Multi-region coordination-based parking resource comprehensive scheduling strategy optimization method and system

Abstract

The present application relates to the field of intelligent traffic parking scheduling, and particularly relates to a parking resource comprehensive scheduling strategy optimization method and system based on multi-region cooperation, which specifically as follows: collecting multi-region and multi-type parking resources, commuting parking demand, illegal parking risk and road network operation data according to scheduling period, constructing parking cooperative monitoring sample through standardization processing; calculating time-sharing sharing potential index, underground priority access intensity and road overflow risk elasticity index, constructing hierarchical directed ant colony search graph with heuristic information written; introducing capacity constraint self-repairing mechanism and congestion degree pheromone updating mechanism, solving Pareto solution set through multi-objective ant colony optimization, screening optimal compromise scheme combining fuzzy membership and dynamic preference weight, outputting parking guidance instruction and realizing rolling closed-loop iteration. The present application can realize time-sharing cooperative deployment of parking resources in residential and office areas, balance various parking load, and improve overall utilization efficiency of parking resources.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation parking scheduling technology, and in particular to a method and system for optimizing comprehensive parking resource scheduling strategies based on multi-regional collaboration. Background Technology

[0002] As the separation of urban work and residence intensifies, the spatial and temporal mismatch between parking resource supply and demand is becoming increasingly prominent. Residential and office areas exhibit significant tidal parking patterns, and different types of parking facilities—underground, surface, and on-street—are under uneven load, often accompanied by problems such as oversaturation of surface parking spaces, spillover of illegal on-street parking, and congestion of surrounding road networks. Urban parking management needs to coordinate various types of parking resources in both residential and office areas, integrating multi-source monitoring data such as parking space status, distribution of illegal parking incidents, and road network operation status to construct a unified scheduling system. This will allow for the exploration of potential for staggered parking sharing, differentiated allocation of parking resources, and ultimately, the multi-objective governance of improving parking turnover efficiency, balancing parking space load, and suppressing the risk of illegal on-street parking.

[0003] Current parking resource scheduling and guidance technologies still have the following shortcomings: (1) Existing parking guidance mainly relies on the real-time remaining parking spaces of a single parking lot as the basis for decision-making. It does not fully explore the potential for staggered complementarity between residential and office areas within a continuous time window. It only relies on the static occupancy rate of a single time period for matching judgment, which is easily affected by short-term vehicle departure fluctuations and noise interference in detection data, resulting in insufficient stability of cross-regional shared scheduling and limited matching accuracy. (2) Existing scheduling methods usually treat different types of parking facilities as homogeneous parking space supply, without distinguishing the differences in governance attributes of underground parking lots, ground parking lots and on-street parking spaces, and without uniformly quantifying and modeling the priority acceptance value of underground parking spaces and the spillover risk of on-street parking. It is difficult to achieve differentiated guidance for priority use of underground parking spaces and risk control of on-street parking spaces, and it is impossible to take into account both the efficiency of resource utilization and the comprehensive management needs of traffic order. (3) Existing parking scheduling schemes using ant colony algorithm have a low degree of coupling with capacity constraints during the deconstruction process, which easily generates infeasible scheduling schemes with overcapacity; and the heuristic information only reflects single factors such as travel distance and number of remaining parking spaces, and cannot simultaneously represent the cross-regional staggered sharing value, underground priority utilization benefits and on-street risk suppression needs, thus limiting the guidance of algorithm optimization and the feasibility of the scheme. (4) The existing multi-objective parking scheduling method updates the pheromone based only on the non-dominated level, which makes it easy for the search results to cluster in the local preference area, and the diversity of Pareto solution set is difficult to guarantee. At the same time, the final scheme selection mostly adopts the fixed weight mode, which cannot dynamically adjust the preference weight according to the needs of different time periods and different governance scenarios, and the scenario adaptability of the scheduling strategy is insufficient.

[0004] Therefore, this invention proposes a method and system for optimizing the comprehensive scheduling strategy of parking resources based on multi-regional collaboration to solve the above problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention proposes a method and system for optimizing the comprehensive scheduling strategy of parking resources based on multi-regional collaboration. This invention can realize the staggered and coordinated allocation of parking resources in residential and office areas, balance the load of various parking spaces, and improve the overall utilization efficiency of parking resources.

[0006] On the one hand, the technical solution of this invention to solve the technical problem is a method for optimizing the comprehensive scheduling strategy of parking resources based on multi-regional collaboration, including the following steps: S1. Collect parking resources, vehicle demand, on-street risks and road network operation data in multiple areas according to the divided scheduling time periods. Based on the collected data, determine the standardized occupancy rate time series curve, parking facility space capacity, number of parking facilities occupied spaces, number of parking facilities remaining spaces, demand for vehicles to be allocated, historical standardized heat value of illegal parking and standardized value of road network service level, and construct parking collaborative monitoring samples. S2. Based on the parking collaborative monitoring samples, calculate the staggered sharing potential index, underground priority access intensity and road spillover risk elasticity index, and then calculate the heuristic information value, and then write the heuristic information value into the hierarchical directed ant colony search graph. S3. Construct an ant scheduling scheme on a hierarchical directed ant colony search graph, introduce a capacity pressure sensing factor and a capacity constraint self-repair mechanism to reduce infeasible schemes, and then maintain the diversity of Pareto schemes through a crowding-sensing pheromone update mechanism to obtain the optimal compromise scheme. S4. Convert the optimal compromise solution into an executable parking guidance instruction, and continuously update the input data and scheduling parameters for the next scheduling period based on the execution results.

[0007] S1 is as follows: The data collection area includes parking facilities in the functional departure area and the office functional purpose area. The types of parking facilities include underground parking lots, surface parking lots, and on-street parking spaces. Parking resource data includes vehicle entry time, vehicle exit time, parking space occupancy status, and parking space capacity. Vehicle entry time, vehicle exit time, and parking space occupancy status in underground and surface parking lots are collected through barrier gate systems, video detection systems, geomagnetic detection systems, or parking fee collection systems. On-street parking space occupancy status is collected through roadside video detection equipment, geomagnetic detection equipment, inspection terminals, or parking fee collection terminals. A standardized occupancy rate is calculated based on vehicle entry time and vehicle exit time. Parking facility capacity, the number of occupied parking spaces, and the number of remaining parking spaces are calculated based on parking space occupancy status and parking space capacity. The vehicle demand data is the parking demand data from the residential area to the office area. The number of vehicles to be allocated is obtained by integrating parking reservation data, mobile terminal navigation destination data, historical commuting traffic prediction results and real-time parking guidance requests. The on-street risk data consists of historical illegal parking incident data, which is obtained by collecting illegal parking capture records, law enforcement records, and manual inspection records. The number of historical illegal parking incidents is obtained by statistical analysis by region and dispatch time period, and the standardized heat map value of historical illegal parking is obtained by normalizing the maximum value. Road network operation data includes surrounding road network operation status data, and the standardized value of road network service level is calculated by comprehensively considering the average vehicle speed, queue length, congestion index and road segment capacity. This results in a parking collaborative monitoring sample that includes standardized occupancy rate time-series curves, parking facility capacity, number of occupied parking spaces, number of remaining parking spaces, demand for vehicles to be allocated, standardized heatmap values ​​of historical illegal parking, and standardized values ​​of road network service levels. , Indicates the first Parking coordination monitoring samples for each scheduling period, The index represents the scheduling period, and its value range is... arrive , This indicates the total number of scheduling periods per day.

