A parking lot linkage control system based on edge computing

Abstract

The application discloses a parking lot linkage control system based on edge computing and relates to the technical field of intelligent traffic, which comprises an edge coordination module, a feature analysis module, an abnormality judgment module and a control execution module. The edge coordination module is used for constructing a distributed coordination network and synchronizing vehicle state data and control instructions. The feature analysis module is used for collecting vehicle features in real time, performing logical comparison based on vehicle archives to complete identity category determination, binding the determination result with the vehicle features, and generating a lightweight identity feature label. The abnormality judgment module is used for receiving the identity feature label, monitoring vehicle behavior in real time, and executing preset judgment rules in combination with the identity feature label and behavior data. The control execution module is used for receiving abnormal event judgment results output by the abnormality judgment module, dynamically matching linkage control strategies based on the abnormal event judgment results, and generating a report containing a complete event chain locally at an edge node.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, specifically a parking lot linkage control system based on edge computing. Background Technology

[0002] With the acceleration of urbanization, parking lot management faces challenges such as low traffic efficiency, high labor costs, and slow response to anomalies. Traditional parking systems mostly adopt a centralized architecture of front-end data collection and cloud processing. Vehicle data needs to be uploaded to a remote server for unified analysis, judgment, and control command issuance. Under this architecture, network latency leads to an excessively long response cycle from the occurrence of an event to the execution of control, resulting in a poor user experience, especially in scenarios requiring immediate intervention such as vehicle lingering or payment anomalies. Furthermore, the system is highly dependent on the continuous stability of the cloud and the network; once the network is down, the entire parking lot operation may be paralyzed. Finally, the continuous uploading of all high-definition video streams puts enormous pressure on network bandwidth and cloud computing power.

[0003] While some existing technologies deploy processing units locally in parking lots, these are mostly isolated lane controllers or toll collection points with simple processing logic, capable of only performing single functions such as license plate recognition and toll collection. Furthermore, there is a lack of efficient collaboration mechanisms between these devices, hindering the implementation of coordinated control based on comprehensive parking lot information. For example, entrance devices cannot share vehicle identification information in real time with exit and in-park monitoring points; when an anomaly occurs at the exit, the control center still needs to manually retrieve data from multiple sources to make a decision, indicating insufficient automation and intelligence. Summary of the Invention

[0004] The purpose of this invention is to provide a parking lot linkage control system based on edge computing to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a parking lot linkage control system based on edge computing, the system comprising: an edge collaboration module, a feature analysis module, an anomaly determination module, a control execution module, and a report generation module;

[0006] The edge collaboration module is used to construct a distributed collaboration network connecting the edge nodes of the parking lot entrance, lanes and exit, ensuring low latency and high reliability communication between nodes in the network, and synchronizing vehicle status data and control commands within the distributed collaboration network; the vehicle status data originates from the processing results of each functional module, and the control commands are directed to specific execution devices;

[0007] The feature analysis module is deployed at the edge node of the parking lot entrance. It is used to collect vehicle features in real time, perform logical comparison based on vehicle files to complete the identity category determination, and bind the determination result with the vehicle features to generate a lightweight identity feature tag. This tag serves as the only data carrier for vehicle identity perception and differentiation processing throughout the parking lot. It is also synchronized to the relevant lane and exit edge nodes through the edge collaboration module to realize the accompanying distribution of vehicle identity information in the edge network.

[0008] The anomaly detection module is deployed at lane and exit edge nodes to receive the identity feature tags and monitor vehicle behavior in real time. Based on the vehicle identity information contained in the received identity feature tags, it executes a preset detection rule in combination with the identity feature tags and behavior data. The core logic of this detection rule is to determine the benchmark for anomaly detection based on the vehicle identity, and then perform event mapping on the benchmark based on the behavior data to output an anomaly event detection result containing event type, event level and vehicle identity information, thus completing intelligent event diagnosis that associates identity with behavior.

[0009] The control execution module is used to receive the abnormal event judgment result output by the abnormal judgment module, and dynamically match the linkage control strategy based on the abnormal event judgment result. The matching process includes: determining the basic control action according to the event type and level, making differentiated adjustments to the execution parameters of the basic control action according to the vehicle identity information carried in the abnormal event judgment result, and controlling the barrier gate, display screen and voice broadcasting device to perform corresponding operations.

[0010] The report generation module is used to query vehicle entry records, vehicle identity information and behavior data in association when the anomaly determination module outputs the anomaly event determination result, so as to construct a full-link event view from identity determination, anomaly occurrence to control execution, and generate a report containing the complete event chain locally at the edge node.

