A bus illegal use behavior analysis system, device, medium and product

Abstract

The application discloses a kind of bus illegal use vehicle behavior analysis system, equipment, medium and product, system includes perception layer, reasoning layer and decision layer, perception layer includes data perception and cleaning intelligent agent, reasoning layer includes knowledge retrieval intelligent agent and trajectory semantic intelligent agent, decision layer includes comprehensive research and judgment intelligent agent, data perception and cleaning intelligent agent are used to denoising to trajectory data;Knowledge retrieval intelligent agent is used to obtain the information associated with the vehicle to be analyzed;Trajectory semantic intelligent agent is used to obtain the interest point category label and regional attribute of the stay point;Comprehensive research and judgment intelligent agent is used to convert the application cause, interest point category label and regional attribute into feature vector, and perform verification, obtain illegal conclusion and evidence chain, the application is analyzed to driving trajectory by multi-agent cooperation, the barrier between application cause and actual trajectory is broken through using the semantic understanding ability of large model, and the precise identification of bus illegal behavior is realized.

Description

Technical Field

[0001] This invention specifically relates to a system, equipment, medium, and product for analyzing illegal use of public vehicles. Background Technology

[0002] The management of official vehicles is of great significance for standardizing official vehicle use, preventing the misuse of official vehicles, and reducing administrative costs. With the deepening of information technology construction, government agencies at all levels have generally installed vehicle global positioning systems (GPS), accumulating massive amounts of vehicle trajectory data. Traditional algorithms cannot understand the textual semantics in vehicle use application forms and can only perform simple geofencing judgments on GPS trajectory data. They cannot achieve logical consistency verification between textual semantics and spatiotemporal trajectories, and at the same time, it is difficult to effectively distinguish between normal operations and violations. Summary of the Invention

[0003] This invention provides a system, device, medium, and product for analyzing illegal use of public vehicles, in order to solve the problem that traditional algorithms are inaccurate in judging illegal use of public vehicles.

[0004] In a first aspect, the present invention provides a public vehicle violation analysis system, the system comprising a perception layer, a reasoning layer and a decision layer, the perception layer comprising a data perception and cleaning agent, the reasoning layer comprising a knowledge retrieval agent and a trajectory semanticization agent, and the decision layer comprising a comprehensive judgment agent.

[0005] The data sensing and cleaning agent is used to acquire the trajectory data of the vehicle to be analyzed, and to use the trajectory denoising algorithm to perform denoising calculation on the trajectory data of the vehicle to be analyzed to obtain the denoised trajectory data. The knowledge retrieval intelligent agent is used to obtain vehicle application forms, whitelists, holiday databases, and interest point knowledge bases associated with the vehicle to be analyzed. The trajectory semantic intelligent agent is used to extract stop points from the denoised trajectory data, obtain the location coordinates and stop duration data of the stop points, and perform reverse geocoding on the location coordinates of the stop points to obtain the point of interest category label data and regional attribute data of the stop points; The intelligent agent comprehensively analyzes and transforms the application reason in the car use application form into the expected behavior feature vector, and transforms the interest point category label data and regional attribute data into the actual behavior feature vector. It performs cross-modal consistency verification on the expected behavior feature vector and the actual behavior feature vector to obtain the consistency verification result. Based on the location coordinates of the stop point, the stay duration, the whitelist database and holiday information, it performs risk classification and time dimension analysis to obtain the violation conclusion and evidence chain.

[0006] By utilizing a data perception and cleaning agent to perform refined noise reduction on the original trajectory, noise data generated by legitimate operations such as temporary ignition and parking in the park is eliminated, solving the technical problem of high false alarm rate. Through a trajectory semanticization agent, the trajectory coordinates are reverse geocoded and semantically mapped to points of interest, transforming the coordinates into computable semantic labels. The comprehensive judgment agent transforms the application reason and the actual trajectory into semantic feature vectors respectively, calculates semantic similarity, and combines whitelists and holiday times for comprehensive judgment, solving the technical problem of traditional solutions lacking reasoning ability, thereby accurately identifying violations and reducing false alarm rate.

[0007] In one optional implementation, the trajectory data includes a trajectory sequence, and the denoising calculation includes removing temporary ignition trajectory data. The data perception and cleaning agent acquires the trajectory data of the vehicle to be analyzed and uses a trajectory denoising algorithm to perform denoising calculations on the trajectory data of the vehicle to be analyzed, including: The data sensing and cleaning agent acquires the trajectory sequence of the vehicle to be analyzed within a time window, and calculates the displacement entropy and velocity variance of each trajectory point based on the trajectory sequence; If the displacement entropy is less than the first preset threshold, and the speed variance is less than the second preset threshold, and the vehicle has an accessory power gear signal jump, the trajectory point is determined to be either the original ignition or idling state. Remove trajectory points from the trajectory data of the vehicle being analyzed when it is in a fixed ignition or idling state.

[0008] By calculating the displacement entropy and velocity variance of trajectory points within a time window, and combining this with the power gear signal jump data, trajectory points in stationary ignition or idling states can be identified and eliminated. This solves the technical problem in traditional solutions that misjudge legitimate operations such as vehicle waiting in place or temporary ignition as illegal stops, thereby reducing the false alarm rate.

[0009] In one optional implementation, the trajectory data includes a set of trajectory points, and the denoising calculation includes removing the park parking trajectory data. The data perception and cleaning agent acquires the trajectory data of the vehicle to be analyzed and uses a trajectory denoising algorithm to perform denoising calculations on the trajectory data of the vehicle to be analyzed, including: The data sensing and cleaning agent acquires the set of trajectory points of the vehicle to be analyzed within a time window and the sequence of polygon vertices in the park. If all trajectory points in the trajectory point set are located within the area formed by the sequence of polygon vertices in the park, and the displacement distance corresponding to the trajectory point set is less than the maximum span threshold of the area formed by the sequence of polygon vertices in the park, then the trajectory point set is determined to be internal scheduling trajectory data of the park. The internal scheduling trajectory data of the park is removed from the trajectory data of the vehicles to be analyzed.