[0008] S2 is as follows: S2.1 Set a time offset window near the current scheduling period, and perform weighted matching of the vacancy supply capacity of the residential function departure area and the parking demand intensity of the office function destination area to obtain the staggered sharing potential index; S2.2 Calculate the underground priority access intensity based on the parking capacity and remaining number of parking spaces in the underground parking lot and the parking capacity and occupied number of parking spaces in the surface parking lot in the office functional area. The spillover risk elasticity index is calculated based on the parking capacity, number of occupied parking spaces, historical standardized heat value of illegal parking, and standardized value of road network service level of the on-street parking area for office function purposes. The underground priority access intensity and the spillover risk elasticity index between roads and outside the road are combined into a collaborative governance factor vector; S2.3. The residential function departure area, the office function destination area, and the parking facility type are jointly encoded as nodes in a hierarchical directed ant colony search graph. The vehicle guidance unit, which is divided according to the demand for vehicles to be allocated, is used as the minimum allocation object for ants to build a scheduling scheme, and a hierarchical directed ant colony search graph is established.

[0009] S2.1 is as follows: A time offset window is selected with the current scheduling period t as the center, and the radius of the time offset window is set to be . The time offset index is h, and the value range of h is... arrive If t+h is less than 1 or greater than 1 Instead, the occupancy rate of the boundary scheduling period is used; The standardized occupancy rate of the residential function departure area in the t+h scheduling period is converted into a free parking rate. The standardized occupancy rate of the office function destination area in the t scheduling period is taken as the current parking demand intensity. By multiplying the standardized free parking rate with the current parking demand intensity, the shared matching intensity between the residential function departure area and the office function destination area at the time offset h is obtained. A decay weight is set for the basic sharing matching strength under different time offsets, and a weighted average is performed within the time offset window to obtain the time-sharing potential index.

[0010] S2.2 is as follows: The remaining number of parking spaces and their capacities in multiple underground parking garages within the office function area are summed, and the ratio of these sums yields the proportion of remaining parking spaces in the underground garages. The occupied number of parking spaces and their capacities in the multiple underground parking garages within the office function area are summed, and the ratio of these sums yields the oversaturation rate of the surface parking garages. Based on the proportion of remaining parking spaces in the underground garages and the oversaturation rate of the surface parking garages, the underground priority access intensity is calculated using the following formula: , in, This represents the underground priority access intensity of the j-th office function destination area during the t-th scheduling period, with a value range of [0,1], where j represents the index of the office function destination area; clip( (,0,1) denotes the truncation function; Indicates the weight of remaining underground capacity; Indicates the ground oversaturation weight; This represents the number of remaining parking spaces in the underground parking lot of the j-th office function destination area during the t-th scheduling period; This represents the parking capacity of the underground parking lot in the j-th office functional area; Represents the minimum stability constant; This represents the number of parking spaces occupied in the ground parking lot of the j-th office function destination area during the t-th scheduling period; This represents the parking capacity of the ground-level parking lot in the j-th office functional area; max( () represents the function that takes the maximum value; For the j-th office function area, read the number and capacity of parking spaces in on-street parking lots, the historical standardized heat value of illegal parking, and the standardized value of road network service level for the t-th scheduling period. Obtain the standardized occupancy rate of on-street parking based on the ratio of the number of parking spaces to the capacity of on-street parking lots. Add one to the negative of the standardized value of road network service level as the road operation pressure. Sum the standardized occupancy rate of on-street parking, the historical standardized heat value of illegal parking, and the road operation pressure with weights to obtain the spillover risk elasticity index between on-street and off-street areas. Finally, the underground priority access intensity and the spillover risk elasticity index between roads and outside the road are merged into a collaborative governance factor vector.

[0011] S2.3 is as follows: Establish a hierarchical directed ant colony search graph , This represents the directed graph structure used for ant colony search during the t-th scheduling period, including the set of nodes and the set of feasible directed edges; First, generate underground parking nodes, surface parking nodes, and on-street parking space nodes by combining each residential function departure area with the office function destination area. Then, screen the nodes for feasibility based on the remaining capacity of parking facilities, gate access, reservation rules, traffic restriction strategies, and management strategies, and retain candidate routes. The demand for vehicles to be allocated is divided into vehicle guidance units according to the proportional granularity, which are used as the smallest vehicle groups to construct parking allocation schemes in ant colony search. Heuristic information values ​​are calculated based on the time-sharing potential index, underground priority access intensity, and spillover risk elasticity index. These heuristic information values ​​are then written into the feasible directed edges of the hierarchical directed ant colony search graph.

[0012] The ant scheduling scheme in S3, which introduces a capacity pressure-aware factor and a capacity constraint self-repair mechanism, is as follows: Initialize the ant population and pheromone concentration. Let the ant index be 'a', ranging from 1 to A, where A represents the total number of ants. Let the iteration index be 'r', ranging from 1 to... , The maximum number of iterations is represented by the pheromone concentration used in the r-th iteration of the t-th scheduling period to travel from the ith residential function departure area to the j-th office function destination area and select the k-th type of parking facility. The pheromone concentration of all feasible paths is initialized to the same constant, and no pheromone concentration is set for impassable nodes and impassable directed edges. The remaining number of parking spaces in various types of parking facilities in each office functional area is compared with the demand for vehicles to be allocated from each residential departure area to each office functional area to obtain the capacity pressure perception factor; the state transition probability is calculated based on pheromone concentration, heuristic information value and capacity pressure perception factor. Each ant selects the parking facility type for the vehicle guidance unit in turn and generates an initial scheduling plan. For each combination of residential function departure area and office function destination area, the ants allocate vehicle guidance units one by one according to the state transition probability using a roulette wheel method. The initial scheduling plan is self-repaired based on capacity constraints. If the initial scheduling plan causes the number of vehicles allocated to a certain type of parking facility in the office function area to exceed the number of remaining parking spaces, the excess vehicles are recorded as vehicles to be repaired, and alternative parking facilities are selected from other feasible parking facility types in the same office function area. Then, a multi-objective evaluation is performed on the repaired ant scheduling plan. The multi-objective evaluation includes parking turnover efficiency, surface parking oversaturation, underground parking utilization, and on-street parking spillover risk. Finally, based on the multi-objective evaluation results, a fast non-dominated sorting was performed on all ant scheduling schemes to obtain multiple Pareto fronts.

