Intelligent parking lot management system based on narrowband internet of things

Abstract

The application relates to the technical field of intelligent parking management, in particular to an intelligent parking lot management system based on narrowband Internet of Things, which comprises the following steps: collecting time sequence and identity data of vehicle parking by using a narrowband Internet of Things sensor node group, and forming standardized parking feature information through base station fusion processing. A parking management server uses a behavior analysis model to identify parking rules and abnormalities from the information, and generates a vector depicting the behavior characteristics of the vehicle. Based on the vector, a dynamic parking space allocation model performs optimization calculation, outputs a real-time parking space allocation scheme and a guidance strategy, and finally forms control instructions for a guidance screen and a gate. The system realizes intelligent management from monitoring the parking space state to analyzing the driving behavior and then dynamically optimizing the parking space resources, and improves the parking space utilization efficiency and the parking experience.

Description

Technical Field

[0001] This invention relates to the field of intelligent parking management technology, and in particular to an intelligent parking management system based on narrowband Internet of Things. Background Technology

[0002] Existing parking management systems generally use single technologies such as geomagnetic, ultrasonic, or video recognition to detect parking space status. These systems typically only provide binary status information indicating whether a parking space is occupied or vacant, and some systems can identify vehicles by combining license plate recognition. Their management logic focuses on responding to instantaneous states, and the data in this mode is isolated point information, lacking continuous and time-sequential recording of the entire process of a vehicle entering, parking, and leaving the parking lot.

[0003] The shortcomings of existing technical solutions lie in their limited data dimensions and rudimentary management. Because they cannot obtain long-term parking behavior data for the same vehicle, the system cannot analyze drivers' parking habits, frequent parking times, and areas. This makes parking allocation and guidance strategies static or reactive, unable to predict and personalize based on users' historical behavior patterns and real-time needs. Parking lot managers struggle to identify abnormal vehicles occupying spaces for extended periods and cannot differentiate between high-frequency turnover spaces and long-term rental spaces for optimized management. This results in room for improvement in the overall utilization efficiency of parking resources, and the user experience is limited to basic space-finding functions.

[0004] There is a need for a system that can collect vehicle parking data continuously and at low cost throughout the entire process, and use this data for in-depth analysis, so as to realize a system that can move from simple status monitoring to intelligent behavior management and dynamic resource optimization decision-making. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing an intelligent parking management system based on narrowband Internet of Things.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an intelligent parking management system based on narrowband Internet of Things, comprising: A narrowband IoT sensor node group deployed in various areas of the parking lot is used to collect a set of vehicle status monitoring data, which includes parking space occupancy status sequence, vehicle parking time sequence and vehicle identification data. Narrowband IoT base station is used to perform multi-source data fusion processing on the vehicle status monitoring data set to generate standardized parking feature information, which includes parking space number with unified timestamp, vehicle identifier and parking duration. The parking management server is used to call a pre-set parking behavior analysis model to perform pattern recognition processing on the standardized parking feature information and generate a parking behavior feature vector. The parking behavior feature vector includes parking pattern, abnormal parking identifier and parking space utilization index. The parking management server is also used to input the parking behavior feature vector into the dynamic parking space allocation model for optimization calculation and processing, and generate a real-time parking space allocation scheme and parking space guidance strategy. The parking management server is also used to generate a parking lot control instruction set based on the real-time parking space allocation scheme and parking space guidance strategy. The parking lot control instruction set includes guidance screen display update instructions and gate control instructions.

[0007] As a further aspect of the present invention, the narrowband IoT base station performs multi-source data fusion processing on the vehicle status monitoring data set to generate standardized parking feature information, including: Extract the time point of state change and the duration of state for each parking space from the parking space occupancy state sequence, and convert the original state data into standard state code through the state parser; The arrival and departure times of vehicles are analyzed from the vehicle parking time sequence to calculate the actual parking duration; Vehicle feature information is extracted from the vehicle identification data and mapped to a unique vehicle identifier. Establish a time coordinate system based on parking lot operating time, and align the timestamps of the standard status code, actual parking duration, and unique vehicle identifier to the time coordinate system. All aligned data are integrated in chronological order to generate standardized parking feature information, including parking space number, vehicle identifier, parking duration, and timestamp.

[0008] As a further aspect of the present invention, the parking management server invokes a pre-set parking behavior analysis model to perform pattern recognition processing on the standardized parking feature information, generating a parking behavior feature vector, including: The standardized parking feature information is sliced ​​along the time dimension to form a subset of parking records within multiple statistical periods; For each subset of parking records, perform the following processing steps: Identify the distribution pattern of vehicle parking duration in the parking record subset; when the distribution pattern conforms to a preset typical pattern, assign a parking pattern pattern identifier to the parking record subset. Detect abnormal parking events in the parking record subset, construct an abnormal event feature map, and extract the abnormal parking identifier from the abnormal event feature map. The abnormal parking identifier includes an abnormality type code and an abnormality degree value. Analyze the spatiotemporal distribution characteristics of parking space usage in the parking record subset, and calculate the parking space utilization rate index; The processing results of all statistical periods are summarized and analyzed to generate the parking behavior feature vector, which includes parking pattern, abnormal parking indicators, and parking space utilization rate indicators.

[0009] As a further aspect of the present invention, the parking management server inputs the parking behavior feature vector into a dynamic parking space allocation model for optimization calculation, generating a real-time parking space allocation scheme and a parking space guidance strategy, including: The parking pattern is matched with historical allocation schemes in the parking space allocation knowledge base to determine the basic allocation scheme score; The abnormal parking identifier is input into the conflict resolution sub-model, which is constructed based on the constraint satisfaction algorithm to predict potential parking space usage conflicts and output the conflict resolution priority. The parking space utilization rate index is compared with the parking space utilization rate optimization objective function to obtain the utilization rate optimization suggestion value; The comprehensive optimization value is calculated by integrating the basic allocation scheme score, the conflict resolution priority, and the utilization optimization suggestion value using a multi-objective optimization algorithm. The real-time parking space allocation scheme is determined based on the numerical range of the comprehensive optimization value. Based on the distribution of conflict resolution priorities and the current parking space status, the parking space guidance strategy is calculated. The parking space guidance strategy includes guidance path planning and guidance information release timing.

[0010] As a further aspect of the present invention, the parking management server generates a set of parking lot control instructions based on the real-time parking space allocation scheme and parking space guidance strategy, including: Based on the real-time parking space allocation scheme, the guidance strategy library is queried to determine the basic attributes of the guidance information, which include guidance level, guidance content, and target display device. Based on the parking guidance strategy, the scope and path of the guidance information are dynamically adjusted, and a guidance screen display update instruction is generated. The guidance screen display update instruction includes updating the guidance screen, display content, and update priority. Based on the real-time parking space allocation scheme and parking space guidance strategy, a matching set of basic control rules is selected from the gate control rule base; Based on the abnormal parking identifier, the parameters of the basic control rule set are optimized to generate a barrier gate control command. The barrier gate control command includes barrier gate opening condition judgment rules and opening duration control parameters. The guidance screen display update command and the barrier gate control command are packaged into the parking lot control command set.