[0011] According to the above scheme, the edge collaboration module includes a network construction unit, a communication management unit, and a data routing unit;

[0012] The network construction unit is used to construct a physical communication topology connecting the edge nodes of the parking lot entrance, lanes and exit, and the physical communication topology includes redundant communication links;

[0013] The communication management unit is used to monitor the connectivity status of the redundant communication links and automatically switch to the backup communication link when the primary communication link fails, so as to maintain the connectivity of the distributed collaborative network.

[0014] The data routing unit is used to perform targeted synchronization of vehicle status data and control commands within the distributed collaborative network according to a preset address mapping and message type. The vehicle status data includes the identity feature tag and vehicle behavior monitoring data, and the control commands include barrier gate control commands and display broadcast commands.

[0015] When the distributed collaborative network and cloud services are interrupted, the edge collaboration module maintains local collaboration and decision-making among edge nodes based on the vehicle status data, ensuring that the core closed loop of identity feature tag transfer, abnormal event judgment and control command generation and execution continues to operate autonomously in offline mode.

[0016] According to the above scheme, the feature analysis module includes a feature acquisition unit, an identity determination unit, and a tag generation synchronization unit;

[0017] The feature acquisition unit is used to acquire vehicle feature data, including vehicle appearance images and license plate information, through sensors deployed at the entrance of the parking lot.

[0018] The identity determination unit is used to logically compare the vehicle feature data with the vehicle archive database, and determine the vehicle identity category as a pre-stored vehicle, a temporary vehicle, or an unlicensed vehicle based on the comparison result.

[0019] The tag generation and synchronization unit is used to associate and encapsulate the identity category determination result with the vehicle feature data to generate the identity feature tag, and call the edge collaboration module to synchronize the encapsulated identity feature tag to the lane and exit edge nodes.

[0020] According to the above scheme, the identity determination unit includes a first determination logic based on license plate matching and a second determination logic based on vehicle appearance features;

[0021] The first determination logic is used to output the determination result of pre-stored vehicles or temporary vehicles based on the matching of license plate information in the vehicle feature data with the vehicle archive database.

[0022] The second determination logic is invoked when the first determination logic cannot output a determination result due to invalid license plate information. It outputs the determination result of the unlicensed vehicle and generates the corresponding vehicle feature identifier to ensure that all vehicles can obtain a unique identity that can be used for subsequent processes.

[0023] According to the above scheme, the anomaly determination module includes a behavior monitoring unit, a rule engine unit, and an event determination unit;

[0024] The behavior monitoring unit is used to acquire behavior monitoring data that characterizes the vehicle's operating status in real time;

[0025] The rule engine unit is used to store judgment rules that associate vehicle identity information, behavior monitoring data and abnormal events. The association relationship is reflected in an event mapping mechanism with vehicle identity information as the first-level index and behavior monitoring data as the second-level index.

[0026] The event determination unit is used to receive the identity feature tag, and based on the behavior monitoring data and the vehicle identity information in the identity feature tag, parse the vehicle identity information from the identity feature tag, and determine its basic sensitivity level through the identity classification rules in the rule engine unit; the basic sensitivity level is combined with the real-time behavior monitoring data and analyzed by the event mapping rules in the rule engine unit to generate the corresponding event type and level; the event type, event level and original vehicle identity information are packaged and output as the abnormal event determination result.

[0027] According to the above scheme, the determination rules include identity classification rules and mapping rules; the identity classification rules are used to determine the basic sensitivity level of the vehicle in the anomaly determination based on the vehicle identity information; the mapping rules are used to map and generate the corresponding abnormal event type and event level based on the combination of the basic sensitivity level and behavior monitoring data. This two-level rule structure forces the vehicle identity to be the primary logical input for anomaly determination.

[0028] According to the above scheme, the control execution module includes a strategy parsing unit, a device control unit, and a feedback unit;

[0029] The strategy parsing unit is used to receive and parse the abnormal event judgment result, and match the corresponding linkage control instruction set from the preset strategy library according to the event type, event level and vehicle identity information contained therein.

[0030] The equipment control unit is used to convert the linkage control instruction set into drive signals and send them to the corresponding barrier gate, display screen and voice broadcasting device;

[0031] The feedback unit is used to monitor the execution status of each controlled device after sending the drive signal, and synchronize the status feedback information to the edge collaboration module.