[0013] The specific pheromone update mechanism for density sensing in S3 is as follows: For all ant scheduling schemes generated in the r-th iteration, perform non-dominated sorting and calculate the congestion distance within each Pareto front. For each Pareto front, sort according to the parking turnover efficiency target, the surface parking oversaturation target, the underground parking utilization level target, and the on-street parking spillover risk target. After normalizing each target, calculate the target difference between adjacent ant scheduling schemes and sum the multiple target differences to form the congestion distance. Construct an elite ant set based on the non-dominated level and the congestion distance, prioritizing the selection of ants with large congestion distances in the first Pareto front, and randomly selecting a small number of ants from the second or third Pareto front. The pheromone evaporation coefficient is determined based on the Pareto front distribution entropy. The ant scheduling schemes in the first Pareto front are then divided according to the target spatial location. There are several intervals, where b represents the index of the target spatial distribution interval. The first Pareto front will fall into the [missing information - likely a specific range or range]. The proportion of schemes in each interval is denoted as Iterate through all the partitioned intervals, based on Calculate the Pareto front entropy, while the ideal uniform distribution entropy is expressed as... The pheromone evaporation coefficient is calculated based on the lower limit of the pheromone evaporation coefficient, the difference between the upper and lower limits of the pheromone evaporation coefficient, the Pareto front distribution entropy, and the ideal uniform distribution entropy; then the pheromone concentration is updated for each feasible directed edge in the hierarchical directed ant colony search graph. Boundary limits are imposed on the updated pheromone concentration, setting a lower and upper limit for the pheromone concentration.

[0014] The optimal compromise solution in S3 is selected as follows: When the ant colony reaches the maximum number of iterations, or when the Pareto scheduling scheme set no longer shows significant improvement after several consecutive rounds, stop ant colony optimization and collect the Pareto scheduling scheme set; convert the four-item target values ​​of each candidate Pareto scheduling scheme into fuzzy membership degrees, set the dynamic preference weights corresponding to the current governance scenario, calculate the overall satisfaction based on the fuzzy membership degrees and dynamic preference weights, and select the optimal compromise scheme. The fuzzy membership calculation process is as follows: The index of the four-objective multi-objective evaluation, including parking turnover efficiency, surface parking oversaturation, underground parking utilization, and on-street parking spillover risk, is denoted as m, with a value range of 1 to 4; the index of the candidate Pareto scheduling scheme is denoted as p, and the fuzzy membership degree of the p-th candidate Pareto scheduling scheme on the m-th objective is denoted as... The target values ​​include benefit-based targets and cost-based targets, and the calculation formula is as follows: For benefit-oriented goals: ; For cost-related objectives: ; in, This represents the maximum objective value of the m-th item in the Pareto scheduling scheme set; Let m represent the minimum objective value of the m-th item in the Pareto scheduling scheme set; This represents the target value of the p-th candidate Pareto scheduling scheme on the m-th target; Denotes the minimum stability constant; if and If they are the same, then the fuzzy membership degree of the target is uniformly set to 1.

[0015] On the other hand, the present invention also provides a parking resource integrated scheduling strategy optimization system based on multi-regional collaboration, including a module for executing the processing instructions of each step in a parking resource integrated scheduling strategy optimization method based on multi-regional collaboration; Data collection sample construction module: Collects raw data on parking facilities, vehicle demand, illegal parking risks, and road network operation in multiple time domains, and generates standardized occupancy rate time series, parking space inventory, unallocated demand, violation heat map, and road network service level indicators through standardization and normalization operations, and outputs parking collaborative monitoring samples for each time period. Collaborative Feature and Ant Colony Graph Modeling Module: Based on monitoring samples, the module calculates the staggered sharing potential index, underground priority access intensity, and spillover risk elasticity index to generate collaborative governance factor vectors; it encodes areas and parking facilities to generate search graph nodes, divides vehicle guidance units, calculates path heuristic information values, and writes them into a hierarchical directed ant colony search graph. Multi-objective ant colony optimization solution module: Initialize ant colony pheromones, introduce capacity pressure sensing factors and capacity constraint self-repair mechanism to generate feasible scheduling schemes; rely on crowding sensing pheromone update mechanism to iteratively optimize, and obtain Pareto solution set through non-dominated sorting; calculate comprehensive satisfaction by multi-objective fuzzy membership degree combined with dynamic preference weight, and output the optimal compromise scheduling scheme. The closed-loop iteration module for guidance instructions converts the optimal scheduling scheme into parking guidance instructions for execution, collects landing feedback data, updates the input data and scheduling parameters for the next time period, and realizes dynamic iterative optimization of the scheduling strategy.

[0016] The effects described in the invention are merely those of the embodiments, and not all the effects of the invention. The above technical solutions have the following advantages or beneficial effects: This invention discloses a comprehensive parking resource scheduling strategy optimization method and system based on multi-regional collaboration. It proposes a method for calculating the staggered-time sharing potential index based on time offset windows and attenuation weights. This method weights and matches the vacant supply capacity of residential areas with the parking demand intensity of office areas in a time dimension, overcoming the limitations of static occupancy rate judgment during a single scheduling period. It mitigates the interference of time-period fluctuations on the matching results, improving the stability and accuracy of cross-regional parking resource sharing matching. Furthermore, this invention constructs a dual-dimensional collaborative governance factor combining underground priority access intensity and on-street spillover risk elasticity index. On the one hand, it guides the priority use of underground parking spaces by combining the remaining capacity of underground parking spaces and the oversaturation of surface parking lots; on the other hand, it integrates on-street parking saturation, illegal parking accumulation intensity, and road network operation pressure to manage the spillover risk of on-street parking. This expands parking scheduling optimization beyond simple capacity allocation. To comprehensively manage and regulate traffic order while balancing resource utilization, this invention integrates the potential for staggered-time sharing, underground priority guidance, and spillover suppression features to construct heuristic information. It also introduces a capacity pressure sensing factor and a capacity constraint self-repair mechanism, performing pre-emptive capacity constraint verification during ant colony search and automatically transferring overcapacity vehicles to other feasible parking facilities in the same area. This significantly reduces infeasible scheduling schemes and improves algorithm efficiency and feasibility. Furthermore, a pheromone update mechanism based on Pareto front congestion perception is established. This mechanism coordinates pheromone release and volatilization intensity through non-dominated levels, congestion distance, and Pareto front distribution entropy, ensuring the diversity of Pareto solutions. Finally, dynamic preference weights and fuzzy membership degrees are used to select the optimal compromise solution, enabling scheduling results to adapt to changing needs in different time periods and governance scenarios, thus enhancing the scenario adaptability and dynamic control capabilities of the scheduling strategy. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0018] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0019] Figure 2 Heat map of potential index for staggered sharing.