[0011] As a further aspect of the present invention, the step of optimizing the parameters of the basic control rule set based on the abnormal parking identifier to generate the barrier gate control command includes: The vehicle identification and parking space number involved in the abnormal parking event are parsed from the abnormal parking sign. For vehicles with abnormal parking records, an enhanced verification rule instruction is generated. This enhanced verification rule instruction is used to add a secondary identity verification process for the vehicle in the barrier gate control system. For the parking space number involved in the abnormal parking event, a status monitoring rule instruction is generated. The status monitoring rule instruction is used to increase the frequency of status monitoring of parking spaces in the parking management system and set abnormal status warning thresholds. The effective time of the rules is adjusted according to the parking pattern. For high-frequency usage periods, a lenient control strategy is adopted; for low-frequency usage periods, a strict control strategy is adopted. The enhanced verification rule instructions, the status monitoring rule instructions, and the rule effective time period configuration are integrated to form specific and executable gate control instructions.

[0012] As a further aspect of the present invention, the step of identifying the distribution pattern of vehicle parking duration in the parking record subset, and assigning a parking pattern pattern identifier to the parking record subset when the distribution pattern conforms to a preset typical pattern, includes: Access the parking pattern library, which stores various typical parking duration distribution patterns such as temporary parking pattern, long-term parking pattern, peak gathering pattern, and uniform distribution pattern; The distribution of vehicle parking duration in the parking record subset is compared with the distribution similarity of each typical parking duration distribution pattern in the parking pattern library. Calculate the matching degree between the subset of parking records and each typical parking duration distribution pattern; Typical parking duration distribution patterns with matching degrees exceeding a threshold are selected, and their corresponding parking pattern patterns are used as candidate identifiers. If multiple candidate identifiers exist, the final parking pattern identifier is determined by comprehensively judging the time span of the parking record subset, the vehicle type distribution, and the parking space utilization rate index.

[0013] As a further aspect of the present invention, the abnormal parking identifier is input into a conflict resolution sub-model. This conflict resolution sub-model is constructed based on a constraint satisfaction algorithm and is used to predict potential parking space usage conflicts and output conflict resolution priorities, including: The abnormal parking identifier is converted into conflict event data, where nodes in the conflict event data represent parking spaces and edges represent observed usage conflict relationships. Each conflict event in the conflict event data is assigned an initial importance attribute, which is determined based on the urgency, scope of impact, and historical resolution effectiveness of the conflict event; The conflict resolution sub-model is used to perform iterative analysis and calculation on the conflict event data. In each iteration, the importance of the conflict event will propagate along the edge to its associated conflict event. The strength of the propagation is determined by the association degree of the edge and the solvability of the received event. After a predetermined number of iterations, each conflict event will converge to a stable resolution priority value, which reflects the urgency of handling the conflict event in the current parking lot state. Extract all conflict events whose resolution priority values ​​exceed the processing threshold, along with their corresponding priority values, and output them as the conflict resolution priority.

[0014] As a further aspect of the present invention, the step of calculating the comprehensive optimization value using a multi-objective optimization algorithm includes: Establish a multi-objective optimization function that includes the objectives of maximizing parking space utilization, minimizing user waiting time, and maximizing parking lot revenue. Each optimization objective is assigned a weight coefficient, which is dynamically adjusted according to the parking lot's operating hours. The basic allocation scheme score is mapped to the input parameters for the goal of maximizing parking space utilization; The conflict resolution priority is converted into a constraint that minimizes user wait time; The proposed utilization rate optimization value is used as the benchmark value for maximizing parking lot revenue. A multi-objective genetic algorithm is used to perform parallel optimization calculations on three optimization objectives, and to find the Pareto optimal solution set under the premise of satisfying various constraints. The solution with the highest overall satisfaction is selected from the Pareto optimal solution set, and its corresponding objective function value is used as the overall optimized value.

[0015] As a further aspect of the present invention, the system further includes: The system continuously collects execution feedback data of the parking lot control command set, which includes guidance screen status feedback, barrier gate execution status records, and actual vehicle parking status. The execution feedback data is correlated with the original parking behavior feature vector and the real-time parking space allocation scheme; The effectiveness of the parking lot control instruction set is evaluated based on the correlation analysis results, and a management effectiveness evaluation report is generated. Based on the management effectiveness evaluation report, the parameters of the dynamic parking space allocation model and the strategies of the guidance strategy library are dynamically optimized and updated.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: By deploying a cluster of narrowband IoT sensor nodes, the system can continuously collect and correlate parking space status change sequences, precise parking durations, and vehicle identification information. The low power consumption and wide coverage of narrowband IoT make large-scale, long-term, and fine-grained data collection economically feasible. This multi-source data, after being fused and processed by base stations, forms standardized parking characteristic information with a unified timestamp. This allows the system to establish a complete digital profile for each vehicle's every parking action, recording the vehicle's detailed trajectory and status changes in both spatial and temporal dimensions, overcoming the limitations of traditional systems that only provide instantaneous snapshots.

[0017] By using a parking behavior analysis model to perform pattern recognition on the aforementioned continuous standardized information, parking patterns of car owners can be extracted from historical data, and abnormal behaviors such as overstaying and unauthorized occupancy can be automatically identified. Simultaneously, the utilization rate of parking spaces in different areas can be accurately calculated. These deep-seated feature vectors are input into a dynamic parking space allocation model for optimization calculations. The model comprehensively considers multiple objectives such as car owner habits and preferences, the current real-time status of parking spaces, anomaly monitoring needs, and overall parking space turnover efficiency to generate personalized parking space allocation schemes and guidance strategies. Parking space allocation has shifted from simple matching based on the nearest geographical distance to intelligent decision-making based on user behavior characteristics and global resource efficiency. Guidance screens and gates execute the results of this optimized decision-making. Managers can implement more targeted management measures based on the parking space utilization rate indicators and anomaly indicators output by the model. Attached Figure Description

[0018] Figure 1 This is a timing diagram of the intelligent parking management system based on narrowband Internet of Things as described in this invention; Figure 2 A flowchart for multi-source data fusion processing; Figure 3 A flowchart for optimizing dynamic parking space allocation calculations; Figure 4 A convergence curve for the iterative assessment of the importance of conflict events; Figure 5 A bar chart comparing the performance of the parking management system before and after optimization. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0021] See Figure 1 The system collects vehicle status monitoring data through a cluster of narrowband IoT sensor nodes deployed in various areas of the parking lot. This data includes parking space occupancy status sequences, vehicle parking time sequences, and vehicle identification data. The narrowband IoT base station receives the data from the sensor nodes and performs multi-source data fusion processing on the vehicle status monitoring data to generate standardized parking characteristic information. This information includes parking space numbers with a unified timestamp, vehicle identifiers, and parking durations. The parking management server connects to the narrowband IoT base station and calls a pre-built parking behavior analysis model to perform pattern recognition processing on the standardized parking characteristic information, generating a parking behavior feature vector. This vector includes parking pattern recognition, abnormal parking indicators, and parking space utilization indicators. The parking management server inputs the parking behavior feature vector into a dynamic parking space allocation model for optimization calculations, generating a real-time parking space allocation scheme and parking guidance strategy. Based on the real-time parking space allocation scheme and parking guidance strategy, the parking management server generates a set of parking lot control instructions, including guidance screen display update instructions and gate control instructions, to achieve intelligent management of the parking lot.