[0010] By acquiring the set of trajectory points of the vehicle to be analyzed within the time window and the sequence of vertices of the park polygon, and determining whether all trajectory points are located inside the park polygon and the displacement distance is less than the maximum span threshold, the scheduling trajectory data inside the park can be identified and eliminated. This solves the technical problem of misjudging legitimate scheduling behaviors such as normal vehicle movement and entering and exiting the garage within the unit's park as illegal vehicle use, and further reduces the false alarm rate.

[0011] In one optional implementation, the trajectory semantic agent performs inverse geocoding on the location coordinates of the stop point to obtain point of interest category label data and regional attribute data for the stop point, including: The trajectory semantic intelligent agent performs inverse geocoding calculations on the location coordinates of the stop point to obtain the address information of the stop point and the information of surrounding points of interest. Based on the surrounding points of interest information, relevant points of interest are identified, and tags are extracted from these points of interest to obtain the point of interest category tag data for the stop location; The location of the stop point is determined based on information about surrounding points of interest.

[0012] By performing reverse geocoding calculations on the location coordinates of the stop points, address information and surrounding points of interest information are obtained. Based on the surrounding points of interest information, relevant points of interest are identified for tag extraction and regional attributes are determined. This transforms abstract latitude and longitude coordinates into point of interest category tags and regional attribute data with clear semantic meanings, providing a semantic data foundation for subsequent comprehensive judgment of the intelligent agent to perform consistency verification and risk classification.

[0013] In one optional implementation, the intelligent agent comprehensively analyzes and transforms the reason for the vehicle use request into a predictable behavioral feature vector, including: A comprehensive analysis of the text explaining the reason for the vehicle use request obtained by the intelligent agent; Key entities are extracted from the application text to obtain destination data, purpose data, and business attribute data; Perform semantic completion and text reconstruction on destination data, purpose data, and business attribute data to generate text describing expected behavior. Input the text describing the expected behavior into a pre-defined text embedding model to obtain the feature vector of the expected behavior.

[0014] By extracting key entities from the application text in the vehicle application form, destination data, purpose data, and business attribute data are obtained. After semantic completion and text reconstruction, the results are input into a text embedding model to generate expected behavior feature vectors. This transforms the unstructured natural language application into a computable numerical vector form, enabling the system to quantitatively determine whether the actual driving behavior matches the application.

[0015] In one optional implementation, the intelligent agent comprehensively analyzes and transforms the interest point category labels and region attributes into actual behavioral feature vectors, including: A comprehensive analysis is conducted to determine the category labels and region attributes of the intelligent agent's points of interest. Text concatenation and semantic expansion are performed on the interest point category labels and region attributes to obtain the actual behavior description text; Input the actual behavior description text into the preset text embedding model to obtain the actual behavior feature vector.

[0016] By concatenating and semantically expanding the text of interest point category labels and regional attributes, the actual behavior description text is generated and then input into a preset text embedding model to obtain the actual behavior feature vector. This process transforms unstructured location category labels and regional attribute information into a structured, computable numerical vector form, enabling the system to compare the similarity between the actual behavior feature vector and the expected behavior feature vector in the same semantic space.

[0017] In one optional implementation, the violation conclusion includes unauthorized stay in sensitive locations and unauthorized parking on holidays. The intelligent agent performs risk classification and time-dimensional analysis based on the location coordinates of the stay point, the duration of stay, the whitelist database, and holiday information to obtain the violation conclusion, including: The system comprehensively analyzes and assesses the data obtained by the intelligent agent, including the range of the unit's sealing points, the location coordinates of the points of stay, the duration of stay, the whitelist database, and holiday information. The current point of interest category label is determined based on the location coordinates and duration of the stop; If the current point of interest category tag belongs to the preset risk location set and the stay duration exceeds the preset duration threshold, it is determined to be a violation of the rules for staying in a sensitive location; If the location coordinates of the stop point are within the whitelist range in the whitelist database, then the stop is deemed compliant. If the current time is a holiday and the location coordinates of the parking spot are not within the unit's sealed-off point data range, then it is determined to be a non-public parking on a holiday.

[0018] By acquiring data on the scope of unit sealing points, the location coordinates of stopping points, the duration of stay, the whitelist database, and holiday information, and based on the relationship between the current point of interest category tag and the preset risk location set, and the relationship between the duration of stay and the preset duration threshold, the system determines illegal staying in sensitive locations. Based on the location coordinates of the stopping point and the scope of the whitelist database, it determines compliant staying. Furthermore, based on whether it is a holiday or not, and the relationship between the location coordinates of the stopping point and the unit sealing point scope data, it determines non-public parking on holidays. Ultimately, this system can identify various violation scenarios such as prolonged staying in sensitive locations and illegal parking on holidays, avoiding false alarms in traditional solutions and improving the accuracy of the system.

[0019] In a second aspect, the present invention provides an electronic device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform a public vehicle violation analysis system according to the first aspect or any corresponding embodiment described above.

[0020] Thirdly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute a public vehicle violation analysis system according to the first aspect or any corresponding embodiment described above.