[0020] Figure 3 A schematic diagram showing the priority access intensity of underground spaces for various office functions and the elasticity index of spillover risk between roads and the outside.

[0021] Figure 4 A schematic diagram illustrating the capacity pressure perception factors for different types of vehicle pick-up facilities in various office functional areas.

[0022] Figure 5 This is a schematic diagram of the parallel coordinates of the multi-objective normalized values ​​of the Pareto scheduling scheme set.

[0023] Figure 6 A schematic diagram of a multi-objective Pareto front for parking turnover efficiency and ground oversaturation.

[0024] Figure 7 A schematic diagram of the multi-objective Pareto front for underground utilization levels and spillover risks inside and outside the road. Detailed Implementation

[0025] To clearly illustrate the technical features of this solution, the invention will be described in detail below through specific implementation methods and in conjunction with the accompanying drawings.

[0026] Example 1 like Figure 1 As shown, a method for optimizing parking resource scheduling strategy based on multi-regional collaboration includes the following steps: S1. Collect parking resources, vehicle demand, on-street risks and road network operation data in multiple areas according to the divided scheduling time periods. Based on the collected data, determine the standardized occupancy rate time series curve, parking facility space capacity, number of parking facilities occupied spaces, number of parking facilities remaining spaces, demand for vehicles to be allocated, historical standardized heat value of illegal parking and standardized value of road network service level, and construct parking collaborative monitoring samples. S2. Based on the parking collaborative monitoring samples, calculate the staggered sharing potential index, underground priority access intensity and road spillover risk elasticity index, and then calculate the heuristic information value, and then write the heuristic information value into the hierarchical directed ant colony search graph. S3. Construct an ant scheduling scheme on a hierarchical directed ant colony search graph, introduce a capacity pressure sensing factor and a capacity constraint self-repair mechanism to reduce infeasible schemes, and then maintain the diversity of Pareto schemes through a crowding-sensing pheromone update mechanism to obtain the optimal compromise scheme. S4. Convert the optimal compromise solution into an executable parking guidance instruction, and continuously update the input data and scheduling parameters for the next scheduling period based on the execution results.

[0027] In a specific implementation, the modeling of the multi-source parking collaborative monitoring data acquisition and scheduling object in S1 is as follows: When coordinating parking resource scheduling across multiple areas, it is necessary to simultaneously monitor the departure areas for residential functions, the destination areas for office functions, underground parking lots, surface parking lots, on-street parking spaces, historical illegal parking incidents, and the operational status of the surrounding road network. This step first collects data on parking resources, vehicle demand, on-street risks, and road network operation, and then constructs a parking coordination monitoring sample. This sample provides a unified input for subsequent coordination feature mining and ant colony search. The specific steps are as follows: S1.1 Scheduling Time Period Division and Regional Parking Resource Data Collection 1) Divide the daily operating time into multiple scheduling periods according to fixed scheduling intervals, and use the scheduling period index... express, The time slice used to identify the coordinated scheduling of parking resources, with a value range of [value range missing]. arrive ; This indicates the total number of scheduling periods per day (e.g., when the scheduling interval is 15 minutes). The current scheduling period refers to the time slice in which the scheduling system is generating parking guidance plans, denoted as the [number]th [time slice]. Each scheduling period.

[0028] In practice, the scheduling interval can be set to 5 minutes, 10 minutes or 15 minutes according to the rate of change in parking demand. When the parking flow in the target area fluctuates rapidly, the scheduling interval is set to 5 minutes or 10 minutes. When the parking flow in the target area is relatively stable, the scheduling interval is set to 15 minutes to reduce the computational burden.

[0029] 2) Obtain vehicle entry time, vehicle exit time, parking space occupancy status, and parking space capacity for the residential departure area and the office destination area. (Index used for residential departure area) express, Used to locate residential areas that can provide off-peak shared parking, with a value range of [value range missing]. arrive ; This indicates the total number of residential departure areas, which can be taken as 12. The index for office function destination areas is used... express, Used to locate office functional areas that generate parking demand and receive parking guidance; the value range is [value range missing]. arrive ; This indicates the total number of functional areas for office use, for example, 8.

[0030] Furthermore, the parking facility type index uses express, Used to distinguish parking facilities with different management attributes within an area designated for office functions, the range of values ​​is underground, surface, and on-street; where underground refers to underground parking lots, surface refers to surface parking lots, and on-street refers to on-street parking spaces. The first office functional area The parking capacity of this type of parking facility is denoted as , Used to characterize the maximum number of parking spaces that the parking facility can accommodate; the The first office functional area Class II parking facilities The number of berths occupied during each scheduling period is recorded as follows: , Used to indicate the number of parking spaces that are already occupied during the current scheduling period; The first office functional area Class II parking facilities The number of remaining berths in each scheduling period is denoted as: , This refers to the parking facility on the [number]th [day / month]. The number of parking spaces that can accommodate additional vehicles during a given scheduling period can be calculated by letting... ,in, This represents the function that takes the maximum value.

[0031] In practical implementation, vehicle entry time, vehicle exit time, and parking space occupancy status in underground and surface parking lots can be collected by barrier gate systems, video detection systems, geomagnetic detection systems, or parking fee collection systems; the occupancy status of on-street parking spaces can be collected by roadside video detection equipment, geomagnetic detection equipment, inspection terminals, or parking fee collection terminals. If multiple underground parking lots exist within the same office function area, the number of occupied and remaining parking spaces in each underground parking lot is first counted separately, and then summed according to the office function area to obtain the underground parking lot capacity, the number of occupied parking spaces, and the number of remaining parking spaces at the target area level.

[0032] In one embodiment, for example, if a certain office functional area There are two underground parking garages. The first garage has a capacity of 200 parking spaces, with 150 currently occupied. The second garage has a capacity of 300 parking spaces, with 180 currently occupied. Therefore, after aggregating by area, what is the total underground parking capacity for this area? For 500, the number of berths occupied There are 330 available, and the number of remaining berths is... There are 170.