[0032] According to the above scheme, the strategy library is pre-set with an event and strategy mapping table and an identity and strategy weight table;

[0033] The event and policy mapping table is used to define the basic control instruction sequence corresponding to different combinations of event types and event levels;

[0034] The identity and strategy weight table is used to define the priority adjustment parameters or execution parameter correction values ​​of control commands for different vehicle identity information, so as to provide differentiated control parameters for vehicles with different identities under the same event.

[0035] When the strategy parsing unit matches the linkage control instruction set, it determines the basic instruction sequence according to the event and strategy mapping table, and modifies the parameters of the basic instruction sequence according to the identity and strategy weight table to generate the linkage control instruction set.

[0036] According to the above scheme, the report generation module includes an event association unit, a data retrieval unit, and a report compilation unit;

[0037] The event association unit is used to extract key index information from the abnormal event judgment results output by the abnormal judgment module.

[0038] The data retrieval unit is used to retrieve associated vehicle entry records, vehicle identity information, and historical behavior data from local storage based on key index information;

[0039] The report compilation unit is used to integrate the abnormal event judgment results with the retrieved related data in chronological and logical order, and generate a structured complete event report locally on the edge node. This report fully records the closed-loop event flow from identity feature binding to final control execution.

[0040] Compared with the prior art, the beneficial effects of the present invention are:

[0041] 1. This invention deploys edge collaboration modules at key nodes within the site, pushing core functions down to the network edge; through real-time data sharing via a distributed collaboration network, the determination and response to abnormal events are completed entirely locally, shortening the response time of execution control and improving passage efficiency and anomaly handling speed.

[0042] 2. This invention integrates vehicle identity information and behavioral data for anomaly detection, and achieves differentiated responses through identity classification rules and mapping rules. It matches appropriate control strategies for vehicles with different identities, ensuring management order and improving user experience.

[0043] 3. This invention synchronizes vehicle identity and feature information through lightweight identity feature tags, reducing redundant data transmission and bandwidth pressure; abnormal event reports are generated locally, integrating entry records, identity information and behavior sequences to form a complete event chain, facilitating rapid tracing and handling, and improving management efficiency. Attached Figure Description

[0044] Figure 1 This is a schematic diagram of the structure of a parking lot linkage control system based on edge computing according to the present invention. Detailed Implementation

[0045] 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.

[0046] Example: Figure 1 As shown, the present invention provides a technical solution, a parking lot linkage control system based on edge computing, which includes: an edge collaboration module, a feature analysis module, an anomaly determination module, a control execution module, and a report generation module;

[0047] The edge collaboration module is used to build a distributed collaborative network connecting the edge nodes of the parking lot entrance, lanes and exit, ensuring low latency and high reliability communication between nodes in the network, and synchronizing vehicle status data and control commands within the distributed collaborative network; vehicle status data originates from the processing results of each functional module, and control commands are directed to specific execution devices;

[0048] Specifically, the edge collaboration module includes a network construction unit, a communication management unit, and a data routing unit. The network construction unit is used to construct the physical communication topology connecting the edge nodes of the parking lot entrance, lanes, and exit. The physical communication topology includes redundant communication links. The communication management unit is used to monitor the connectivity status of the redundant communication links and automatically switch to the backup communication link when the primary communication link fails, maintaining the connectivity of the distributed collaboration network. The data routing unit is used to perform targeted synchronization of vehicle status data and control commands within the distributed collaboration network according to preset address mappings and message types. Vehicle status data includes identity feature tags and vehicle behavior monitoring data, and control commands include gate control commands and display broadcast commands.

[0049] Furthermore, when the distributed collaborative network and cloud services are interrupted, the edge collaboration module maintains local collaboration and decision-making among edge nodes based on vehicle status data, ensuring that the core closed loop of identity feature tag transfer, abnormal event judgment and control command generation and execution continues to operate autonomously in offline mode.

[0050] For example: the network construction unit of the edge collaboration module establishes a star physical communication topology based on the TCP / IP protocol according to the pre-configured IP address list, such as ingress node: 192.168.1.10, lane node A: 192.168.1.20, and egress node: 192.168.1.30, and configures the 4G module as a backup communication link; the communication management unit starts a dual-link heartbeat monitoring mechanism, sending test data packets to the primary and backup communication links once per second; if no valid response is received from the primary communication link for 5 consecutive cycles, If the primary link fails, the threshold for these five cycles is determined based on statistics of historical communication link failure cases in the parking lot. By analyzing the distribution of failure recovery time, it was found that these five cycles are the optimal judgment cycle, which can control the false judgment rate below 0.5%. The network switch is automatically executed, and all subsequent communication traffic is directed to the backup communication link to ensure the connectivity of the distributed collaborative network. The data routing unit loads the preset address mapping table. For example, messages of type VEHICLE_TAG need to be broadcast to all lanes and exit nodes, and the message queue is initialized. This is only an example and is not a limitation.