[0021] Fourthly, the present invention provides a computer program product, including computer instructions, which are used to cause a computer to execute a public vehicle violation analysis system according to the first aspect or any corresponding embodiment described above. Attached Figure Description

[0022] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0023] Figure 1 This is a system schematic diagram of a public vehicle violation analysis system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0025] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0026] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0027] This invention provides a system for analyzing illegal use of public vehicles, such as... Figure 1 As shown, the system includes a perception layer 101, an inference layer 102, and a decision layer 103. The perception layer 101 includes a data perception and cleaning agent 1011; the inference layer 102 includes a knowledge retrieval agent 1021 and a trajectory semanticization agent 1022; and the decision layer 103 includes a comprehensive judgment agent 1031. The data perception and cleaning agent 1011 acquires the trajectory data of the vehicle to be analyzed and uses a trajectory denoising algorithm to denoise the trajectory data, obtaining denoised trajectory data. The knowledge retrieval agent 1021 acquires vehicle application forms, whitelists, holiday databases, and point-of-interest knowledge bases associated with the vehicle to be analyzed. The trajectory semanticization agent 1022 is used to... The denoised trajectory data is used to extract stop points, obtaining their location coordinates and duration. Reverse geocoding is then performed on the stop point coordinates to obtain point-of-interest (POI) category labels and regional attribute data. The intelligent agent 1031 is then used to transform the application reason in the vehicle use application form into a predicted behavioral feature vector, and the POI category label data and regional attribute data into an actual behavioral feature vector. Cross-modal consistency checks are performed on the predicted and actual behavioral feature vectors to obtain consistency check results. Based on the stop point location coordinates, duration, whitelist database, and holiday information, risk grading and time-dimensional analysis are performed to obtain violation conclusions and evidence chains.

[0028] This invention provides a public vehicle violation analysis system. By combining trajectory denoising, semantic understanding and logical reasoning, it achieves semantic understanding and then judges violations from the original trajectory data, thereby making the judgment of violation behavior more accurate.

[0029] According to an embodiment of the present invention, a system for analyzing illegal use of public vehicles is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0030] This embodiment provides a system for analyzing illegal use of public vehicles. Figure 1This is a flowchart of a public bus violation analysis system according to an embodiment of the present invention, such as... Figure 1 As shown, the system includes a perception layer, an inference layer, and a decision layer. The perception layer includes a data perception and cleaning agent, the inference layer includes a knowledge retrieval agent and a trajectory semanticization agent, and the decision layer includes a comprehensive judgment agent.

[0031] The data sensing and cleaning agent 1011 is used to acquire the trajectory data of the vehicle to be analyzed and to use a trajectory denoising algorithm to perform denoising calculations on the trajectory data of the vehicle to be analyzed, so as to obtain the denoised trajectory data.

[0032] The vehicle to be analyzed refers to an official vehicle for which illegal use of the vehicle needs to be analyzed. Trajectory data refers to spatiotemporal sequence data continuously collected and transmitted back by satellite positioning equipment installed on the vehicle to be analyzed at preset time intervals, which may include the position coordinates, timestamps, and speeds at each sampling time. Trajectory denoising algorithm refers to an algorithm that identifies and removes noisy trajectory data generated by legitimate operations from the original trajectory data. Denoising calculation refers to the process of using the trajectory denoising algorithm to process the original trajectory data of the vehicle to be analyzed, removing noisy trajectory points determined to be in a temporary ignition state or an internal scheduling state within the park. Denoised trajectory data refers to the remaining valid trajectory data after the original trajectory data has undergone denoising calculation.

[0033] The knowledge retrieval agent 1021 is used to obtain vehicle application forms, whitelists, holiday databases, and point-of-interest knowledge bases associated with the vehicle to be analyzed.

[0034] A vehicle use application form is an electronic document recording the reason for applying for official vehicle use. As an example, it may include the text of the application reason, the time of use, and destination information, serving as a compliance reference for determining violations. A whitelist database is a pre-stored collection of knowledge about compliant locations and related business documents. As an example, it may include the names, addresses, coordinates, and corresponding business document numbers of designated repair shops, designated gas stations, and partner units, used for compliance verification of vehicle parking behavior. A holiday database is a calendar database storing national statutory holidays, weekends, and their corresponding time ranges, used to mark whether the analysis time is a non-working day. A point-of-interest (POI) knowledge base is used to map latitude and longitude coordinates to location types with semantic meaning.

[0035] The trajectory semantic intelligent agent 1022 is used to extract stop points from the denoised trajectory data, obtain the location coordinates and stop duration data of the stop points, and perform reverse geocoding on the location coordinates of the stop points to obtain the point of interest category label data and regional attribute data of the stop points.

[0036] Stop point extraction refers to the process of identifying spatially clustered and temporally continuous location sets of vehicles from denoised trajectory data using density clustering algorithms, and extracting the positions and durations in which vehicles remain relatively stationary within a certain time period. Stop point location coordinate data refers to the center point coordinates obtained by calculating the average of the latitude and longitude coordinates of all trajectory points belonging to the same stop point cluster. Stop duration data refers to the length of a continuous period of time during which a vehicle remains stationary or engages in small-scale activity within the same stop area. Reverse geocoding refers to the process of inputting the stop point location coordinates into a geocoding system interface and converting them into readable address descriptions and surrounding points of interest (POIs). POI category label data refers to the category labels of POIs based on their function and attributes. As an example, POI category label data can include a primary category, such as "Leisure and Entertainment," and secondary categories, such as "KTV," reflecting the type of venue at the stop point. Regional attribute data refers to the functional attributes of the area where the stop point is located, determined based on the address components in the address information.

[0037] For example, the density clustering algorithm can be calculated using the spatiotemporal density clustering algorithm (ST-DBSCAN algorithm), including parameter settings: setting a spatial distance threshold. As an example, The value can be 100-300 meters, or a time span threshold. Minimum number of points Spatiotemporal distance metric calculation: For any two GPS points, the spherical distance is calculated using the Haversine formula, and the temporal distance is the timestamp difference; Core point identification: Traverse all GPS points and count the number of points in their spatiotemporal neighborhood. If the core point is found... These are then marked as core points; Cluster expansion and boundary point processing: Starting from each core point, recursively visit all points in its spatiotemporal neighborhood to form a complete cluster; Noise point identification: GPS points that do not belong to any cluster are marked as noise points; Stop point feature extraction: For each cluster, extract feature information such as the center coordinates of the stop points, the stop duration, and the stop time range; Stop point quality assessment: Evaluate the quality of the extracted stop points and remove low-quality stop points. As an example, The value can range from 100 to 300 meters. The value can be 5-10 minutes. The value can be 5-10 points.