[0022] See Figure 2In one embodiment of the present invention, a narrowband IoT base station performs multi-source data fusion processing on a vehicle status monitoring dataset to generate standardized parking feature information. This processing includes extracting the status change time point and duration of each parking space from the parking space occupancy status sequence, and converting the original status data into a standard status code using a status parser. It also involves parsing the vehicle arrival and departure times from the vehicle parking time sequence to calculate the actual parking duration. Furthermore, it extracts vehicle feature information from vehicle identification data and maps this information to a unique vehicle identifier. A time coordinate system based on parking lot operating time is established, and the timestamps of the standard status code, actual parking duration, and unique vehicle identifier are aligned to this time coordinate system. All aligned data are integrated in chronological order to generate standardized parking feature information containing the parking space number, vehicle identifier, parking duration, and timestamp. The parking management server then calls a pre-set parking behavior analysis model to perform pattern recognition processing on the standardized parking feature information, generating a parking behavior feature vector. This processing includes slicing the standardized parking feature information along a time dimension to form subsets of parking records within multiple statistical periods. For each subset of parking records, the distribution pattern of vehicle parking duration within the subset is identified. When the distribution pattern matches a preset typical pattern, a parking pattern pattern identifier is assigned to the subset. Abnormal parking events are detected within the subset, an abnormal event feature map is constructed, and abnormal parking identifiers are extracted from the feature map. These identifiers include an abnormality type code and an abnormality severity value. The spatiotemporal distribution characteristics of parking space usage within the subset are analyzed, and parking space utilization rate is calculated. The processing results from all statistical periods are summarized and analyzed to generate a parking behavior feature vector containing parking pattern, abnormal parking identifier, and parking space utilization rate. Identifying the distribution pattern of vehicle parking duration within the subset involves accessing a parking pattern database, which stores various typical parking duration distribution patterns, such as temporary parking, long-term parking, peak-hour clustering, and uniform distribution. The distribution similarity of the vehicle parking duration distribution in the subset is then compared with the distribution patterns of each typical parking duration distribution pattern in the parking pattern database. Calculate the matching degree between the parking record subset and each typical parking duration distribution pattern. Select typical parking duration distribution patterns with matching degrees exceeding a threshold, and use their corresponding parking pattern patterns as candidate identifiers. If multiple candidate identifiers exist, determine the final parking pattern pattern identifier based on a comprehensive judgment of the parking record subset's time span, vehicle type distribution, and parking space utilization rate.

[0023] In practical implementation, the intelligent parking management system based on narrowband IoT collects vehicle status monitoring data sets through a group of narrowband IoT sensor nodes deployed in various areas of the parking lot. The vehicle status monitoring data sets include parking space occupancy status sequences, vehicle parking time sequences, and vehicle identification data. The narrowband IoT base station receives the vehicle status monitoring data sets and performs multi-source data fusion processing on the vehicle status monitoring data sets to generate standardized parking feature information. The standardized parking feature information includes parking space numbers with a unified timestamp, vehicle identifiers, and parking duration. The parking management server calls a pre-set parking behavior analysis model to perform pattern recognition processing on the standardized parking feature information to generate parking behavior feature vectors. The parking behavior feature vectors include parking pattern, abnormal parking indicators, and parking space utilization rate indicators. In some embodiments, the narrowband IoT base station performs multi-source data fusion processing on the vehicle status monitoring data set to generate standardized parking feature information. This includes extracting the status change time point and status duration of each parking space from the parking space occupancy status sequence, converting the original status data into a standard status code through a status parser, parsing the vehicle arrival and departure time points from the vehicle parking time sequence, calculating the actual parking duration, extracting vehicle feature information from vehicle identification data, mapping the vehicle feature information to a unique vehicle identifier, establishing a time coordinate system based on the parking lot operating time, aligning the timestamps of the standard status code, actual parking duration, and unique vehicle identifier to the time coordinate system, and integrating all aligned data in chronological order to generate standardized parking feature information containing parking space number, vehicle identifier, parking duration, and timestamp. It is understandable that the parking management server calls a pre-set parking behavior analysis model to perform pattern recognition processing on standardized parking feature information, generating a parking behavior feature vector. This includes slicing the standardized parking feature information in the time dimension to form parking record subsets within multiple statistical periods. For each parking record subset, it identifies the distribution pattern of vehicle parking duration. When the distribution pattern matches a preset typical pattern, it assigns a parking pattern identifier to the parking record subset. It detects abnormal parking events in the parking record subset, constructs an abnormal event feature map, extracts abnormal parking identifiers from the abnormal event feature map (the abnormal parking identifier includes an abnormality type code and an abnormality degree value), analyzes the spatiotemporal distribution characteristics of parking space usage in the parking record subset, calculates the parking space utilization rate index, and summarizes and analyzes the processing results of all statistical periods to generate a parking behavior feature vector containing parking pattern, abnormal parking identifier, and parking space utilization rate index.

[0024] Optionally, the distribution pattern of vehicle parking duration within the parking record subset is identified. When the distribution pattern conforms to a preset typical pattern, a parking pattern pattern identifier is assigned to the parking record subset. This includes accessing a parking pattern pattern library, which stores various typical parking duration distribution patterns such as temporary parking, long-term parking, peak-hour clustering, and uniform distribution. The distribution similarity of the vehicle parking duration distribution of the parking record subset is compared with each typical parking duration distribution pattern in the parking pattern pattern library. The matching degree between the parking record subset and each typical parking duration distribution pattern is calculated. Typical parking duration distribution patterns with matching degrees exceeding a threshold are selected, and their corresponding parking pattern patterns are used as candidate identifiers. If multiple candidate identifiers exist, a comprehensive judgment is made based on the time span of the parking record subset, vehicle type distribution, and parking space utilization rate to determine the final parking pattern pattern identifier. In specific implementation, to quantify the distribution similarity comparison process, a similarity calculation formula is used to evaluate the matching degree between the parking record subset and the typical parking duration distribution pattern. The formula is:

[0025] in, Represents the similarity coefficient. Indicates the subset of parking records in the th order. The percentage of vehicles in each time period This represents the distribution pattern of a typical parking duration in the parking pattern library at the [number]th [time]. The percentage of vehicles in each time period This represents the average percentage of vehicles in a subset of parking records. This represents the average percentage of vehicles exhibiting a typical parking duration distribution pattern. The similarity coefficient represents the total number of time intervals. The value ranges from -1 to 1, with values ​​closer to 1 indicating more similar distributions. In some embodiments, by comparing data from example scenarios, a subset of parking records containing 150 records shows two distinct peaks in vehicle parking duration distribution: 8:00 AM to 10:00 AM and 5:00 PM to 7:00 PM. Similarity calculations are performed with the peak clustering pattern distribution, and the similarity coefficient R is 0.88, exceeding the preset threshold of 0.75. Therefore, the peak clustering pattern is selected as a candidate identifier. Furthermore, the time span of this parking record subset covers weekday morning and evening peak hours, private cars account for over 90% of the vehicle type distribution, and the parking space utilization rate is 85%. Based on this comprehensive assessment, the parking pattern is identified as the peak clustering pattern. It can be understood that distribution similarity comparison relies on a predefined mathematical model in the parking pattern library. This library is constructed by analyzing historical parking data and includes typical parking duration distribution patterns under different scenarios. The similarity calculation formula provides an objective measurement basis for pattern recognition, avoiding bias caused by subjective judgment. In the multi-source data fusion process, the state parser converts the original state data in the parking space occupancy state sequence into standard state codes, such as "occupied" being coded as 1 and "idle" as 0. The time point parsing in the vehicle parking time sequence is based on sensor timestamps and logical rules. Vehicle feature information in the vehicle identification data is mapped to a fixed-length unique vehicle identifier using a hash function. The time coordinate system uses the daily operation start point of the parking lot as the zero point, converting all timestamps into offsets relative to the zero point to ensure the consistency of the data's time dimension. In specific implementation, after the standardized parking feature information is generated, the parking behavior analysis model further processes the subset of parking records. Abnormal parking identifiers are extracted based on an abnormal event feature map. The abnormal event feature map uses a clustering algorithm to identify outliers in the subset of parking records, such as records where the parking duration exceeds a preset upper limit or falls below a preset lower limit. The parking space utilization rate index is obtained by calculating the ratio of the total parking space occupancy time to the total duration of the statistical period. Through the above methods, Example 1 fully describes the specific implementation process from data collection to behavior feature vector generation.