[0051] The feature analysis module is deployed at the edge node of the parking lot entrance to collect vehicle features in real time, perform logical comparison based on vehicle files to complete the identity category determination, and bind the determination result with the vehicle features to generate a lightweight identity feature label. This label serves as the only data carrier for vehicle identity perception and differentiation processing throughout the entire parking lot, and is synchronized to the relevant lane and exit edge nodes through the edge collaboration module to realize the accompanying distribution of vehicle identity information in the edge network.

[0052] Specifically, the feature analysis module includes a feature collection unit, an identity determination unit, and a label generation synchronization unit; the feature collection unit is used to obtain vehicle feature data including vehicle appearance images and license plate information through sensors deployed at the parking lot entrance; for example: the feature collection unit captures the front image of the vehicle through a high-definition camera and triggers the work; the feature collection unit calls the object detection algorithm based on convolutional neural network to locate the vehicle area in the image; applies the optical character recognition algorithm to the vehicle area to extract license plate number characters; at the same time, extracts the color histogram and scale-invariant feature transform key point description from the vehicle area image to jointly form the vehicle feature data; if the recognized license plate number characters are Beijing A12345, this is only for illustrative purposes and not for limitation; the identity determination unit is used to logically compare the vehicle feature data with the vehicle archive and determine the vehicle identity category as a pre-stored vehicle, a temporary vehicle, or a vehicle without a license plate according to the comparison result; for example: the first determination logic of the identity determination unit compares the string Beijing A12345 character by character with the vehicle archive stored in the local database; there is a pre-stored record in the vehicle archive with the license plate number Beijing A12345 and the status marked as monthly rental car; the comparison is consistent, so the first determination logic outputs the identity category as a pre-stored vehicle; since the license plate is valid and the match is successful, the second determination logic is not triggered, this is only for illustrative purposes and not for limitation; the label generation synchronization unit is used to associate and encapsulate the identity category determination result with the vehicle feature data, generate an identity feature label, and call the edge collaboration module to synchronize the encapsulated identity feature label to the lane and exit edge nodes in a targeted manner; for example: the label generation synchronization unit associates and encapsulates the pre-stored vehicle identity category determination result with the key vehicle features: license plate number Beijing A12345, color feature vector, and vehicle model contour feature vector; the encapsulation process includes calculating a simplified hash value of the feature vector, such as calculating the concatenated string of the feature vector using the MD5 algorithm to generate a 128-bit hash digest; generating a lightweight JSON format identity feature label, this is only for illustrative purposes and not for limitation;

[0053] Furthermore, the identity determination unit includes a first determination logic based on license plate matching and a second determination logic based on vehicle appearance features; the first determination logic is used to output the determination result of a pre-stored vehicle or a temporary vehicle according to the matching situation between the license plate information in the vehicle feature data and the vehicle archive; the second determination logic is used to be called when the first determination logic cannot output a determination result due to invalid license plate information, output the determination result of a vehicle without a license plate, and generate a corresponding vehicle feature identifier to ensure that all vehicles can obtain a unique identity identifier that can be used in subsequent processes.

[0054] The anomaly detection module is deployed at lane and exit edge nodes to receive identity feature tags and monitor vehicle behavior in real time. Based on the vehicle identity information contained in the received identity feature tags, it combines the identity feature tags with behavioral data to execute preset judgment rules. The core logic of these judgment rules is to determine the benchmark for anomaly detection based on vehicle identity, and then perform event mapping on the benchmark based on behavioral data to output anomaly event detection results that include event type, event level, and vehicle identity information, thus completing intelligent event diagnosis that associates identity with behavior.

[0055] Specifically, the anomaly detection module includes a behavior monitoring unit, a rule engine unit, and an event detection unit. The behavior monitoring unit acquires behavior monitoring data representing the vehicle's operating status in real time. The rule engine unit stores detection rules that associate vehicle identity information, behavior monitoring data, and abnormal events. The association is represented by an event mapping mechanism with vehicle identity information as the primary index and behavior monitoring data as the secondary index. The event detection unit receives identity feature tags and, based on the behavior monitoring data and the vehicle identity information in the identity feature tags, parses the vehicle identity information from the identity feature tags. It then determines the basic sensitivity level using the identity grading rules in the rule engine unit. The basic sensitivity level is combined with the real-time behavior monitoring data and analyzed by the event mapping rules in the rule engine unit to generate the corresponding event type and level. Finally, the event type, event level, and original vehicle identity information are packaged and output as the anomaly event detection result.