[0033] 3) Calculate the standardized occupancy rate time-series curve based on vehicle entry and exit times. The residential function starting area is in the first The standardized occupancy rate for each scheduling period is denoted as: , Used to characterize the The saturation level of parking resources in each residential area, with a value range of [value range missing]. ;No. The office functional area is in the first The standardized occupancy rate for each scheduling period is denoted as: , Used to characterize the The parking demand intensity of each office functional area, with a value range of [value missing]. .

[0034] In specific implementation, for the first Starting from each residential function area, first count the... The number of berths occupied during the first scheduling period, divided by the number of berths in the second scheduling period. The total number of parking spaces in each residential departure area is obtained. For the first For each office functional area, first sum the number of occupied parking spaces in the underground parking lot, surface parking lot, and on-street parking area, then divide by the number of occupied spaces in each office functional area. The total number of all legal parking spaces in an office-purpose area was obtained. If the detection equipment experiences a short-term missing test, the average occupancy rate of two adjacent scheduling periods can be used to fill the gap; if the missing test continues to exceed a preset threshold (e.g., 30 minutes), the corresponding parking facility will be marked as a low-reliability data source, and the parking facility's selectability priority in subsequent ant colony searches will be reduced.

[0035] S1.2 Collection of parking demand to be allocated and road network operation data 1) Collect parking demand data from residential areas to office areas to obtain the demand for vehicles to be allocated. The scheduling period starts from the first The residential departure area leads to the first The number of vehicles to be allocated for each office functional area is denoted as follows: , It refers to the first The number of vehicles that need to be allocated by the dispatch system or guided across regions during each dispatch period.

[0036] In practical implementation, This can be obtained by fusing parking reservation data, mobile terminal navigation destination data, historical commuter traffic prediction results, and real-time parking guidance requests. In one implementation, the number of vehicles that have submitted parking reservations or navigation requests within the current scheduling period is first counted, denoted as [the number of vehicles]. , Indicates the first Parking reservations or navigation requests submitted during the dispatch period are from the [number]th [period]. The residential departure area leads to the first The real-time number of vehicles in each office function area; then, based on historical records of the same week type, same holiday type, and adjacent time slots, the number of vehicles arriving in a short period of time is predicted, denoted as... , Indicates the first The scheduling period is predicted based on historical data of the same week type, same holiday type, and adjacent time slices, starting from the [number]th [time slot]. The residential departure area leads to the first The short-term predicted number of vehicles in each office functional area is then calculated; the vehicle demand to be allocated is then obtained using a weighted average method, expressed as follows: .in, This represents the real-time demand weight, used to control the impact of real-time parking requests on the number of vehicles to be allocated; a value of 0.7 is acceptable. This represents the predicted demand weight, used to control the impact of historical forecasts on the demand for vehicles to be allocated; a value of 0.3 is acceptable. It also satisfies the following conditions: When real-time booking or navigation data coverage is low, the speed can be appropriately increased. To avoid underestimating parking demand; This represents the floor function, used to convert the fusion result into the number of vehicles.

[0037] 2) Collect historical data on illegal parking incidents to obtain standardized heatmap values ​​for historical illegal parking. The office functional area is in the first The historical standardized heat value of illegal parking during each scheduling period is recorded as follows: , Used to characterize the The clustering intensity of illegal parking incidents in a single office functional area under similar time conditions, with a value range of [value missing]. .

[0038] In practice, the system first collects records of illegal parking violations captured by cameras, enforcement records, and manual inspection records from the past 30, 60, or 90 days. Then, it statistically analyzes these records according to office function areas and scheduling periods to obtain the historical number of illegal parking incidents. Finally, it uses maximum value normalization to map the historical number of illegal parking incidents to... Interval. When using maximum value normalization, the first interval can be... The office functional area is in the first The number of historical illegal parking incidents in a given scheduling period divided by the maximum number of historical illegal parking incidents in all office functional areas across all scheduling periods yields the following result: Through this process, office-oriented areas with a high historical incidence of illegal parking will receive a higher risk impact in the subsequent spillover risk elasticity index.

[0039] In one embodiment, for example, if the statistical scope includes all historical illegal parking events in all office functional areas during all scheduling periods, the maximum number of historical illegal parking events in a single area and time period is 50, and the number of events in the first area is... The office functional area is in the first If the number of historical illegal parking incidents during a given scheduling period is 30, then the standardized heatmap value of historical illegal parking in that area during this period is... .

[0040] 3) Collect the operational status of the surrounding road network to obtain standardized values ​​for the road network service level. The office functional area is in the first The standardized value of the road network service level for each scheduling period is denoted as: , Used to characterize the The carrying capacity of the surrounding roads of each office functional area to accommodate new parking detours and on-street parking interference is defined as follows: A higher value indicates a better road condition.

[0041] In practical implementation, the standardized value of the road network service level can be calculated comprehensively based on the average vehicle speed, queue length, congestion index, and road segment capacity. In one implementation method, the first... Key road segments within a predetermined radius surrounding the designated office functional area are used as the evaluation road segment set; then, the ratio of the actual average vehicle speed to the free-flow vehicle speed is calculated for each evaluation road segment; finally, a weighted average is calculated for all evaluation road segments based on segment length to obtain... .

[0042] In one embodiment, as an example, suppose the first There are three road segments evaluated around the office function area: Segment 1, 500 meters long, with a free-flow speed of 60 km / h and a current average speed of 45 km / h, with a ratio of 0.75; Segment 2, 300 meters long, with a free-flow speed of 50 km / h and a current average speed of 30 km / h, with a ratio of 0.60; and Segment 3, 200 meters long, with a free-flow speed of 60 km / h and a current average speed of 48 km / h, with a ratio of 0.8. Therefore, the standardized value of the road network service level is... If the average speed of roads surrounding a certain office function area is close to the free-flow speed, then Approaching 1; if the surrounding roads have long queues and a high congestion index, then With the reduction, the allocation of on-street parking will be more strongly suppressed in the future.

[0043] 4) Construct parking cooperative monitoring samples. The parking cooperative monitoring samples are denoted as... , It refers to the first The multi-source data set used for parking collaborative scheduling during each scheduling period includes standardized occupancy rate time series curves, parking facility space capacity, number of occupied parking spaces, number of remaining parking spaces, demand for vehicles to be allocated, standardized heat values ​​of historical illegal parking, and standardized values ​​of road network service levels.