[0026] In one embodiment of the present invention, the parking management server inputs parking behavior feature vectors into a dynamic parking space allocation model for optimization calculation, generating a real-time parking space allocation scheme and a parking space guidance strategy. (See also...) Figure 3The process includes matching parking pattern characteristics with historical allocation schemes in a parking space allocation knowledge base to determine a basic allocation scheme score. Abnormal parking identifiers are input into a conflict resolution sub-model, built based on a constraint satisfaction algorithm, to predict potential parking space usage conflicts and output conflict resolution priorities. Parking space utilization rate indicators are compared with a parking space utilization optimization objective function to obtain a utilization rate optimization suggestion value. The basic allocation scheme score, conflict resolution priorities, and utilization rate optimization suggestion value are integrated, and a multi-objective optimization algorithm is used to calculate a comprehensive optimization value. Based on the numerical range of the comprehensive optimization value, a real-time parking space allocation scheme is determined. Based on the distribution of conflict resolution priorities and the current parking lot status, a parking guidance strategy is calculated, which includes guidance path planning and guidance information dissemination timing.

[0027] In practical implementation, the parking management server inputs parking behavior feature vectors into the dynamic parking space allocation model for optimization calculation, generating real-time parking space allocation schemes and parking guidance strategies. The dynamic parking space allocation model's processing includes matching parking pattern characteristics with historical allocation schemes in the parking space allocation knowledge base to determine the basic allocation scheme score; inputting abnormal parking identifiers into the conflict resolution sub-model, which is built based on a constraint satisfaction algorithm to predict potential parking space usage conflicts and output conflict resolution priorities; comparing the parking space utilization rate index with the parking space utilization rate optimization objective function to obtain a utilization rate optimization suggestion value; integrating the basic allocation scheme score, conflict resolution priorities, and utilization rate optimization suggestion value; and using a multi-objective optimization algorithm to calculate the comprehensive optimization value; determining the real-time parking space allocation scheme based on the numerical range of the comprehensive optimization value; and calculating the parking guidance strategy based on the distribution of conflict resolution priorities and the current parking lot parking space status. The parking guidance strategy includes guidance path planning and guidance information release timing. In some embodiments, the determination of the basic allocation scheme score is achieved through a pattern matching algorithm. For example, if the parking pattern in the current parking behavior feature vector is a "peak gathering pattern," and the parking space allocation knowledge base stores multiple historical peak-period parking space allocation schemes and their corresponding operational performance scores, the system retrieves the historical scheme with the highest similarity to the current time and traffic density. After weighting and correcting its performance score based on the current total number of parking spaces and vehicle type, the system outputs it as the basic allocation scheme score. It can be understood that the input to the conflict resolution sub-model is an abnormal parking identifier, which includes an abnormality type code and an abnormality severity value. The conflict resolution sub-model parses each abnormal parking identifier as a potential conflict event. For example, an identifier with an abnormality type code of "overtime occupancy" and an abnormality severity value of "high" may indicate that a parking space will be occupied for a long time, thus conflicting with subsequent reservations or high-frequency usage needs. The conflict resolution sub-model analyzes the correlation between all conflict events based on a constraint satisfaction algorithm and calculates a quantified conflict resolution priority value for each event.

[0028] The utilization rate optimization suggestion value is obtained by comparing the current parking space utilization rate index with a preset parking space utilization rate optimization objective function. This objective function defines the ideal utilization rate curve under different total traffic flow ranges. The system inputs the currently monitored total traffic flow into the function to obtain the corresponding ideal utilization rate value. Then, the current actual parking space utilization rate index is compared with this ideal value, and the difference, after normalization, is output as the utilization rate optimization suggestion value. Integrating the basic allocation scheme score, conflict resolution priority, and utilization rate optimization suggestion value, a multi-objective optimization algorithm is used to calculate the comprehensive optimization value. This process is achieved by establishing a multi-objective optimization function, the formula of which is:

[0029] in, This represents the overall optimization value. Indicates the score of the basic allocation scheme. Indicates the priority of conflict resolution. This indicates the recommended value for utilization optimization. This represents a standardized mapping function for scoring the basic allocation scheme. This represents a benefit transfer function for prioritizing conflict resolution. This represents the contribution function to the utilization optimization recommendation value. , , These represent the dynamic weighting coefficients corresponding to the basic allocation scheme score, conflict resolution priority, and utilization optimization suggestion value, respectively. These dynamic weighting coefficients are dynamically adjusted according to the parking lot's operating hours, such as during peak traffic periods. The value of is increased to emphasize the urgency of resolving the conflict, especially during off-peak hours. The value of is increased to emphasize the benefits of improved utilization. In practical implementation, an example scenario is used to illustrate this. Suppose a parking lot with 200 spaces receives a parking behavior feature vector at 9:00 AM. The basic allocation scheme score is 85 points, the highest priority conflict resolution item is 8.7, and the utilization optimization suggestion value is 92%. Based on the judgment that it is currently the morning rush hour, the system sets the dynamic weight coefficient to . , , After calculation by various functions, the comprehensive optimization value F is 86.4, which falls within the preset "efficient allocation" value range [85, 95], thereby triggering a corresponding real-time parking space allocation scheme.

[0030] In some embodiments, after the real-time parking space allocation scheme is determined, the system calculates a parking guidance strategy based on the distribution of conflict resolution priorities and the current parking lot status. For example, if it is identified that the area with the highest conflict resolution priority is concentrated in zone B, while the current parking space status shows that there are still many empty spaces in zone A, the guidance path planning in the parking guidance strategy will generate the optimal path from the entrance to zone A, avoiding guiding vehicles through zone B. The timing of guidance information release is sorted according to the predicted arrival time of vehicles at the entrance and the path length, ensuring the timeliness and orderliness of information release. It can be understood that the entire process from inputting parking behavior feature vectors to outputting real-time parking space allocation schemes and parking guidance strategies is iterative and continuous. The dynamic parking space allocation model periodically updates the output based on new input data to adapt to the dynamic changes in the parking lot status. Through the above methods, Embodiment 2 fully describes the specific implementation process of the dynamic parking space allocation model performing optimization calculations and generating schemes and strategies.