[0056] Furthermore, the judgment rules include identity classification rules and mapping rules; identity classification rules are used to determine the basic sensitivity level of a vehicle in anomaly judgment based on its identity information; mapping rules are used to map and generate corresponding abnormal event types and event levels based on the combination of basic sensitivity level and behavior monitoring data. This two-level rule structure forces vehicle identity to be the primary logical input for anomaly judgment.

[0057] For example: A vehicle enters a lane inside the parking lot; the anomaly detection module deployed in the lane continues to operate; the behavior monitoring unit analyzes the video stream and uses a multi-target tracking algorithm based on inter-frame difference and Kalman filter to monitor vehicle behavior; the multi-target tracking algorithm detects moving targets by calculating continuous inter-frame pixel changes and uses Kalman filter to predict and update target trajectories; if the algorithm calculates that a vehicle target is within a preset exit payment area geofence, the coordinate range of the exit payment area geofence is measured and set by parking lot engineers based on the actual lane size and the deployment location of the payment equipment, the speed drops to below 0.2 m / s, and a continuous timer begins; the speed threshold of 0.2 m / s is determined based on the walking speed of pedestrians and the slow driving speed of vehicles in the parking lot, used to distinguish between stationary stagnation and slow passage; at the same time, the event determination unit of this node has received the vehicle's identity feature tag through the edge network; when the behavior monitoring unit reports a vehicle stagnation event, and the duration t reaches 20 seconds, the event determination unit is triggered to perform rule matching;

[0058] The event determination unit first parses the vehicle identity information from the tags and then queries the pre-stored identity classification rules in the rule engine unit. The identity classification rule is represented as a function mapping S=F(C), where: C represents the vehicle identity category; F represents the classification function; and S represents the basic sensitivity level, with lower values ​​indicating lower sensitivity. Therefore, the current vehicle's S=F=1. The event determination unit combines the basic sensitivity level S=1 with specific real-time behavior monitoring data: {Behavior type: Detention, Duration t: 20 seconds, Area: Exit toll payment area}. This real-time behavior monitoring data combination is then fed into the event mapping rules of the rule engine unit for matching. The event mapping rules are in the form of a decision tree or rule table.

[0059] For example, the event mapping rule is described as follows: When the behavior type is determined to be a delay and the occurrence area is located in the exit payment area, the rule engine will perform differentiated event mapping based on the determined basic sensitivity level (S) and the specific delay duration (t): If the vehicle's basic sensitivity level is 1 (corresponding to a low-sensitivity identity, such as a monthly rental car), and the delay time reaches or exceeds 30 seconds, then the behavior is mapped to a payment reminder event and assigned a level 1 (low-level reminder); if the vehicle's basic sensitivity level is 3 (corresponding to a high-sensitivity identity, such as a temporary vehicle), and the delay time reaches or exceeds 15 seconds, then the behavior is mapped to an exit congestion risk event and assigned a level 3 (higher-level alarm); Since S=1 and t=20 seconds, the threshold of 30 seconds has not been reached, and no abnormal event output is generated in this judgment;

[0060] In some embodiments, if the behavior monitoring unit reports that a temporary vehicle (S=3) has been stuck in the same area for 18 seconds, then according to the event mapping rules, the event type is mapped to "Exit Congestion Risk" and the event level is 3; the event judgment unit outputs a structured abnormal event judgment result {Event Type: Exit Congestion Risk, Event Level: 3, Identity Information: Temporary Vehicle, Identity Feature Tag: T_20250001_002}; this is only an example and is not a limitation.

[0061] The control execution module is used to receive the abnormal event judgment results output by the abnormal judgment module, and dynamically match the linkage control strategy based on the abnormal event judgment results. The matching process includes: determining the basic control action according to the event type and level, making differentiated adjustments to the execution parameters of the basic control action according to the vehicle identity information carried in the abnormal event judgment results, and controlling the barrier gate, display screen and voice broadcasting equipment to perform corresponding operations.

[0062] Specifically, the control execution module includes a strategy parsing unit, an equipment control unit, and a feedback unit. The strategy parsing unit receives and parses the abnormal event judgment results, and matches the corresponding linkage control instruction set from the pre-set strategy library based on the event type, event level, and vehicle identity information contained therein. The event and strategy mapping table in the strategy library is designed by parking lot operation experts in combination with safety specifications and user experience requirements, and is optimized with reference to historical abnormal event handling effect data. The equipment control unit is used to convert the linkage control instruction set into drive signals and send them to the corresponding barrier gate, display screen, and voice broadcasting equipment. The feedback unit is used to monitor the execution status of each controlled device after sending the drive signal and synchronize the status feedback information to the edge collaboration module.