[0044] In a specific implementation, the S2 parking resource collaborative feature mining and hierarchical directed ant colony search graph construction are as follows: When coordinating parking resource scheduling, it's not only necessary to know where there are available parking spaces, but also to determine which areas have shared value during off-peak hours, which destination areas are suitable for priority access to underground parking, and which destination areas would amplify spillover risks if on-street parking were continued. This step calculates the off-peak sharing potential index, underground priority access intensity, and on-street spillover risk elasticity index based on parking coordination monitoring samples, and writes these coordination characteristics into a hierarchical directed ant colony search graph. The specific steps are as follows: S2.1 Calculation of the Potential Index for Time-Sharing Because residential departure areas and office destination areas typically exhibit opposite parking demand patterns—residential departure areas have more vacant parking spaces during working hours, while office destination areas have higher parking demand—meeting a shared parking situation based solely on single-point occupancy rates during the current scheduling period is susceptible to interference from short-term departures, early arrivals, or detection noise. This step establishes a time offset window near the current scheduling period to weightedly match the vacant supply capacity of residential departure areas with the parking demand intensity of office destination areas, yielding a staggered-time sharing potential index. The specific steps are as follows: 1) Based on the current scheduling period Select the time offset window centered on. The time offset index is used... express, Used to iterate through adjacent time slices near the current scheduling period, with a value range of [value range missing]. arrive ; This represents the time offset window radius, used to limit the search range for time-shifting complementarity relationships; a value of 4 is acceptable. When the scheduling interval is 15 minutes and... At that time, the time offset window covers 60 minutes before and after the current scheduling period.

[0045] In practical implementation, if Less than 1 or greater than This can be replaced by using the occupancy rate of the boundary scheduling period. If Then take Make That is, the standardized occupancy rate of the first scheduling period is used to replace the original occupancy rate value that exceeds the lower limit; if Then take Make That is, using the first The standardized occupancy rate for each scheduling period replaces the original occupancy rate value exceeding the upper limit. Through boundary processing, it can be ensured that the staggered sharing potential index can also be calculated for boundary time periods such as early morning or late at night.

[0046] 2) Calculate the basic shared matching strength based on idle rate and demand intensity. In specific implementation, for the first... The residential function starting area is in the first Standardized occupancy rate per scheduling period Perform an idle transition to obtain the first... The residential function starting area is in the first The standardized idle rate for each scheduling period, the standardized idle rate is The standardized idle rate is used to characterize the first... Each residential departure area can release its berth supply capacity for cross-regional shared scheduling.

[0047] Furthermore, the first The office functional area is in the first Standardized occupancy rate per scheduling period As the current parking demand intensity, the standardized vacancy rate can be multiplied by the current parking demand intensity to obtain the [number]. The residential function starting area and the first The time shift of each office functional area The basic sharing matching strength is the key factor. The higher the basic sharing matching strength, the more vacant parking spaces can be released in the residential starting area, the stronger the parking demand in the office destination area, and the higher the cross-regional sharing incentive value.

[0048] 3) Attenuation weights are applied to the basic sharing matching strength under different time offsets, and a weighted average is performed within the time offset window to obtain the time-sharing potential index. The time-sharing potential index is denoted as... , Indicates the first The residential function starting area and the first The office functional area is in the first The staggered-time sharing potential index for each scheduling period is used to measure the matching strength between the supply of vacant parking spaces in residential departure areas and the parking demand in office destination areas on the time axis, with a value range of [value missing]. .

[0049] In one implementation, first through The calculation method is as follows: for each time offset Constructing time decay weights , Time offset The corresponding decay weights; where, This represents the time offset attenuation coefficient, used to control the impact of data far from the current scheduling period on the time-sharing potential index, and can be set to 0.35; Indicates time offset index The absolute value; This represents the natural exponential function. Then, based on the decay weights, the basic shared matching strength is normalized and weighted to obtain the time-sharing potential index, calculated as follows: ; in, Used to characterize the The residential function starting area is in the first Parking saturation level during each scheduling period; Used to characterize the The office functional area is in the first The intensity of parking demand during each scheduling period. Based on this, the vacancy status of residential areas closer to the current scheduling period has a greater impact on the potential index of staggered-time sharing, while the occupancy status further away from the current scheduling period has a smaller impact, thus taking into account both short-term fluctuations and stable complementary patterns.

[0050] like Figure 2 As shown, a heatmap of the potential index for time-sharing is analyzed, displaying the distribution of the potential index between the starting areas of each residential function and the destination areas of each office function. The horizontal axis represents the index of the destination area of ​​the office function (dimensionless), and the vertical axis represents the index of the starting area of ​​the residential function (dimensionless). The colors from red to yellow indicate that the value of the potential index for time-sharing (dimensionless) decreases from large to small. The experiment shows that some cross-regional combinations have a high potential index for time-sharing, which can be used to guide these combinations to prioritize cross-regional sharing.

[0051] 4) Incorporate the time-sharing potential index into the cross-regional sharing characteristic table. The cross-regional sharing characteristic table is indexed by residential function. Office Function Purpose Area Index and scheduling period index As a multidimensional index key, it is used to quickly read the time-sharing potential index of the corresponding cross-region combination during the initialization of the hierarchical directed ant colony search graph.

[0052] In one embodiment, for example, if the first If the standardized occupancy rate of the residential function departure area is 0.25 during the corresponding scheduling period at 10:00 AM, then the standardized vacancy rate is 0.75; if the first... If the standardized occupancy rate of each office functional area is 0.92 during the same scheduling period, then the... The parking demand intensity is high in the office function area. If the time offset window radius is 4 and the attenuation coefficient is 0.35, the first... Each residential function starting area is in If the standardized occupancy rate within the time offset window remains stable at 0.25, then after time offset weighted calculation, This indicates that the cross-regional combination has strong potential for staggered sharing, and the scheduling system prioritizes the combination of the residential function departure area and the office function destination area in the cross-regional sharing incentive candidate.

[0053] It should be noted that the staggered-time sharing potential index can extend the static judgment of "whether there are currently vacant berths" to a dynamic judgment of "whether there is supply and demand complementarity in the current and adjacent time periods". By using time offset windows and time decay weights, the impact of abnormal occupancy values ​​in a single scheduling period on the sharing matching judgment can be reduced, thereby improving the stability of cross-regional shared scheduling.

[0054] S2.2 Calculation of Underground Priority Access Intensity and Spillover Risk Elasticity Index for Roadside and Inland Areas Because underground parking lots, surface parking lots, and on-street parking spaces have different governance attributes, underground parking lots have concentrated carrying capacity, and prioritizing their use helps alleviate surface parking pressure. On-street parking spaces directly occupy road space; when on-street occupancy rates are high, historical illegal parking incidents are concentrated, or the surrounding road conditions are poor, continuing to allocate on-street parking spaces can easily induce illegal parking spillover and deteriorate road conditions. This step calculates the underground priority access intensity and the on-street spillover risk elasticity index, respectively, as follows: 1) Read the first The scheduling period is the first Number of remaining parking spaces in the underground parking lot of the office functional area , No. The underground parking capacity of the office functional area , No. The scheduling period is the first Number of parking spaces occupied in the ground parking lot of each office functional area and the The ground parking capacity of the office functional area .in, Used to characterize the The underground parking garage in the office function area has the capacity to accommodate additional vehicles; Used to standardize the number of remaining parking spaces in underground parking lots; Used to characterize the The ground parking pressure in the office function area; Used to determine whether a surface parking lot has become oversaturated.