[0031] In one embodiment of the present invention, the parking management server generates a set of parking lot control instructions based on a real-time parking space allocation scheme and a parking guidance strategy. This generation process includes querying a guidance strategy library based on the real-time parking space allocation scheme to determine the basic attributes of the guidance information, including guidance level, guidance content, and target display device. Based on the parking guidance strategy, the server dynamically adjusts the scope and path of the guidance information's publication, generating a guidance screen display update instruction. This instruction includes updating the guidance screen, display content, and update priority. According to the real-time parking space allocation scheme and the parking guidance strategy, a matching set of basic control rules is selected from the gate control rule library. The basic control rule set is then optimized based on abnormal parking identifiers to generate gate control instructions, which include gate opening condition judgment rules and opening duration control parameters. The guidance screen display update instructions and the gate control instructions are packaged into a parking lot control instruction set. Again, the basic control rule set is optimized based on abnormal parking identifiers to generate gate control instructions. This optimization includes parsing the vehicle identifier and parking space number involved in the abnormal parking event from the abnormal parking identifiers. For vehicle identifiers with abnormal parking records, enhanced verification rule instructions are generated to add a secondary identity verification process to the barrier gate control system. For parking space numbers involved in abnormal parking events, status monitoring rule instructions are generated to increase the frequency of parking space status monitoring in the parking management system and set abnormal status warning thresholds. The effective time periods of the rules are adjusted according to parking pattern characteristics; a lenient control strategy is adopted for high-frequency usage periods, while a strict control strategy is adopted for low-frequency usage periods. The enhanced verification rule instructions, status monitoring rule instructions, and rule effective time period configurations are integrated to form specific and executable barrier gate control instructions.

[0032] In practical implementation, the guidance strategy library stores various preset guidance information templates. Each template is associated with a specific real-time parking space allocation scheme type. For example, when the real-time parking space allocation scheme is "centralized guidance to Area A," the basic attributes of the guidance information determined after querying the guidance strategy library are: guidance level set to "Level 1," guidance content set to "Please proceed to Area A for parking," and target display devices set to the main entrance guidance screen and the sub-screens at each intersection. It can be understood that the scope and path of guidance information dissemination are dynamically adjusted based on the parking guidance strategy. The guidance path planning in the parking guidance strategy determines the physical scope of information dissemination, and the timing of guidance information dissemination determines the time path of information dissemination. For example, if the parking guidance strategy plans to guide newly arriving vehicles to Area C on the third floor, then the scope of the guidance screen update instruction is limited to all guidance screens on the path leading to Area C on the third floor, and the dissemination path updates the content of each screen sequentially from the entrance to the target area. Based on abnormal parking identifiers, the basic control rule set is optimized to generate barrier gate control commands. This optimization includes parsing the vehicle identifier and parking space number involved in the abnormal parking event from the abnormal parking identifiers. For vehicle identifiers with abnormal parking records, enhanced verification rule commands are generated to add a secondary identity verification process for vehicles in the barrier gate control system. For parking space numbers involved in abnormal parking events, status monitoring rule commands are generated to increase the frequency of parking space status monitoring in the parking management system and set abnormal status warning thresholds. The effective time period of the rules is adjusted according to parking pattern. For high-frequency usage periods, a lenient control strategy is adopted; for low-frequency usage periods, a strict control strategy is adopted. The enhanced verification rule commands, status monitoring rule commands, and rule effective time period configurations are integrated to form specific and executable barrier gate control commands. In practical implementation, an example scenario illustrates the parameter optimization process. Assume an abnormal parking record is parsed, with vehicle identifier "V_7F3A", parking space number "B205", anomaly type code "unauthorized occupancy", and anomaly severity value "high". The enhanced verification rule generated for vehicle identifier "V_7F3A" states: "When vehicle 'V_7F3A' attempts to enter again, in addition to license plate recognition, the gate can only be opened after on-duty personnel confirm via video or scan a QR code." Regarding parking space number "B205", the rule further specifies the required verification criteria. The status monitoring rule command generated for "B205" is to "increase the status monitoring frequency of parking space 'B205' from once every 5 minutes to once every 30 seconds, and set an alert to be triggered if the status is continuously occupied for more than 24 hours." At the same time, based on the current parking pattern of "weekday peak mode", the rule's effective period has been adjusted to adopt a lenient control strategy from 7:00 am to 7:00 pm on weekdays, with a default gate opening time of 10 seconds; and a strict control strategy from 7:00 pm to 7:00 am the next day, with a default gate opening time of 5 seconds and a complete verification process required.The specific adjustment of the gate opening duration control parameter can be achieved through a parameter optimization function. This function takes the abnormality level value and parking pattern in the abnormal parking indicator as input variables, and the formula is:

[0033] in, This indicates the adjustment amount of the barrier gate opening time relative to the base time. A positive value means extending the opening time for easier verification, while a negative value means shortening the opening time to speed up passage. This represents the degree of abnormality parsed from the abnormal parking sign. This indicates the currently active parking pattern. and These are the weighting coefficients configured in the system, and the function... Parking patterns are mapped to a coefficient related to the strictness of time-based control. For example, "high-frequency usage periods" are mapped to negative values ​​to encourage fast passage, while "low-frequency usage periods" are mapped to positive values ​​to strengthen control. The calculated adjustment amount is then... The final opening duration control parameter is obtained by adding it to the basic opening duration and then written into the barrier gate control command. It can be understood that the update priority in the guidance screen's update command determines the controller's processing order when multiple update commands arrive simultaneously. The update priority is usually determined by both the guidance level and the timing of guidance information release. For example, "Level 1" guidance commands that need to be released immediately have the highest update priority and will be executed before "Level 2" guidance commands or commands released on a timed basis. Through the above method, Example 3 fully describes the specific implementation process of generating a specific set of parking lot control commands from the real-time parking space allocation scheme and parking space guidance strategy, and optimizing the parameters of the barrier gate control rules.

[0034] In one embodiment of the invention, abnormal parking identifiers are input into a conflict resolution sub-model. This sub-model, constructed based on a constraint satisfaction algorithm, predicts potential parking space usage conflicts and outputs conflict resolution priorities. The process includes converting abnormal parking identifiers into conflict event data, where nodes represent parking spaces and edges represent observed usage conflict relationships. Each conflict event in the conflict event data is assigned an initial importance attribute, determined based on the urgency, scope of impact, and historical resolution effectiveness of the conflict event. The conflict event data is iteratively analyzed and calculated using the conflict resolution sub-model. In each iteration, the importance of a conflict event propagates along edges to its associated conflict events, with the propagation strength determined by the edge's relevance and the solvability of the received event. After a predetermined number of iterations, each conflict event converges to a stable resolution priority value, reflecting the urgency of handling the conflict event in the current parking lot state. Conflict events with resolution priority values ​​exceeding a processing threshold, along with their corresponding priority values, are extracted and output as the conflict resolution priorities.

[0035] A multi-objective optimization algorithm is employed to calculate the comprehensive optimization value. This calculation involves establishing a multi-objective optimization function encompassing three objectives: maximizing parking space utilization, minimizing user waiting time, and maximizing parking lot revenue. Weight coefficients are assigned to each objective, dynamically adjusted according to the parking lot's operating hours. The basic allocation scheme score is mapped to the input parameters of the parking space utilization maximization objective. Conflict resolution priorities are converted into constraints for the user waiting time minimization objective. The utilization optimization recommendation value is used as the benchmark value for the parking lot revenue maximization objective. A multi-objective genetic algorithm is used to perform parallel optimization calculations on the three objectives, searching for a Pareto optimal solution set while satisfying all constraints. The solution with the highest overall satisfaction is selected from the Pareto optimal solution set, and its corresponding objective function value is used as the comprehensive optimization value.