[0063] Furthermore, the strategy library contains a pre-set event and strategy mapping table and an identity and strategy weight table. The event and strategy mapping table defines the basic control instruction sequence corresponding to different event types and event level combinations. The identity and strategy weight table defines the priority adjustment parameters or execution parameter correction values ​​of control instructions based on different vehicle identity information, providing differentiated control parameters for vehicles with different identities under the same event. The parameters of the identity and strategy weight table are set by the parking lot management according to the customer level service agreement, and are calibrated in conjunction with historical service satisfaction data. When the strategy parsing unit matches the linkage control instruction set, it determines the basic instruction sequence according to the event and strategy mapping table, and corrects the parameters of the basic instruction sequence according to the identity and strategy weight table to generate the linkage control instruction set.

[0064] For example: The strategy parsing unit of the control execution module receives the above abnormal event judgment result; based on the event type as exit congestion risk and level 3, it queries the preset event and strategy mapping table; the event and strategy mapping table stipulates that for this event, the basic control instruction sequence is: [turn on the lane red warning light, display warning information on the exit screen, and broadcast traffic guidance voice]; the strategy parsing unit queries the identity and strategy weight table based on the vehicle identity information; the identity and strategy weight table stipulates that for temporary vehicles, a temporary vehicle identifier must be added to the displayed information, and the repetition number parameter n of the voice broadcast is adjusted from the default 1 time to 2 times; the strategy parsing unit corrects the basic instruction sequence and generates the final linkage control instruction set: [turn on the lane red warning light, display on the exit screen: temporary vehicle - do not stay for a long time, broadcast voice (repeated 2 times): please leave as soon as possible]. The device control unit converts the instruction set into specific device protocol instructions; for example, sending a command to turn on the red light to the lane warning light controller via the Modbus RTU protocol; controlling the output display screen to update the displayed content via serial port commands; triggering the voice broadcaster to play a specified audio file via an audio cable or network protocol; after the instruction is sent, the feedback unit reads the return status register or confirmation message of each device to confirm successful execution, and synchronizes the status feedback information of the executed instruction to the network through the edge collaboration module.

[0065] The report generation module is used to query vehicle entry records, vehicle identity information and behavior data in conjunction with the abnormal event judgment results output by the abnormal judgment module, so as to build a full-link event view from identity judgment, abnormal occurrence to control execution, and generate a report containing the complete event chain locally at the edge node.

[0066] Specifically, the report generation module includes an event association unit, a data retrieval unit, and a report assembly unit. The event association unit is used to extract key index information from the abnormal event judgment results output by the abnormal judgment module. The data retrieval unit is used to retrieve associated vehicle entry records, vehicle identity information, and historical behavior data from local storage based on the key index information. The report assembly unit is used to integrate the abnormal event judgment results and the retrieved associated data in chronological and logical order, and generate a structured and complete event report locally on the edge node. This report fully records the closed-loop event flow from identity feature binding to final control execution.

[0067] For example, when an abnormal event determination result is generated, the event association unit of the report generation module is triggered; it extracts the identity feature tag T_20250001_002 from the determination result as key index information; the data retrieval unit uses the identity feature tag as a clue to perform retrieval: it searches for the vehicle's entry record in the local log, such as timestamps and thumbnails of the entrance capture image; it initiates a query request to the entrance node through the edge collaborative network to obtain the complete identity feature binding raw data corresponding to the identity feature tag, including raw image features and determination logs; it extracts the historical behavior data sequence of the vehicle from entry to the triggering of the abnormality from the behavior monitoring unit database of this node, such as the position coordinates and speed per second; the report assembly unit integrates all the collected data in chronological order: entry time - identity binding - in-field movement trajectory - exit area lingering behavior - abnormal determination triggering - issued control commands - equipment execution feedback, and generates a structured report file, which is stored in the non-volatile memory of this edge node.

[0068] In some embodiments, when the communication management unit detects an interruption in all connections with external cloud services, the edge collaboration module immediately switches to offline autonomous mode. In this mode, the data routing unit stops forwarding any data to the cloud, but ensures normal TCP / IP network communication between local nodes. The feature analysis module, anomaly judgment module, control execution module, and report generation module will rely on their locally cached vehicle archives, judgment rule base, policy base, and locally stored historical data to operate. The entire core business closed loop continues to run in the parking lot's internal network, unaffected by external network interruptions. The locally generated structured reports will be asynchronously and in batches uploaded to the cloud by the edge collaboration module after the network is restored.