[0038] The intelligent agent 1031 is used to transform the application reason in the car use application form into the expected behavior feature vector, and the interest point category label data and regional attribute data into the actual behavior feature vector. Cross-modal consistency verification is performed on the expected behavior feature vector and the actual behavior feature vector to obtain the consistency verification result. Based on the location coordinates of the stop point, the stay duration, the whitelist database and holiday information, risk classification and time dimension analysis are performed to obtain the violation conclusion and evidence chain.

[0039] The reason for application refers to the textual description of the purpose of vehicle use recorded in the vehicle use application form. The expected behavior feature vector is a high-dimensional numerical vector obtained by inputting the text of the reason for application from the vehicle use application form, after key entity extraction, semantic completion, and text reconstruction, into a pre-trained text embedding model. It is used to represent the semantic meaning of the vehicle's expected behavior for this trip. The actual behavior feature vector is a high-dimensional numerical vector obtained by inputting the text concatenation and semantic expansion of the point of interest category labels and region attributes of the stop point into a pre-trained text embedding model. It is used to represent the semantic meaning of the vehicle's actual stop behavior. Cross-modal consistency verification involves calculating the semantic similarity between the expected behavior feature vector and the actual behavior feature vector. By comparing the distance between the two in the semantic space, it determines whether the vehicle's reason for application matches the actual driving behavior. The consistency verification result is the conclusion drawn after calculating the semantic similarity between the expected behavior feature vector and the actual behavior feature vector, indicating whether they match. Risk grading is used to distinguish the severity of violations. Time dimension analysis involves determining whether the vehicle violated regulations during non-application periods or holidays based on the application time, actual usage time, and holiday periods. A conclusion of non-compliance refers to the final judgment made after comprehensively considering the results of cross-modal consistency verification, risk classification, and time-dimensional analysis. A chain of evidence refers to the complete set of evidence supporting the conclusion of non-compliance.

[0040] For example, consistency verification results can be obtained by calculating semantic distance. If Dist > Threshold, the judgment result is semantically irrelevant. The output result is: "The cause and trajectory do not match, suspected fraudulent application." Here, Threshold is the semantic dissimilarity threshold, which can take a value of 0.4. This is the expected behavior feature vector. This represents the feature vector of actual behavior.

[0041] This embodiment provides a public vehicle violation analysis system. By utilizing a data perception and cleaning agent to perform refined noise reduction on the original trajectory, noise data generated by legitimate operations such as temporary ignition and vehicle relocation within the park is eliminated, solving the technical problem of high false alarm rate. Through a trajectory semanticization agent, the trajectory coordinates are reverse geocoding and point-of-interest semantic mapping are performed, transforming the coordinates into computable semantic tags. The comprehensive judgment agent transforms the application reason and the actual trajectory into semantic feature vectors respectively, calculates semantic similarity, and combines whitelists and holiday times for comprehensive judgment, solving the technical problem of traditional solutions lacking reasoning ability, thereby accurately identifying violations and reducing the false alarm rate.

[0042] In an optional embodiment, the trajectory data includes a trajectory sequence, and the denoising calculation includes removing temporary ignition trajectory data. The data perception and cleaning agent 1011 acquires the trajectory data of the vehicle to be analyzed and uses a trajectory denoising algorithm to perform denoising calculations on the trajectory data of the vehicle to be analyzed. The specific steps include: Step a1: The data perception and cleaning agent acquires the trajectory sequence of the vehicle to be analyzed within the time window, and calculates the displacement entropy and velocity variance of each trajectory point based on the trajectory sequence.

[0043] Displacement entropy measures the dispersion of the displacement of a vehicle's trajectory points within a time window, representing the distance the vehicle travels. Velocity variance measures the fluctuation of the vehicle's velocity values ​​within the time window, representing the velocity variation of the vehicle.

[0044] For example, the time window can be 5 minutes, and there is no limitation on this.

[0045] For example, the displacement entropy can be calculated according to the following formula:

[0046] in, For displacement entropy, The probability value is the probability that a trajectory point falls into the i-th displacement range. Specifically, it is the proportion of the number of trajectory points whose displacement falls into the i-th range after dividing the displacement of all trajectory points in the time window into n ranges.

[0047] Step a2: If the displacement entropy is less than the first preset threshold, and the speed variance is less than the second preset threshold, and the vehicle has an accessory power gear signal jump, the trajectory point is determined to be either the original ignition or idling state.

[0048] The first preset threshold is a pre-set displacement entropy threshold, used to determine whether the dispersion of the trajectory point displacement distribution is low enough. If the displacement entropy is less than this threshold, it indicates that the vehicle's displacement change within the time window is minimal, and it is in a state of almost no movement. The second preset threshold is a pre-set speed variance threshold, used to determine whether the fluctuation of the trajectory point speed is low enough. If the speed variance is less than this threshold, it indicates that the vehicle's speed change within the time window is minimal. The accessory power signal transition refers to the effective state switching of the vehicle's accessory power signal (ACC signal) within the time window.

[0049] For example, in addition to using a first preset threshold and a second preset threshold, the determination of displacement entropy and velocity variance can also be made by using... And velocity variance The judgment is made in this way.

[0050] For example, the power supply signal transition refers to a valid state change (e.g., from "off" to "on / ignition" or from "on / ignition" to "off") occurring at least once within a time window, where the transition duration exceeds a preset threshold, the displacement entropy is less than a first preset threshold, and the velocity variance is less than a second preset threshold. As an example, the preset threshold can be 3-5 seconds to filter out instantaneous signal jitter.

[0051] Step a3: Remove the trajectory points of the vehicle in the original ignition or idling state from the trajectory data of the vehicle to be analyzed.