[0055] In practical implementation, if the first If an office function area contains multiple underground parking lots, then the sum of the remaining parking spaces in all underground parking lots will yield the result. The parking capacity of multiple underground parking lots is summed to obtain Ground parking data is aggregated in the same way according to office function and purpose areas.

[0056] 2) Calculate the underground priority access intensity based on the remaining parking space ratio of underground parking lots and the oversaturation ratio of surface parking lots. The underground priority access intensity is denoted as... , Indicates the first The office functional area is in the first The underground priority access intensity for each scheduling period is used to measure the appropriateness of prioritizing vehicles to underground parking lots, with a value range of [value range missing]. .

[0057] In the specific implementation, first Divide by The remaining parking space ratio in the underground parking lot is obtained; then the calculation is performed. When the difference is greater than 0, it indicates that the surface parking lot has exceeded its capacity limit; when the difference is less than or equal to 0, it indicates that the surface parking lot has not exceeded its capacity limit. The excess capacity is then divided by... The oversaturation ratio of ground parking lots is obtained, and the calculation method is expressed as follows: ; in, This represents the weight of the remaining underground capacity, used to control the impact of the proportion of remaining parking spaces in underground parking lots on the intensity of underground priority access; it can be taken as 0.6. This represents the surface oversaturation weight, used to control the influence of surface parking pressure on underground priority access intensity, and can be taken as 0.4; and satisfies... ; This represents the minimum stability constant, used to prevent the denominator from being zero; it can be taken as... ; This represents the function that takes the maximum value. This represents a truncation function, used to restrict the calculation result to a specific range. Based on this, when there are still many vacant spaces in underground parking lots and the pressure on surface parking lots increases, the priority access intensity for underground parking lots increases, and subsequent ant colony searches tend to choose underground parking lot paths.

[0058] It should be noted that, The item is the underground remaining capacity item, which represents the proportion of empty spaces in the underground parking lot that can accommodate vehicles. The larger this item is, the greater the underground capacity potential. The term is a ground oversaturation penalty term, which represents the severity of the ground parking lot exceeding its own capacity. This term is positive only when the occupancy exceeds the capacity, thereby triggering stronger guidance for underground parking.

[0059] 3) Regarding the first Each office functional area, read the first The standardized occupancy rate of on-street parking, the standardized heat value of historical illegal parking, and the standardized value of road network service level for each scheduling period. The office functional area is in the first The standardized occupancy rate of on-street parking during each scheduling period is recorded as follows: , Used to characterize the saturation level of on-street parking spaces, with a value range of [value range missing]. Historical standardized thermal values ​​of illegal parking Used to characterize the historical clustering intensity of illegal parking incidents; Standardized value of road network service level. Used to characterize the operating status of surrounding roads.

[0060] In practice, the standardized occupancy rate of on-street parking is obtained by dividing the number of occupied on-street parking spaces by the total number of on-street parking spaces. If on-street parking adopts segmented charging or zoned management, the standardized occupancy rate of on-street parking can be calculated first by segment, and then weighted averaged by the number of parking spaces to obtain the rate at the level of office functional areas. .

[0061] In one embodiment, for example, if there are three road segments with on-street parking spaces within the j-th office functional area, with road segment 1 having a total of 20 parking spaces, 18 of which are occupied (occupancy rate of 0.9); road segment 2 having a total of 30 parking spaces, 24 of which are occupied (occupancy rate of 0.8); and road segment 3 having a total of 40 parking spaces, 30 of which are occupied (occupancy rate of 0.75), then the standardized on-street parking occupancy rate for this area is... .

[0062] 4) Calculate the spillover risk elasticity index between on-street and off-street areas based on the standardized occupancy rate of on-street parking, the standardized heat map value of historical illegal parking, and the standardized value of road network service level. The spillover risk elasticity index between on-street and off-street areas is denoted as... , Indicates the first The office functional area is in the first The sensitivity of continuing to allocate on-street parking resources during a given scheduling period to illegal parking and deterioration of road conditions is categorized into values ​​ranging from [value range missing]. .

[0063] In one implementation, the on-street parking saturation level, the intensity of historical illegal parking accumulation, and road operational pressure are weighted and summed to obtain the on-street spillover risk elasticity index, which is calculated as follows: ; in, This represents the weight of on-street occupancy rate, used to control the impact of on-street parking saturation on the elasticity index of spillover risk from on-street parking, and can be taken as 0.45; This represents the historical illegal parking heat map weight, used to control the impact of the intensity of historical illegal parking clusters on the spillover risk elasticity index, and can be taken as 0.35; This represents the weight of the road network service level, used to control the impact of road operating pressure on the elasticity index of spillover risks inside and outside the road network, and can be taken as 0.20; and satisfies... ; This represents a truncation function, used to restrict the calculation result to a specific range. Interval. Because... A larger value indicates a better road condition, therefore it is used... It represents the pressure on road operation, and the higher the elasticity index of spillover risks inside and outside the road is when the road condition is worse.

[0064] It should be noted that, The item is the on-street saturation risk item, which reflects the current saturation level of on-street parking spaces. The higher the occupancy rate, the higher the immediate risk of overflow and illegal parking. The item is the historical inertia risk of illegal parking, which reflects the intensity of illegal parking incidents in the area in the past. The more historical illegal parking there is, the worse the parking order in the area is, and the greater the risk of continuing to increase on-street parking. This item represents the vulnerability risk of the road network, mapping the level of service of the road network to operational pressure; the more congested the roads (…), the lower the risk level. The smaller the value, the larger the value, indicating that the road's ability to handle new on-street parking interference is weaker and the risk of spillover is higher.

[0065] 5) The underground priority access intensity and the spillover risk elasticity index between roads and other areas are combined into a collaborative governance factor vector. The collaborative governance factor vector is denoted as... , It refers to the first The office functional area is in the first The governance-oriented feature vector for each scheduling period can be represented as: It is used to simultaneously provide guidance on underground parking priority and on-street parking risk mitigation for subsequent ant colony search processes.

[0066] In one embodiment, as an example, if the remaining parking space ratio in the underground parking lot of a certain office function area is 0.65 and the oversaturation ratio in the surface parking lot is 0.20, then... , The underground priority access intensity is approximately If the standardized occupancy rate of on-street parking in the area designated for office functions is 0.90, the standardized heat map value of historical illegal parking is 0.80, and the standardized value of road network service level is 0.40, then take... , , The spillover risk elasticity index is then... This indicates that there is a high risk of continuing to allocate on-street parking spaces in this office function area, and subsequent ant colony searches should reduce the probability of selecting on-street parking space paths.