[0069] In some embodiments, the behavior monitoring unit deployed in the lane area operates continuously; by analyzing video streams and utilizing background modeling and foreground extraction algorithms, such as Gaussian mixture models, it stably detects moving vehicles; combined with deep learning-driven behavior recognition models, such as action classifiers based on 3D convolutional neural networks, and trajectory analysis algorithms, it generates multi-dimensional behavior monitoring data in real time; in this embodiment, the behavior monitoring unit successively identifies two behavior sequences of the vehicle in a short period of time: Behavior 1: by comparing the vehicle's movement direction with the lane's preset direction vector, the calculated directional angle continuously exceeds a threshold and is determined to be driving in the wrong direction; Behavior 2: by tracking the displacement of the bounding box to calculate the instantaneous speed, it continuously exceeds the speed limit standard in the parking lot and is determined to be speeding;

[0070] The event determination unit receives the vehicle's identity feature tag through the edge network and parses it to determine that the vehicle's identity information is a temporary vehicle. According to the preset identity classification rules, its basic sensitivity level S is determined to be level 3. The event determination unit combines the identity level S=3, behavior 1, behavior 2 and key time context information and submits it to the rule engine unit for comprehensive determination.

[0071] The rule engine unit pre-stores judgment rules applicable to such complex scenarios. The judgment rules are composite rules combining multiple conditions and temporal correlations. For example, if a vehicle's basic sensitivity level S≥2, and within a continuous time window T, it triggers either wrong-way driving or speeding, the system should determine that the vehicle's behavior constitutes a high-risk driving abnormal event and automatically assign it the highest event level of 5, while also marking that a full-field linkage response must be executed immediately. The judgment rules include identity conditions, behavior combination conditions, and temporal conditions. Only when all conditions are met will a specific event conclusion be mapped.

[0072] Since the current vehicle meets all the above conditions, the rule engine unit executes the judgment rule and outputs the judgment result: the event type is high-risk driving and the event level is 5. The event judgment unit packages this result with the vehicle identity information to form an abnormal event judgment result and sends it to the control execution module. The strategy parsing unit queries the event and strategy mapping table based on the high-risk driving event and the highest level 5. The basic strategies obtained include: continuously flashing red warning lights in the lane where the vehicle is located and adjacent lanes, publishing dynamic warning graphics and text on all public information displays in the venue, and playing safety reminder voice messages in a loop through the venue's broadcast system. Based on the vehicle's identity, the strategy parsing unit queries the identity and strategy weight table, specifically notes the temporary vehicle in the displayed information, and increases the emergency tone parameter of the broadcast voice message. The equipment control unit executes these modified instructions to achieve warning linkage throughout the venue.

[0073] The report generation module starts synchronously, and the generated report includes: the occurrence time, duration and specific location coordinates of behavior 1; the occurrence time, peak speed and occurrence segment of behavior 2, the time interval between the two and the proof that the timing condition T in the rule is met; the specific process and logical output of the rule engine matching the judgment rule; and the set of full-field linkage control instructions and execution feedback triggered by this.

[0074] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A parking lot linkage control system based on edge computing, characterized in that: The system includes: an edge collaboration module, a feature analysis module, an anomaly detection module, a control execution module, and a report generation module; The edge collaboration module is used to construct a distributed collaboration network connecting the edge nodes of the parking lot entrance, lanes and exit, and to synchronize vehicle status data and control commands within the distributed collaboration network. The edge collaboration module includes a network construction unit, a communication management unit, and a data routing unit. The network construction unit constructs a physical communication topology connecting the parking lot entrance, lanes, and exit edge nodes, including redundant communication links. The communication management unit monitors the connectivity of the redundant communication links and automatically switches to a backup communication link when the primary link fails, maintaining the connectivity of the distributed collaboration network. The data routing unit performs targeted synchronization of vehicle status data and control commands within the distributed collaboration network based on preset address mappings and message types. The vehicle status data includes identity feature tags and vehicle behavior monitoring data, and the control commands include gate control commands and display / broadcast commands. When the distributed collaboration network and cloud services are interrupted, the edge collaboration module maintains local collaboration and decision-making among edge nodes based on vehicle status data. The feature analysis module is deployed at the edge node of the parking lot entrance to collect vehicle features in real time, perform logical comparison based on vehicle files, bind the judgment result with the vehicle features, generate a lightweight identity feature label, and synchronize it to the relevant lane and exit edge nodes through the edge collaboration module. The anomaly detection module is deployed at lane and exit edge nodes to receive the identity feature tags, monitor vehicle behavior in real time, combine the identity feature tags and behavior data to execute preset detection rules, and output anomaly event detection results including event type, event level and vehicle identity information. The control execution module is used to receive the abnormal event judgment result output by the abnormal judgment module and dynamically match the linkage control strategy. The report generation module is used to query vehicle entry records, vehicle identity information, and behavioral data in conjunction with the abnormal event determination results output by the abnormal event determination module, and generate a report containing a complete event chain.