[0052] After identifying and judging the temporary ignition trajectory points, the system removes these trajectory points, which are judged to be in a stationary ignition or idling state, from the complete trajectory data of the vehicle to be analyzed, and they will no longer be used as the basis for subsequent violation analysis. For data judged to be in a stationary ignition or idling state, although there is an ACC signal jump, that is, the vehicle has made an ignition action, the vehicle has not actually moved, and is essentially in a non-moving state. If these false data are not deleted, the system may mistakenly think that the vehicle has "abnormal stop" or "illegal parking" at these locations, thus generating false alarms.

[0053] This embodiment provides a public vehicle violation analysis system. By calculating the displacement entropy and velocity variance of trajectory points within a time window, and combining this with the power gear signal jump data, the system identifies and eliminates trajectory points in the stationary ignition or idling state. This solves the technical problem in traditional solutions of misjudging legitimate operations such as waiting in place or temporary ignition as illegal stops, thereby reducing the false alarm rate.

[0054] In an optional embodiment, the trajectory data includes a set of trajectory points, and the denoising calculation includes removing the park parking trajectory data. The data perception and cleaning agent 1011 acquires the trajectory data of the vehicle to be analyzed and uses a trajectory denoising algorithm to perform denoising calculations on the trajectory data of the vehicle to be analyzed. The specific steps include: Step b1: The data perception and cleaning agent acquires the set of trajectory points of the vehicle to be analyzed within the time window and the sequence of polygon vertices in the park.

[0055] A trajectory point set refers to the collection of multiple trajectory points continuously collected by the vehicle to be analyzed within a time window. A park polygon vertex sequence refers to a sequence of coordinates of closed polygon vertices connected in a clockwise or counterclockwise order, used to represent the administrative area of ​​a park or the boundary of historical normal parking points.

[0056] For example, the sequence of polygon vertices in the park can be represented as .

[0057] Step b2: If all trajectory points in the trajectory point set are located within the area formed by the polygon vertex sequence of the park, and the displacement distance corresponding to the trajectory point set is less than the maximum span threshold of the area formed by the polygon vertex sequence of the park, then the trajectory point set is determined to be the internal scheduling trajectory data of the park.

[0058] Displacement distance refers to the straight-line distance between the first and last trajectory points in the trajectory point set within a time window, used to measure the overall displacement of the vehicle within the time window. The maximum span threshold refers to the maximum straight-line distance between any two points within the area formed by the polygon vertex sequence of the park, serving as the basis for determining whether a vehicle is only dispatched short distances within the park. Park-internal dispatch trajectory data refers to trajectory data corresponding to park-internal dispatch behavior where all trajectory points in the trajectory point set are located within the area formed by the polygon vertex sequence of the park, and the displacement distance is less than the maximum span threshold of that area.

[0059] For example, the trajectory point set can be determined using the ray method. With the vertex sequence of the park polygon The topological relationship is as follows: from Take the trajectory point to be judged from the middle. From point Draw a horizontal ray R along the positive X-axis; calculate the intersection of ray R and the polygon. The number of intersection points N of all edges; if N is odd, then determine the point. lie in Internally, if N is an even number containing 0, then the decision point is... lie in External; Traversal For all trajectory points, calculate the proportion of points inside the trajectory. If all points are inside the trajectory, then the trajectory point set is considered complete. and The topological relationship is "internal containment"; if all trajectory points are located within... Internal, and displacement distance It was determined to be "internal scheduling within the park," marked with the IGNORE tag, and will not proceed to the subsequent violation analysis process. This is the maximum span within the park. Displacement distance. The set of trajectory points within a time window The Euclidean distance between the first and last trajectory points.

[0060] Step b3: Remove the internal scheduling trajectory data of the park from the trajectory data of the vehicles to be analyzed.

[0061] After identifying and determining the internal scheduling trajectory data of the park, the system will delete the trajectory point set corresponding to the internal scheduling trajectory data from the complete trajectory data of the vehicle to be analyzed, and will no longer use it as the basis for subsequent violation analysis.

[0062] This embodiment provides a public vehicle violation analysis system. By acquiring the trajectory point set of the vehicle to be analyzed within a time window and the vertex sequence of the park polygon, and determining whether all trajectory points are located inside the park polygon and the displacement distance is less than the maximum span threshold, the system identifies and removes scheduling trajectory data within the park. This solves the technical problem of misjudging legitimate scheduling behaviors such as normal vehicle movement and entering / exiting the garage within the unit's park as violation of vehicle use, and further reduces the false alarm rate.

[0063] In an optional embodiment, the trajectory semantic agent 1022 performs inverse geocoding on the location coordinates of the stop point to obtain point of interest category label data and regional attribute data of the stop point. The specific steps include: Step c1: The trajectory semantic agent performs inverse geocoding calculation on the location coordinates of the stop point to obtain the address information of the stop point and the information of surrounding points of interest.

[0064] Address information refers to structured address description data returned after reverse geocoding of the location coordinates of the stop point, used to determine the specific administrative location and regional attributes of the stop point. Nearby points of interest information refers to a list of nearby points of interest returned after reverse geocoding of the location coordinates of the stop point, used to filter out the points of interest most relevant to the stop point and extract category tags.

[0065] Step c2: Determine relevant points of interest based on surrounding point of interest information, extract tags from relevant points of interest, and obtain point of interest category tag data for the stop.

[0066] Relevant points of interest (POIs) refer to selecting a POI that represents the actual semantic meaning of the location at the point of stay from the list of surrounding POIs returned by reverse geocoding. This POI serves as the object for subsequent tag extraction and semantic analysis. Tag extraction refers to the process of reading and outputting the corresponding category labels from the selected POI information according to a preset classification system, including primary categories and secondary categories, thereby obtaining the POI category label data for the point of stay.

[0067] Step c3: Determine the regional attribute data of the stop point based on the surrounding points of interest information.