[0036] In practice, abnormal parking indicators are input into a conflict resolution sub-model. This sub-model, built based on a constraint satisfaction algorithm, is used to predict potential parking space usage conflicts and output conflict resolution priorities. This process includes converting abnormal parking indicators into conflict event data. Nodes in the conflict event data represent parking spaces, and edges represent observed usage conflict relationships. Each conflict event in the conflict event data is assigned an initial importance attribute, determined based on the urgency, scope of impact, and historical resolution effectiveness of the conflict event. The conflict resolution sub-model is used to iteratively analyze and calculate the conflict event data. In each iteration, the importance of the conflict event propagates along the edges to its associated conflict events. The propagation strength is determined by the association degree of the edge and the solvability of the received event. After a predetermined number of iterations, each conflict event converges to a stable resolution priority value. The resolution priority value reflects the urgency of handling the conflict event in the current parking lot state. Conflict events with resolution priority values ​​exceeding a processing threshold and their corresponding priority values ​​are extracted and used as the output of the conflict resolution priorities. In some embodiments, the abnormal parking identifier includes an abnormality type code and an abnormality severity value. The conversion process instantiates each abnormal parking identifier as a conflict event. For example, an abnormal event identified as "Parking space B203 reservation time exceeded" is converted into a conflict event with parking space "B203" as the node. This node is connected to future reservation event nodes that also depend on parking space B203 or its adjacent areas. The initial importance attribute is calculated and assigned by querying the historical event database to find the resolution time and the number of parking spaces affected by similar events. It can be understood that the correlation of edges is quantified by analyzing the frequency and causal relationship strength of two conflict events occurring simultaneously in historical data, and the resolvability of the received event is dynamically evaluated based on factors such as the number of spare parking spaces available in the current parking lot resource pool to alleviate the conflict and scheduling flexibility.

[0037] The core iterative computation process of the conflict resolution sub-model follows a specific propagation and convergence rule to simulate the diffusion and stabilization of conflict effects in the network. Its mathematical expression is as follows:

[0038] in, Indicates the first Conflict events after the next iteration Importance value, Indicates the first Conflict events after the next iteration Importance value, It is the attenuation retention factor of the importance of the conflict event itself. It is the global gain coefficient of network propagation effect. Indicates conflict events The set of all other collision events directly connected by edges. Indicates from conflict events Spread to conflict events The propagation strength, which is determined by the edge correlation degree and the event. The type determines this together. Indicates conflict events The solvability factor is inversely proportional to the solvability of the event; that is, the greater the difficulty of solving the problem, the higher the factor value. This is achieved after a predetermined number of attempts. After several iterations, the network is considered to have converged when the change in the importance value of all conflict events is less than the stability threshold. At this point, the importance value of each conflict event... This involves resolving the priority value. The process is illustrated using input data from an example scenario. Assume that five related conflict events are derived from the current abnormal parking flag; their initial data and relationships are shown in Table 1.

[0039] Table 1: Conflict Event Data Table

[0040] In practical implementation, system configuration parameters , Preset number of iterations The stability threshold is 0.01. The propagation strength value is calculated for each pair of related events according to the edge association rule, and the solvability factor value is evaluated for each conflict event based on the current vacancy status of each area of ​​the parking lot. For example, if the area involved in event C1 has no spare parking space, its solvability factor value is set to be high. The initial data is substituted into the iterative formula for calculation. After 7 iterations, the network reaches a stable state. Finally, the resolution priority values ​​of C1, C4, C2, C3, and C5 are calculated to be 8.9, 7.5, 5.8, 4.3, and 2.9. If the processing threshold is set to 5.0, the output conflict resolution priority includes events C1, C4, and C2 and their corresponding values. A multi-objective optimization algorithm is used to calculate the comprehensive optimization value. This calculation includes establishing a multi-objective optimization function that includes the objectives of maximizing parking space utilization, minimizing user waiting time, and maximizing parking lot revenue. Weight coefficients are assigned to each optimization objective, and these weight coefficients are dynamically adjusted according to the parking lot's operating hours. The basic allocation scheme score is mapped to the input parameters of the parking space utilization maximization objective. The conflict resolution priority is converted into the constraint condition of the user waiting time minimization objective. The utilization optimization suggestion value is used as the benchmark value of the parking lot revenue maximization objective. The three optimization objectives are optimized in parallel using a multi-objective genetic algorithm. Under the premise of satisfying all constraints, a Pareto optimal solution set is found. The solution with the highest overall satisfaction is selected from the Pareto optimal solution set, and its corresponding objective function value is used as the comprehensive optimization value.

[0041] In some embodiments, the constraint construction for minimizing user waiting time directly utilizes the output of conflict resolution priority. For example, parking areas involving events with resolution priority values ​​higher than a threshold are set as high-latency sensitive areas in the optimization model, and stricter maximum waiting time upper limits are imposed on vehicle allocation schemes in these areas. The search for the Pareto optimal solution set is achieved through a multi-objective genetic algorithm. The algorithm maintains a population of candidate solutions, evolves through selection, crossover, and mutation operations, and uses congestion ranking to maintain the distribution of solutions in the objective space. The selection of the solution with the highest overall satisfaction is based on a preset satisfaction evaluation function, which comprehensively considers the parking lot management's preference weights for different objectives and the position of each solution on the Pareto front. It can be understood that the entire processing flow described in Embodiment 4 is executed sequentially. The conflict resolution priority output by the conflict resolution sub-model serves as one of the key inputs and is used by the subsequent multi-objective optimization algorithm. The comprehensive optimization value finally generated by the multi-objective optimization algorithm is used to determine the final real-time parking space allocation scheme. In the above manner, Embodiment 4 fully describes the internal iterative calculation details of the conflict resolution sub-model and the specific implementation process of how the multi-objective optimization algorithm uses its output for calculation.

[0042] See Figure 4 This is an iterative convergence curve of conflict event importance, fully demonstrating the calculation process of the conflict resolution sub-model in the parking management system. The black dashed line in the graph represents the processing threshold (5.0). Only the final importance of C1, C4, and C2 exceeds this threshold and are marked by the system as conflict events requiring priority processing. This result perfectly matches the conflict resolution priority output in the implementation example. After the 6th iteration, the importance change of all events is less than 0.01, indicating that the system has reached a stable convergence state, verifying the computational efficiency and reliability of the conflict resolution sub-model. The graph transforms abstract conflict events into quantifiable importance values, providing the parking management system with a clear priority ranking for conflict processing and avoiding the subjectivity of manual judgment. The stability of the convergence curve and the rationality of the results verify the effectiveness of the propagation rules and iterative algorithm in the conflict resolution sub-model, which can be used to further optimize the model's convergence speed and accuracy.

[0043] In one embodiment of the present invention, the system further includes continuously collecting execution feedback data of the parking lot control command set. This execution feedback data includes guidance screen status feedback, barrier gate execution status records, and actual vehicle parking conditions. The execution feedback data is then correlated with the original parking behavior feature vector and the real-time parking space allocation scheme. The effectiveness of the parking lot control command set is evaluated based on the correlation analysis results, and a management effectiveness evaluation report is generated. Based on the management effectiveness evaluation report, the parameters of the dynamic parking space allocation model and the strategies in the guidance strategy library are dynamically optimized and updated.