2. The parking lot linkage control system based on edge computing according to claim 1, characterized in that: The feature analysis module includes a feature acquisition unit, an identity determination unit, and a tag generation synchronization unit; The feature acquisition unit is used to acquire vehicle feature data, including vehicle appearance images and license plate information, through sensors deployed at the entrance of the parking lot. The identity determination unit is used to logically compare the vehicle feature data with the vehicle archive database, and determine the vehicle identity category as a pre-stored vehicle, a temporary vehicle, or a vehicle without a license plate based on the comparison result. The tag generation and synchronization unit is used to associate and encapsulate the vehicle identity category determination result with the vehicle feature data, generate the identity feature tag, and call the edge collaboration module to synchronize the identity feature tag to the lane and exit edge nodes.

3. The parking lot linkage control system based on edge computing according to claim 2, characterized in that: The identity determination unit includes a first determination logic based on license plate matching and a second determination logic based on vehicle appearance features; The first determination logic is used to output the determination result of pre-stored vehicles or temporary vehicles based on the matching of license plate information in the vehicle feature data with the vehicle archive database. The second determination logic is used to output the determination result of the unlicensed vehicle and generate the corresponding vehicle feature identifier when the first determination logic cannot output the determination result due to invalid license plate information.

4. The parking lot linkage control system based on edge computing according to claim 1, characterized in that: The anomaly detection module includes a behavior monitoring unit, a rule engine unit, and an event detection unit; The behavior monitoring unit is used to acquire behavior monitoring data that characterizes the vehicle's operating status in real time; The rule engine unit is used to store judgment rules that associate vehicle identity information, behavior monitoring data and abnormal events; The event determination unit is used to receive the identity feature tag, and based on the behavior monitoring data and the vehicle identity information in the identity feature tag, call the rule engine unit to perform determination and output the abnormal event determination result.

5. A parking lot linkage control system based on edge computing according to claim 4, characterized in that: The determination rules include identity classification rules and mapping rules; the identity classification rules are used to determine the basic sensitivity level of a vehicle in anomaly determination based on its identity information. The mapping rules are used to map and generate corresponding abnormal event types and event levels based on the combination of the basic sensitivity level and behavior monitoring data.

6. The parking lot linkage control system based on edge computing according to claim 1, characterized in that: The control execution module includes a strategy parsing unit, a device control unit, and a feedback unit; The strategy parsing unit is used to receive and parse the abnormal event judgment result, and match the corresponding linkage control instruction set from the preset strategy library according to the event type, event level and vehicle identity information contained therein. The equipment control unit is used to convert the linkage control instruction set into drive signals and send them to the corresponding barrier gate, display screen and voice broadcasting device; The feedback unit is used to monitor the execution status of each controlled device after sending the drive signal, and synchronize the status feedback information to the edge collaboration module.

7. A parking lot linkage control system based on edge computing according to claim 6, characterized in that: The policy library contains a pre-set event and policy mapping table and an identity and policy weight table. The event and policy mapping table is used to define the basic control instruction sequence corresponding to different combinations of event types and event levels; The identity and strategy weight table is used to define the priority adjustment parameters or execution parameter correction values ​​of control commands based on different vehicle identity information. When the strategy parsing unit matches the linkage control instruction set, it determines the basic instruction sequence according to the event and strategy mapping table, and modifies the parameters of the basic instruction sequence according to the identity and strategy weight table to generate the linkage control instruction set.

8. A parking lot linkage control system based on edge computing according to claim 1, characterized in that: The report generation module includes an event association unit, a data retrieval unit, and a report compilation unit; The event association unit is used to respond to the abnormal event determination result output by the abnormal event determination module and extract key index information from the abnormal event determination result; The data retrieval unit is used to retrieve associated vehicle entry records, vehicle identity information, and historical behavior data from local storage based on the key index information. The report compilation unit is used to integrate the abnormal event determination results with the retrieved related data in chronological and logical order, and generate a structured complete event report locally on the edge node.

Images