[0068] For example, by performing inverse geocoding on the location coordinates of the stop point to obtain the point of interest category label data and regional attribute data of the stop point, the calculation can be performed according to the following process, including: obtaining the stop point coordinates: the stop point center extracted based on the ST-DBSCAN clustering algorithm. The process involves: obtaining the latitude and longitude coordinates (lng, lat); calling the reverse geocoding API: selecting a suitable map open platform, calling its reverse geocoding API, and passing in the coordinates and search radius parameters; parsing the API response data: the API returns response data in JavaScript object representation (JSON format), including formatted address, address components, and a list of nearby points of interest; filtering and assessing the confidence level of points of interest: selecting the most relevant points of interest from the returned list as semantic information for the stop point; extracting category labels for points of interest: extracting category labels from the selected point of interest information, including primary categories and secondary categories; determining regional attributes: based on the address components returned by reverse geocoding, determining the regional attributes of the stop point, including city level, region type, and functional zoning.

[0069] This embodiment provides a public vehicle violation analysis system. By performing reverse geocoding calculations on the location coordinates of the stopping points, it obtains address information and surrounding points of interest information. Based on the surrounding points of interest information, it determines relevant points of interest for tag extraction and determines regional attributes. This transforms abstract latitude and longitude coordinates into point of interest category tags and regional attribute data with clear semantic meanings, providing a semantic data foundation for subsequent comprehensive judgment of intelligent agents to perform consistency verification and risk classification.

[0070] In an optional embodiment, the intelligent agent 1031 comprehensively analyzes and transforms the reason for the vehicle use application into a feature vector of expected behavior. The specific steps include: Step d1: Comprehensively analyze and determine the text of the application reason in the vehicle application form obtained by the intelligent agent.

[0071] The application reason text refers to the natural language description recorded in the vehicle use application form, which explains the purpose and intended use of the official vehicle. As an example, the application reason text could be "Going to Village A for exploration".

[0072] Step d2 involves extracting key entities from the application reason text to obtain destination data, purpose data, and business attribute data.

[0073] Key entity extraction refers to the process of inputting the application reason text into a large language model, and using prompts and engineering guidance to identify and extract structured key information related to vehicle usage behavior from unstructured natural language text. Destination data is used to characterize the target location of the vehicle trip. Purpose data is used to characterize the task of the vehicle trip. Business attribute data is used to characterize the scenario category of the vehicle trip.

[0074] As an example, when the application reason text is "Exploration trip to Village A", the key entity extraction is: .

[0075] Step d3 involves semantic completion and text reconstruction of the destination data, purpose data, and business attribute data to generate a description text of the expected behavior.

[0076] Semantic completion refers to the process of contextualizing and expanding the implicit information of extracted key entities. Text reconstruction refers to the process of reorganizing the semantically completed key entities according to a preset text template or logical order to generate a complete descriptive text with a clear structure and coherent semantics. Expected behavior description text refers to the complete semantic description of the vehicle's expected driving behavior generated by semantic completion and text reconstruction, which is used as input to a preset text embedding model to generate expected behavior feature vectors.

[0077] For example, during semantic completion, a large model is used to complete the implicit contextual scene features, such as "exploration" implying "outdoor / non-commercial area" and "meeting" implying "business / office building"; during text reconstruction, the purpose attribute and the destination area attribute are concatenated to generate standardized expected behavior description text, such as "outdoor exploration work + rural area".

[0078] Step d4: Input the expected behavior description text into the preset text embedding model to obtain the expected behavior feature vector.

[0079] Text embedding models are deep learning models that convert natural language text into fixed-dimensional numerical vectors.

[0080] For example, inputting the text describing the expected behavior into a pre-trained text embedding model to generate a high-dimensional feature vector of the expected behavior can be represented as: ,in, This is the expected behavior feature vector. Describe the expected behavior.

[0081] This embodiment provides a public vehicle violation analysis system. By extracting key entities from the application text in the vehicle application form, destination data, purpose data, and business attribute data are obtained. After semantic completion and text reconstruction, the results are input into a text embedding model to generate expected behavior feature vectors. This transforms the unstructured natural language application into a computable numerical vector form, enabling the system to quantitatively determine whether the actual driving behavior matches the application.

[0082] In an optional embodiment, the intelligent agent 1031 comprehensively evaluates and transforms the interest point category labels and region attributes into actual behavior feature vectors. The specific steps include: Step e1: Comprehensively analyze the intelligent agent to obtain the interest point category labels and region attributes.

[0083] The point of interest category labels and region attributes are derived from the inverse geocoding of the location coordinates of the stop points by the trajectory semantic agent.

[0084] Step e2 involves concatenating and semantically expanding the text of the interest point category labels and region attributes to obtain the actual behavior description text.

[0085] Text concatenation refers to the process of merging point-of-interest category label data and regional attribute data according to a preset text template or connection method to form a combined text containing multiple pieces of information. Semantic expansion refers to the process of adding contextual associations and higher-level concepts to the combined text generated after text concatenation, making the semantics of the actual behavior description text more complete. Actual behavior description text refers to natural language text that provides a complete semantic description of the actual stopping behavior of vehicles.

[0086] Step e3: Input the actual behavior description text into the preset text embedding model to obtain the actual behavior feature vector.

[0087] The actual behavior description text is input into a pre-defined text embedding model, which maps the natural language text into a fixed-dimensional vector representation, thereby obtaining the actual behavior feature vector.

[0088] For example, the actual behavior feature vector can be represented as Furthermore, the actual behavior feature vector can be normalized so that its length is 1.

[0089] This embodiment provides a public bus violation behavior analysis system. By concatenating and semantically expanding the text of the point of interest category labels and regional attributes, the system generates actual behavior description text, which is then input into a preset text embedding model to obtain actual behavior feature vectors. This process transforms unstructured location category labels and regional attribute information into structured, computable numerical vectors, enabling the system to compare the similarity between the actual behavior feature vectors and the expected behavior feature vectors in the same semantic space.