[0044] In practical implementation, the intelligent parking management system based on narrowband IoT also includes continuously collecting execution feedback data of the parking control command set. This execution feedback data includes guidance screen status feedback, barrier gate execution status records, and actual vehicle parking status. The parking management server performs correlation analysis on the execution feedback data with the original parking behavior feature vector and real-time parking space allocation scheme. Based on the correlation analysis results, it evaluates the effectiveness of the parking control command set and generates a management effectiveness evaluation report. Based on this report, it dynamically optimizes and updates the parameters of the dynamic parking space allocation model and the strategies in the guidance strategy library. In some embodiments, guidance screen status feedback is reported by each guidance screen controller, including screen number, command received time, successful content refresh status, and a snapshot of the current displayed content. Barrier gate execution status records are provided by the barrier gate controller log, including barrier gate number, time of receiving opening command, actual opening action time, opening duration, and abnormal alarm information. Actual vehicle parking status is re-collected by a narrowband IoT sensor node group after the command execution cycle, forming a new parking space occupancy status sequence and vehicle identification information, which is used for comparison with the expected allocation result of the command. It is understandable that the correlation analysis process establishes data mapping relationships, associating a guide screen update command with the status feedback of the corresponding guide screen, associating a gate control command with the execution status record of the corresponding gate, and associating the target parking space allocated for a specific vehicle or vehicle type in the real-time parking space allocation plan with the parking space where the vehicle is finally parked in the actual parking situation, thereby forming a full-link tracking dataset from the issuance of the command to the on-site execution effect.

[0045] The effectiveness of parking lot control command sets is evaluated based on the full-link tracking dataset generated after correlation analysis. The evaluation process quantifies the deviation between the command execution results and the expected goals. For example, for a command to guide a vehicle to a parking space in Zone A, the effectiveness evaluation needs to calculate multiple dimensions such as whether the vehicle travels along the expected path, whether it successfully parks in a parking space in Zone A, and whether the total time from receiving the guidance information to parking in the space is within the expected range. For a gate control command, the effectiveness evaluation needs to analyze whether the gate opens on time, whether the opening duration meets the parameter settings, and whether any abnormal alarms are triggered. The evaluation results of these dimensions are combined to generate a comprehensive management effectiveness evaluation report. In practical implementation, an example scenario is used to illustrate data comparison and correlation analysis. Assume that during the statistical period from 10:00 AM to 10:30 AM, the system generates a real-time parking allocation plan aimed at evenly guiding newly arriving vehicles to the east and west zones, and issues corresponding guidance screen update instructions and gate control instructions. Continuously collected execution feedback data shows that the guidance screen at the east entrance successfully updated its guidance information, while the guidance screen at the west entrance failed to update immediately due to network latency. Gate execution records show that all vehicles were allowed to pass normally. However, actual vehicle parking data shows that 50 vehicles entered the area during this period, with 38 parking in the east zone and 12 in the west zone, a significant deviation from the expected "even guidance" plan. Correlation analysis links this deviation to the failure of the west zone guidance screen to update, and the management effectiveness evaluation report explicitly points out that the instruction set has a defect in the synchronization of guidance information, leading to uneven parking allocation.

[0046] In some embodiments, the management effectiveness evaluation report adopts a structured format, including the evaluation period, identifiers of the instruction sets involved, numerical values ​​of evaluation indicators for each dimension, descriptions of key deviation events, and preliminary attribution analysis. The numerical values ​​of the evaluation indicators are calculated using a preset evaluation function to quantify the overall effectiveness of the instruction set. The general form of this function is:

[0047] in, This represents the overall effectiveness score of the parking lot control instruction set within the current evaluation period. This represents the total number of evaluation dimensions, such as instruction arrival rate, execution accuracy rate, and target achievement rate. Indicates the first The weighting coefficients of each evaluation dimension, Indicates the first The total number of instructions or events that need to be evaluated under each dimension. Indicates the first The i-th instruction or expected goal under each dimension Indicates and The corresponding actual execution feedback data, function f( , This is used to calculate the execution effect score of a single instruction or event. The score is normalized according to specific dimension rules. For example, for a boot screen update instruction, The function can compare the degree of match between the content requested to be displayed and the snapshot of the actual displayed content. For parking space allocation schemes, The function can calculate the consistency ratio between the expected parking space allocation and the actual number of parking spaces, and the final management effectiveness evaluation report will include a comprehensive effectiveness score. And detailed sub-scoring for each dimension. It can be understood that the parameters of the dynamic parking space allocation model and the strategies in the guidance strategy library are dynamically optimized and updated based on the management effectiveness evaluation report. This optimization and update is a closed-loop learning process. When the management effectiveness evaluation report indicates that uneven allocation is caused by asynchronous guidance information release, the system may adjust the weight parameters regarding the credibility of the guidance screen status in the dynamic parking space allocation model, or add a constraint to the relevant strategies in the guidance strategy library that "the instruction is considered effective only after receiving confirmation feedback from all target screens." The specific content of the update is directly related to the attribution analysis conclusions in the evaluation report. The system maintains a historical update record library to track the evolution of parameters and strategies and avoid invalid or contradictory updates. Through the above methods, Example 5 fully describes the specific implementation process of the system through feedback collection, correlation analysis, evaluation report generation, and ultimately, dynamic optimization and updating of the model and strategies.

[0048] See Figure 5 This is a bar chart comparing the performance of a parking management system before and after optimization, quantifying the benefits of system optimization across four core evaluation dimensions. The chart visually demonstrates the improvements across multiple dimensions, with the most significant increase in target achievement rate, indicating that the optimization of the dynamic parking space allocation model and conflict resolution sub-model played a crucial role in achieving core operational goals. The simultaneous improvement across all four dimensions verifies the systematic nature and effectiveness of the optimization solution. This not only improves the resource utilization and revenue of the parking lot but also enhances the user parking experience through more timely response and more accurate execution. The quantified comparative data provides clear directions for improvement in subsequent system iterations, such as further optimization of the "target achievement rate" by adjusting the weight allocation of the multi-objective optimization algorithm or the conflict resolution rules.

[0049] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. An intelligent parking management system based on narrowband Internet of Things, characterized in that, The system includes: A narrowband IoT sensor node group deployed in various areas of the parking lot is used to collect a set of vehicle status monitoring data, which includes parking space occupancy status sequence, vehicle parking time sequence and vehicle identification data. Narrowband IoT base station is used to perform multi-source data fusion processing on the vehicle status monitoring data set to generate standardized parking feature information, which includes parking space number with unified timestamp, vehicle identifier and parking duration. The parking management server is used to call a pre-set parking behavior analysis model to perform pattern recognition processing on the standardized parking feature information and generate a parking behavior feature vector. The parking behavior feature vector includes parking pattern, abnormal parking identifier and parking space utilization index. The parking management server is also used to input the parking behavior feature vector into the dynamic parking space allocation model for optimization calculation and processing, and generate a real-time parking space allocation scheme and parking space guidance strategy. The parking management server is also used to generate a parking lot control instruction set based on the real-time parking space allocation scheme and parking space guidance strategy. The parking lot control instruction set includes guidance screen display update instructions and gate control instructions.

2. The intelligent parking management system based on narrowband Internet of Things according to claim 1, characterized in that, The narrowband IoT base station performs multi-source data fusion processing on the vehicle status monitoring data set to generate standardized parking feature information, including: Extract the time point of state change and the duration of state for each parking space from the parking space occupancy state sequence, and convert the original state data into standard state code through a state parser; The arrival and departure times of vehicles are analyzed from the vehicle parking time series to calculate the actual parking duration; Vehicle feature information is extracted from the vehicle identification data and mapped to a unique vehicle identifier. Establish a time coordinate system based on parking lot operating time, and align the timestamps of the standard status code, actual parking duration, and unique vehicle identifier to the time coordinate system. All aligned data are integrated in chronological order to generate standardized parking feature information, including parking space number, vehicle identifier, parking duration, and timestamp.

