequency, and time distribution entropy are recalculated to generate updated current risk feature points.
[0100] Based on the updated current risk feature points, the associated feature points are redefined, and a new user behavior profile is generated.
[0101] Based on the corrected user behavior profile, the verification priority and processing path of the corresponding supplementary payment records are re-determined, and the updated differentiated access management results are output.
[0102] It should be noted that when a subsequent payment result for a certain supplementary payment record is obtained, or when the vehicle generates a new passage result, a correction process for the user behavior profile of that vehicle is immediately triggered. After the correction process is completed, the old profile of the vehicle is replaced with the newly generated user behavior profile, and the current risk feature points used in this correction are added to the historical risk feature point set for subsequent similarity matching. Each correction only applies to supplementary payment records that are not yet closed; records that have been paid and have no new anomalies are not corrected again.
[0103] It should also be noted that the differentiated access management results can be release review, priority verification, continuous tracking, manual intervention review, or other management results corresponding to the risk level. Through the feedback update mechanism, user behavior profiles can be prevented from remaining in a historical state for extended periods, thereby improving the system's responsiveness to new behavioral changes.
[0104] This invention corrects and updates user behavior profiles by using subsequent supplementary payment results and re-pass results, which can continuously correct historical judgment biases and thus improve the real-time performance and reliability of differentiated access management results.
[0105] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.
Claims
1. A user behavior profile generation and management system based on ETC, characterized in that, include: The identification and differentiation module identifies abnormal transaction events based on vehicle transaction records, gantry identification results, OBU status information, and payment records, and classifies them into equipment abnormal events and payment abnormal events. The event classification module extracts the payment delay, payment frequency, and time distribution entropy value of abnormal payment events, and performs normalization processing to generate the current risk feature points; The profile building module extracts the set of historical risk feature points, calculates the Euclidean distance between the current risk feature point and each historical risk feature point, identifies historical risk feature points whose Euclidean distance is less than a preset distance threshold as associated feature points, and generates user behavior profiles based on the minimum Euclidean distance and the number of associated feature points. The processing and scheduling module determines the verification priority of supplementary payment records based on user behavior profiles and selects the corresponding processing path based on the verification priority. Update the output module to correct the user behavior profile based on the subsequent supplementary payment results and the results of the second passage, and output differentiated passage management results.
2. The user behavior profile generation and management system based on ETC according to claim 1, characterized in that, The process of identifying abnormal transaction events and distinguishing them into equipment abnormal events and reimbursement abnormal events is as follows: Obtain the communication status identifier and the billing status identifier from the OBU status information, and perform consistency verification between the vehicle identifier in the gantry identification result and the vehicle identifier in the vehicle transaction record. If the vehicle identifier is inconsistent, or the communication status identifier is abnormal and the billing status identifier is failed, it is determined to be an abnormal device identification event. When no abnormal event is identified by the equipment, the gantry identification result is compared item by item with the gantry sequence corresponding to the vehicle transaction record; If there is no corresponding transaction record for a gantry, or if the entrance information and exit information do not belong to the same passage path, and there is a supplementary payment record, it is determined to be a supplementary payment abnormal event.
3. The user behavior profile generation and management system based on ETC according to claim 1, characterized in that, The process of generating the current risk feature point is as follows: Obtain the set of abnormal payment events for the vehicle to be processed within a preset statistical period, calculate the time difference between the payment time and the time of the abnormality for each payment record in the set, and use it as the corresponding payment delay. Take the median of all payment delays as the payment delay feature value of the current abnormal payment event. Extract the number of supplementary payments due to abnormal events within the statistical period, and use the cumulative number of supplementary payments as the supplementary payment frequency feature value; Arrange the completion times of each supplementary payment in chronological order to form a supplementary payment time sequence. Calculate the time difference between adjacent completion times to obtain a supplementary payment time interval sequence. Reconstruct the supplementary payment time interval sequence using the permutation entropy algorithm, statistically analyze the distribution probability of each symbol permutation pattern, and calculate the time distribution entropy value based on the dispersion of each symbol permutation pattern. The feature values of the payment delay, the frequency of payment, and the time distribution entropy are normalized to extreme values and mapped to the interval [0, 1] to form a three-dimensional vector as the current risk feature point.
4. The user behavior profile generation and management system based on ETC according to claim 3, characterized in that, The specific process for determining historical risk feature points whose Euclidean distance is less than a preset distance threshold as associated feature points is as follows: The database is used to retrieve all historical risk feature points corresponding to the marked supplementary payment anomaly events of the corresponding vehicle within a preset historical time period, forming a set of historical risk feature points. Each historical risk feature point is represented by the same three-dimensional vector as the current risk feature point. Calculate the Euclidean distance between the current risk feature point and each historical risk feature point in the historical risk feature point set; Historical risk feature points with an Euclidean distance less than a preset distance threshold are selected as associated feature points, and the generation time and Euclidean distance of each associated feature point are recorded.