[0090] In an optional embodiment, the violation conclusion includes unauthorized stay in sensitive locations and unauthorized parking on holidays. The intelligent agent 1031 comprehensively analyzes the location coordinates of the stay point, the duration of stay, the whitelist database, and holiday information to perform risk grading and time-dimensional analysis, obtaining the violation conclusion. The specific steps include: Step f1 involves comprehensively analyzing the data obtained by the intelligent agent regarding the range of the unit's sealing points, the location coordinates of the dwell points, the dwell time, the whitelist database, and holiday information.

[0091] The data on the scope of the unit's sealing points refers to the coordinate data of the designated parking locations for official vehicles during statutory holidays or public holidays as defined within the unit.

[0092] Step f2: Determine the current point of interest category label based on the location coordinates and duration of the stop.

[0093] This step is the same as obtaining the point of interest category labels in step c2, so it will not be repeated here.

[0094] Step f3: If the current point of interest category tag belongs to the preset risk location set and the stay duration exceeds the preset duration threshold, it is determined to be a violation of the rules for staying in a sensitive location.

[0095] The preset risk location set refers to a pre-defined set of point-of-interest (POI) category labels for locations deemed high-risk or sensitive. The preset duration threshold refers to a pre-set threshold for the length of time a vehicle remains at a location, used to determine whether such a stay constitutes a violation.

[0096] For example, methods for determining unauthorized lingering in sensitive locations may include: if If the whitelist is not matched, the Points of Interest (POI) database will be searched. The POI database stores basic identification information: unique POI ID and name; spatial location information: longitude, latitude, and administrative division code; semantic classification tags: primary industry categories (e.g., "Leisure and Entertainment," "Catering Services"), secondary industry categories (e.g., "KTV," "Bar"), and tertiary industry subcategories; and auxiliary attribute information: business status, venue rating, and average spending per person. If the POI tag ∈ {KTV, Sauna, Bar, Golf…} and T duration >T threshold T durationFor the duration of stay, T threshold A preset stay duration threshold is set, which can be 30 minutes. The output result can be: "Suspected unauthorized use of a vehicle to enter or exit an entertainment venue," along with a screenshot as evidence.

[0097] Step f4: If the location coordinates of the stop point are within the whitelist range in the whitelist database, then the stop is determined to be compliant.

[0098] For example, constructing a dynamic whitelist knowledge graph This can include designated repair shops, designated gas stations, and partner companies. As an example, the steps to build a whitelist are as follows: Data access: Extract the names, addresses, and associated business order numbers of designated repair shops, designated gas stations, and partner companies from the enterprise's internal system; Geocoding: Convert the address text into latitude and longitude coordinates using a geocoding API; Ontology definition: Define the graph ontology structure, including "whitelist location entities," "business order entities," and "relationships"; Knowledge storage and update: Store the entities and relationships in the graph database and set up a scheduled synchronization task to automatically update the graph nodes and edges when the business system adds or changes partner companies.

[0099] As an example, calculate the dwell point. The spatial distance from the whitelist node can be: based on Based on the latitude and longitude and whitelist type, a set of candidate whitelist nodes for the same city or region is retrieved from the graph database; the Haversine formula is used to calculate... The spherical distance to each candidate whitelist node; the minimum distance in the calculation results is taken as the final spatial distance Distance.

[0100] For example, the judgment logic of the intelligent agent can be as follows: If Distance < δ, trigger the "supporting evidence retrieval" tool to check if there is a corresponding repair work order or refueling record. If there is a record: mark it as compliant and terminate the analysis. If there is no record: mark it as "abnormal stay at a whitelist location (no work order support)" and increase the risk level. δ is a preset spatial distance threshold.

[0101] Step f5: If the current time is a holiday and the location coordinates of the stopping point are not within the unit's sealed point range data, then it is determined to be a non-public parking on a holiday.

[0102] For example, non-public parking on holidays can be determined as follows: The holiday rule base is read, and the current time slice attribute is marked. If it is a holiday or public holiday, and the vehicle is parked in a... Combining semantic analysis of points of interest: If The risk level is medium, suspected of using a public vehicle for personal purposes to go home; if... The risk level is high, indicating a possible case of a public bus being used for tourism. Output: "Illegal parking during holidays," recording the specific location and duration.

[0103] This embodiment provides a public vehicle violation analysis system. By acquiring data on the scope of unit-sealed areas, the location coordinates of stopping points, the duration of stay, a whitelist database, and holiday information, and based on the relationship between the current point of interest category tag and a preset risk location set, and the relationship between the duration of stay and a preset duration threshold, it determines illegal stays in sensitive locations. Based on the location coordinates of the stopping points and the scope of the whitelist database, it determines compliant stays. Furthermore, based on whether it is a holiday or not, and the relationship between the location coordinates of the stopping points and the unit-sealed area data, it determines non-public parking on holidays. Ultimately, it can identify various violation scenarios such as prolonged stays in sensitive locations and illegal parking on holidays, avoiding false alarms in traditional solutions and improving the accuracy of the system.

[0104] Figure 2 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0105] The following is a detailed reference. Figure 2 The diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 201, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 202 or a program loaded from memory 208 into random access memory (RAM) 203. The RAM 203 also stores various programs and data required for the operation of the electronic device. The processor 201, ROM 202, and RAM 203 are interconnected via a bus 204. An input / output (I / O) interface 205 is also connected to the bus 204.

[0106] Typically, the following devices can be connected to I / O interface 205: input devices 206 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 207 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 208 including, for example, magnetic tapes, hard disks, etc.; and communication devices 209. Communication device 209 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 2 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0107] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 209, or installed from memory 208, or installed from ROM 202. When the computer program is executed by processor 201, it performs the functions defined in a public vehicle violation behavior analysis system according to embodiments of the present invention.