3. The intelligent parking management system based on narrowband Internet of Things according to claim 1, characterized in that, The parking management server calls a pre-set parking behavior analysis model to perform pattern recognition processing on the standardized parking feature information, generating a parking behavior feature vector, including: The standardized parking feature information is sliced ​​along the time dimension to form a subset of parking records within multiple statistical periods; For each subset of parking records, perform the following processing steps: Identify the distribution pattern of vehicle parking duration in the parking record subset; when the distribution pattern conforms to a preset typical pattern, assign a parking pattern pattern identifier to the parking record subset. Detect abnormal parking events in the parking record subset, construct an abnormal event feature map, and extract the abnormal parking identifier from the abnormal event feature map. The abnormal parking identifier includes an abnormality type code and an abnormality degree value. Analyze the spatiotemporal distribution characteristics of parking space usage in the parking record subset, and calculate the parking space utilization rate index; The processing results of all statistical periods are summarized and analyzed to generate the parking behavior feature vector, which includes parking pattern, abnormal parking indicators, and parking space utilization rate indicators.

4. The intelligent parking management system based on narrowband Internet of Things according to claim 1, characterized in that, The parking management server inputs the parking behavior feature vector into the dynamic parking space allocation model for optimization calculation, generating a real-time parking space allocation scheme and parking space guidance strategy, including: The parking pattern is matched with historical allocation schemes in the parking space allocation knowledge base to determine the basic allocation scheme score; The abnormal parking identifier is input into the conflict resolution sub-model, which is constructed based on the constraint satisfaction algorithm to predict potential parking space usage conflicts and output the conflict resolution priority. The parking space utilization rate index is compared with the parking space utilization rate optimization objective function to obtain the utilization rate optimization suggestion value; The comprehensive optimization value is calculated by integrating the basic allocation scheme score, the conflict resolution priority, and the utilization optimization suggestion value using a multi-objective optimization algorithm. The real-time parking space allocation scheme is determined based on the numerical range of the comprehensive optimization value. Based on the distribution of conflict resolution priorities and the current parking space status, the parking space guidance strategy is calculated. The parking space guidance strategy includes guidance path planning and guidance information release timing.

5. The intelligent parking management system based on narrowband Internet of Things according to claim 1, characterized in that, The parking management server generates a set of parking lot control instructions based on the real-time parking space allocation scheme and parking space guidance strategy, including: Based on the real-time parking space allocation scheme, the guidance strategy library is queried to determine the basic attributes of the guidance information, which include guidance level, guidance content, and target display device. Based on the parking guidance strategy, the scope and path of the guidance information are dynamically adjusted, and a guidance screen display update instruction is generated. The guidance screen display update instruction includes updating the guidance screen, display content, and update priority. Based on the real-time parking space allocation scheme and parking space guidance strategy, a matching set of basic control rules is selected from the gate control rule base; Based on the abnormal parking identifier, the parameters of the basic control rule set are optimized to generate a barrier gate control command. The barrier gate control command includes barrier gate opening condition judgment rules and opening duration control parameters. The guidance screen display update command and the barrier gate control command are packaged into the parking lot control command set.

6. The intelligent parking management system based on narrowband Internet of Things according to claim 5, characterized in that, The step of optimizing the parameters of the basic control rule set based on the abnormal parking identifier and generating the barrier gate control command includes: The vehicle identification and parking space number involved in the abnormal parking event are parsed from the abnormal parking sign. For vehicles with abnormal parking records, an enhanced verification rule instruction is generated. This enhanced verification rule instruction is used to add a secondary identity verification process for the vehicle in the barrier gate control system. For the parking space number involved in the abnormal parking event, a status monitoring rule instruction is generated. The status monitoring rule instruction is used to increase the frequency of status monitoring of parking spaces in the parking management system and set abnormal status warning thresholds. The effective time of the rules is adjusted according to the parking pattern. For high-frequency usage periods, a lenient control strategy is adopted; for low-frequency usage periods, a strict control strategy is adopted. The enhanced verification rule instructions, the status monitoring rule instructions, and the rule effective time period configuration are integrated to form specific and executable gate control instructions.

7. The intelligent parking management system based on narrowband Internet of Things according to claim 3, characterized in that, The step involves identifying the distribution pattern of vehicle parking durations within the parking record subset. When the distribution pattern conforms to a preset typical pattern, a parking pattern pattern identifier is assigned to the parking record subset, including: Access the parking pattern library, which stores various typical parking duration distribution patterns such as temporary parking pattern, long-term parking pattern, peak gathering pattern, and uniform distribution pattern; The distribution of vehicle parking duration in the parking record subset is compared with the distribution similarity of each typical parking duration distribution pattern in the parking pattern library. Calculate the matching degree between the subset of parking records and each typical parking duration distribution pattern; Typical parking duration distribution patterns with matching degrees exceeding a threshold are selected, and their corresponding parking pattern patterns are used as candidate identifiers. If multiple candidate identifiers exist, the final parking pattern identifier is determined by comprehensively judging the time span of the parking record subset, the vehicle type distribution, and the parking space utilization rate index.

8. The intelligent parking management system based on narrowband Internet of Things according to claim 4, characterized in that, The abnormal parking identifier is input into the conflict resolution sub-model, which is constructed based on a constraint satisfaction algorithm to predict potential parking space usage conflicts and output conflict resolution priorities, including: The abnormal parking identifier is converted into conflict event data, where nodes in the conflict event data represent parking spaces and edges represent observed usage conflict relationships. Each conflict event in the conflict event data is assigned an initial importance attribute, which is determined based on the urgency, scope of impact, and historical resolution effectiveness of the conflict event; The conflict resolution sub-model is used to perform iterative analysis and calculation on the conflict event data. In each iteration, the importance of the conflict event will propagate along the edge to its associated conflict event. The strength of the propagation is determined by the association degree of the edge and the solvability of the received event. After a predetermined number of iterations, each conflict event will converge to a stable resolution priority value, which reflects the urgency of handling the conflict event in the current parking lot state. Extract all conflict events whose resolution priority values ​​exceed the processing threshold, along with their corresponding priority values, and output them as the conflict resolution priority.

9. The intelligent parking management system based on narrowband Internet of Things according to claim 4, characterized in that, The calculation of the comprehensive optimization value using a multi-objective optimization algorithm includes: Establish a multi-objective optimization function that includes the objectives of maximizing parking space utilization, minimizing user waiting time, and maximizing parking lot revenue. Each optimization objective is assigned a weight coefficient, which is dynamically adjusted according to the parking lot's operating hours. The basic allocation scheme score is mapped to the input parameters for the goal of maximizing parking space utilization; The conflict resolution priority is converted into a constraint that minimizes user wait time; The proposed utilization rate optimization value is used as the benchmark value for maximizing parking lot revenue. A multi-objective genetic algorithm is used to perform parallel optimization calculations on three optimization objectives, and to find the Pareto optimal solution set under the premise of satisfying various constraints. The solution with the highest overall satisfaction is selected from the Pareto optimal solution set, and its corresponding objective function value is used as the overall optimized value.

10. The intelligent parking management system based on narrowband Internet of Things according to claim 1, characterized in that, The system also includes: The system continuously collects execution feedback data of the parking lot control command set, which includes guidance screen status feedback, barrier gate execution status records, and actual vehicle parking status. The execution feedback data is correlated with the original parking behavior feature vector and the real-time parking space allocation scheme; The effectiveness of the parking lot control instruction set is evaluated based on the correlation analysis results, and a management effectiveness evaluation report is generated. Based on the management effectiveness evaluation report, the parameters of the dynamic parking space allocation model and the strategies of the guidance strategy library are dynamically optimized and updated.

Images