5. The user behavior profile generation and management system based on ETC according to claim 4, characterized in that, The specific process for generating a user behavior profile based on the minimum Euclidean distance and the number of associated feature points is as follows: Sort the Euclidean distances corresponding to each associated feature point from smallest to largest, count the number of associated feature points, and take the minimum value as the minimum Euclidean distance; Use the historical risk feature points corresponding to the minimum Euclidean distance as reference feature points; Calculate the differences between the current risk feature point and the reference feature point in terms of the feature value of payment delay, the feature value of payment frequency, and the feature value of time distribution entropy. Then, encode the location according to positive change, negative change, and no change. Form a behavior offset identifier in the order of payment delay feature value, payment frequency feature value, and time distribution entropy value.
6. The user behavior profile generation and management system based on ETC according to claim 5, characterized in that, The specific process of generating user behavior profiles based on minimum Euclidean distance and the number of associated feature points also includes: The associated feature points are sorted according to their generation time, the time interval sequence between adjacent associated feature points is extracted, and the number of associated feature points and the time interval sequence are sequentially encoded as behavior reproduction identifiers. The minimum Euclidean distance, behavior offset identifier, and behavior reproduction identifier are combined and encoded to generate a user behavior profile.
7. The user behavior profile generation and management system based on ETC according to claim 6, characterized in that, The process of determining the verification priority of supplementary payment records based on user behavior profiles is as follows: Obtain the user behavior profile corresponding to each supplementary payment record in the current verification batch, and parse it to obtain the minimum Euclidean distance, behavior offset identifier and behavior recurrence identifier; The supplementary payment records are sorted first according to the minimum Euclidean distance from smallest to largest. When the minimum Euclidean distances are the same, a second sort is performed based on the number of associated feature points from largest to smallest. When the number of associated feature points is the same, obtain the original time interval sequence corresponding to the behavior reproduction identifier; If the time interval sequence is not empty, the arithmetic mean of all time intervals in the sequence is calculated as the average time interval, and then sorted in the third order according to the average time interval from smallest to largest. If the time interval sequence is empty, the average time interval of the supplementary payment record will be ranked last in the third sorting. When the average time intervals are the same, the corresponding position codes are compared in the order of payment delay, payment frequency and time distribution entropy value in the behavior offset identifier, and the order is determined according to positive change, no change and reverse change. A verification sequence is formed based on the sorting results, and the verification priority is determined according to the sorting position of each supplementary payment record in the verification sequence. The earlier the sorting position, the higher the verification priority.
8. The user behavior profile generation and management system based on ETC according to claim 7, characterized in that, The process of selecting the corresponding processing path based on verification priority is as follows: Read the sort position of the target supplementary payment record in the verification sequence; When the target supplementary payment record is at the top of the verification sequence, the processing path of the target supplementary payment record is determined as the key verification path, and the gantry identification result review, OBU status backtracking and passage path restoration processing are called. When the target supplementary payment record is not at the first position in the verification sequence and a user behavior profile has been generated, the processing path for the target supplementary payment record is determined to be the regular verification path. The consistency of the gantry recognition result corresponding to the supplementary payment record with the vehicle transaction record is verified, and the validity of the supplementary payment completion status and the supplementary payment completion time is verified. When the associated feature point of the target supplementary payment record is not determined, the processing path of the target supplementary payment record is determined to be the tracking and verification path, and the field integrity verification sub-process is called to write the corresponding supplementary payment record into the subsequent passage tracking queue. The retrieved processing path identifier will be used as the processing path for the corresponding supplementary payment record.
9. The user behavior profile generation and management system based on ETC according to claim 1, characterized in that, The process of correcting the user behavior profile based on the subsequent supplementary payment results and the results of re-access is as follows: Obtain the subsequent payment results corresponding to the target payment record and the re-pass results of the target vehicle after the current verification is completed. The subsequent payment results include the payment completion status and payment completion time, and the re-pass results include the transaction records corresponding to the re-pass and the judgment results of whether there are any new payment anomalies. When the subsequent supplementary payment result indicates that the supplementary payment has been completed, the supplementary payment completion time will be written into the corresponding supplementary payment exception event data; When the result of the passage indicates the existence of a new supplementary payment exception event, the new supplementary payment exception event will be written into the supplementary payment exception event set. Based on the updated set of abnormal events for supplementary payments, the feature values of supplementary payment delay, frequency, and time distribution entropy are recalculated to generate updated current risk feature points.
10. A user behavior profile generation and management system based on ETC according to claim 9, characterized in that, The process of correcting the user behavior profile based on the subsequent supplementary payment results and the results of re-access also includes: Based on the updated current risk feature points, the associated feature points are re-determined, and a new user behavior profile is generated. Based on the corrected user behavior profile, the verification priority and processing path of the corresponding supplementary payment records are re-determined, and the updated differentiated access management results are output.