[0108] Figure 2 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0109] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, it implements the bus violation behavior analysis system shown in the above embodiments.

[0110] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0111] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A system for analyzing illegal use of public vehicles, characterized in that, The system comprises a perception layer, a reasoning layer, and a decision layer. The perception layer includes a data perception and cleaning agent; the reasoning layer includes a knowledge retrieval agent and a trajectory semanticization agent; and the decision layer includes a comprehensive judgment agent. The data sensing and cleaning intelligent agent is used to acquire the trajectory data of the vehicle to be analyzed, and to use a trajectory denoising algorithm to perform denoising calculation on the trajectory data of the vehicle to be analyzed to obtain denoised trajectory data. The knowledge retrieval intelligent agent is used to obtain vehicle application forms, whitelists, holiday databases, and interest point knowledge bases associated with the vehicle to be analyzed. The trajectory semantic intelligent agent is used to extract stop points from the denoised trajectory data, obtain the location coordinates and stop duration data of the stop points, and perform reverse geocoding on the location coordinates of the stop points to obtain the point of interest category label data and regional attribute data of the stop points. The comprehensive judgment intelligent agent is used to convert the application reason in the vehicle application form into an expected behavior feature vector, convert the point of interest category label data and regional attribute data into an actual behavior feature vector, perform cross-modal consistency verification on the expected behavior feature vector and the actual behavior feature vector, obtain the consistency verification result, and perform risk classification and time dimension analysis based on the location coordinates of the stop point, the stay duration, the whitelist database and holiday information to obtain the violation conclusion and evidence chain.

2. The system according to claim 1, characterized in that, The trajectory data includes trajectory sequences, and the denoising calculation includes removing temporary ignition trajectory data. The data perception and cleaning agent acquires the trajectory data of the vehicle to be analyzed and performs denoising calculations on the trajectory data of the vehicle to be analyzed using a trajectory denoising algorithm, including: The data sensing and cleaning agent acquires the trajectory sequence of the vehicle to be analyzed within the time window, and calculates the displacement entropy and velocity variance of each trajectory point based on the trajectory sequence. If the displacement entropy is less than the first preset threshold, and the velocity variance is less than the second preset threshold, and the vehicle has an accessory power gear signal jump, the trajectory point is determined to be in the original ignition or idling state. The trajectory points of the original ignition or idling state are removed from the trajectory data of the vehicle to be analyzed.

3. The system according to claim 1, characterized in that, The trajectory data includes a set of trajectory points. The denoising calculation includes removing park parking trajectory data. The data perception and cleaning agent acquires the trajectory data of the vehicle to be analyzed and performs denoising calculation on the trajectory data of the vehicle to be analyzed using a trajectory denoising algorithm, including: The data sensing and cleaning agent acquires the trajectory point set of the vehicle to be analyzed within the time window and the sequence of polygon vertices in the park. If all trajectory points in the trajectory point set are located within the region formed by the sequence of polygon vertices in the park, and the displacement distance corresponding to the trajectory point set is less than the maximum span threshold of the region formed by the sequence of polygon vertices in the park, then the trajectory point set is determined to be internal scheduling trajectory data of the park. The internal scheduling trajectory data of the park is removed from the trajectory data of the vehicle to be analyzed.

4. The system according to claim 1, characterized in that, The trajectory semantic agent performs inverse geocoding on the location coordinates of the stop point to obtain the point of interest category label data and regional attribute data of the stop point, including: The trajectory semantic intelligent agent performs inverse geocoding calculation on the location coordinates of the stop point to obtain the address information of the stop point and the information of surrounding points of interest. Based on the surrounding point of interest information, relevant points of interest are determined, and tags are extracted from the relevant points of interest to obtain the point of interest category tag data of the stop; The location attribute data of the stop point is determined based on the surrounding points of interest information.

5. The system according to claim 1, characterized in that, The comprehensive analysis and judgment agent transforms the application reason in the vehicle use application form into a predictive behavioral feature vector, including: The comprehensive analysis and judgment agent obtains the application reason text in the vehicle use application form; Key entities are extracted from the application text to obtain destination data, purpose data, and business attribute data; Semantic completion and text reconstruction are performed on the destination data, purpose data, and business attribute data to generate expected behavior description text; The expected behavior description text is input into a preset text embedding model to obtain the expected behavior feature vector.

6. The system according to claim 1, characterized in that, The comprehensive judgment agent transforms interest point category labels and region attributes into actual behavioral feature vectors, including: The comprehensive analysis agent acquires the interest point category label and the region attribute; The text is concatenated and semantically expanded on the interest point category labels and the region attributes to obtain the actual behavior description text; The actual behavior description text is input into a preset text embedding model to obtain the actual behavior feature vector.

7. The system according to claim 1, characterized in that, The violation conclusions include unauthorized stays in sensitive locations and unauthorized parking during holidays. The comprehensive judgment agent performs risk classification and time-dimensional analysis based on the location coordinates of the stay point, the duration of stay, the whitelist database, and holiday information to obtain violation conclusions, including: The comprehensive analysis and judgment intelligent agent obtains data on the range of unit sealing points, the location coordinates of the dwell points, the dwell time, the whitelist database, and holiday information; The current point of interest category label is determined based on the location coordinates and duration of the stop point; If the current point of interest category tag belongs to a preset risk location set and the stay duration exceeds a preset duration threshold, it is determined to be a violation of the rules for staying in a sensitive location. If the location coordinates of the stop point are within the whitelist range in the whitelist database, then the stop is determined to be compliant. If the current time is a holiday and the location coordinates of the stop point are not within the unit's sealed-off point range data, then it is determined to be a non-public parking on a holiday.

8. An electronic device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform a public vehicle violation analysis system according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the public vehicle violation analysis system according to any one of claims 1 to 7.

10. A computer program product, characterized in that, It includes computer instructions for causing a computer to execute a public vehicle violation analysis system according to any one of claims 1 to 7.

Images