[0067] like Figure 3 As shown in the bar chart, the intensity of underground priority access and the elasticity index of spillover risk between roads and outside roads are analyzed for each office function area, showing the comparison of the two governance characteristic indicators for each office function area; the horizontal axis is the index of the office function area (dimensionless), and the vertical axis is the index value (dimensionless). Figure 3 Blue bars represent the intensity of underground priority access, and red bars represent the elasticity index of spillover risk from roads. The experiment shows that the governance characteristics of different regions are significantly different. Some regions have high intensity of underground priority access and low spillover risk, while others are the opposite, proving that the collaborative governance factor vector can reflect the differentiated governance needs of regions.

[0068] It should be noted that the underground priority access intensity and the spillover risk elasticity index can expand parking resource allocation from a capacity-oriented approach to a governance-oriented approach. The underground priority access intensity enables the system to proactively increase the proportion of underground parking guidance when there is sufficient remaining underground capacity or when surface parking pressure increases; the spillover risk elasticity index enables the system to automatically suppress the allocation of on-street parking spaces when the risk of on-street parking is high, thereby reducing the spillover of illegal parking and the deterioration of road operation.

[0069] S2.3 Initialization of hierarchical directed ant colony search graph In the multi-target ant colony search process, it is necessary to express in the search space which residential function departure area a vehicle originates from, which office function destination area it travels to, which type of parking facility it enters, and how many vehicles are allocated. If search nodes are directly established for each vehicle and each parking space, the search scale will be too large, which is not conducive to real-time scheduling. This step encodes the residential function departure area, the office function destination area, and the parking facility type together as nodes in a hierarchical directed ant colony search graph, and uses the vehicle guidance unit as the minimum allocation object for ants to construct a scheduling scheme. The specific steps are as follows: 1) Construct a hierarchical directed ant colony search graph. The hierarchical directed ant colony search graph is denoted as... , It refers to the first A directed graph structure used for ant colony search during a scheduling period, comprising a set of nodes and a set of feasible directed edges. Nodes in the graph are indexed by their residential function originating region. Office Function Purpose Area Index Parking facility type index Together, they determine a corresponding candidate parking allocation combination.

[0070] In practical implementation, the dispatching system first generates underground parking nodes, surface parking nodes, and on-street parking space nodes by combining each residential departure area with an office destination area; then, it performs feasibility screening of the nodes based on the remaining capacity of parking facilities, gate access permissions, reservation rules, traffic restriction strategies, and management strategies. If the... The first office functional area If a parking facility has zero remaining parking spaces, or the vehicle lacks access permissions, or there is a conflict with the reservation rules, the corresponding node is marked as an impassable node; impassable nodes do not participate in ant path construction. Based on this, the hierarchical directed ant colony search graph only retains candidate paths that can form an executable scheduling scheme.

[0071] 2) Divide the demand for vehicles to be allocated into vehicle guidance units. A vehicle guidance unit is the smallest group of vehicles used in ant colony search to construct a parking allocation scheme. Starting from the residential function area to the first The office functional area is in the first The number of vehicle guidance units per scheduling period is denoted as , This is used to determine how many parking facility selections the ant needs to make for this origin-destination combination.

[0072] In one implementation, vehicle guidance units can be divided according to a proportional granularity. For example, with a minimum adjustment granularity of 5%, each departure-destination combination can be divided into a maximum of 20 vehicle guidance units. With a minimum adjustment granularity of 5%, each vehicle guidance unit corresponds to 4 vehicles. Ants indirectly form the vehicle guidance ratios for underground parking lots, surface parking lots, and on-street parking spaces by selecting parking facility types for each of the 20 vehicle guidance units. By dividing the system into vehicle guidance units, direct searching of continuous ratio variables can be avoided, reducing the complexity of the ant colony search.

[0073] 3) Calculate heuristic information values ​​based on the time-sharing potential index, underground priority access intensity, and spillover risk elasticity index. The heuristic information values ​​are denoted as... , Indicates the first The scheduling period starts from the first The residential departure area leads to the first Select the first office functional area. The heuristic information value for parking facilities is used to guide ants to prioritize parking allocation paths with high sharing potential, high underground utilization value, and low risk of spillover from on-street to off-street areas. The calculation method is expressed as follows: ; in, Represents the underground parking lot indicator variable, when the... The value is 1 when the parking facility is an underground parking lot, and 0 otherwise. This is used to control whether the underground priority access intensity is included in the heuristic information calculation. Represents the variable indicating on-street parking spaces, when the first... The value is 1 when the parking facility is an on-street parking space, and 0 otherwise. This is used to control whether the spillover risk elasticity index between on-street and off-street parking facilities participates in the heuristic information calculation. This represents the weight of the time-sharing potential, used to control the influence of the time-sharing potential index on path selection, and can be set to 1.2. This represents the underground priority access weight, used to control the enhancement intensity of underground priority access intensity on underground parking path, and can be set to 1.0; This represents the weight for suppressing spillover risks inside and outside the road, used to control the intensity of the suppression of the spillover risk elasticity index on the in-street parking space path, and can be taken as 1.5.

[0074] It should be noted that, The item is a time-sharing guidance item, which uses the time-sharing potential of residential areas and office areas as the base value. The higher the potential, the greater the overall attractiveness of choosing the route from the residential area to the office area for parking. This item is an underground priority enhancement item, only applicable if the target parking facility... When it is an underground parking lot ( This feature is activated only after [certain conditions are met], and the underground priority access intensity is [specifically determined]. The larger the value, the stronger the effect of this feature on selecting routes to underground parking lots; The term is an overflow suppression term for roadside and internal / external parking facilities, which only applies when the target parking facility... When it is an on-street parking space ( This is activated only after [the event], and the spillover risk elasticity index [is also considered]. The larger the value, the stronger the inhibitory effect of this property on selecting on-street parking routes.

[0075] 4) Write the heuristic information values ​​into the feasible directed edges of the hierarchical directed ant colony search graph. For underground parking paths, the higher the underground priority access strength, the larger the heuristic information value; for on-street parking paths, the higher the on-street spillover risk elasticity index, the smaller the heuristic information value; for surface parking paths, the underground priority access strength and on-street spillover risk elasticity index do not directly amplify or inhibit the path, but are mainly adjusted through the time-sharing potential index, capacity pressure perception factor, and subsequent multi-objective evaluation.

[0076] In one embodiment, for example, if the first The residential function starting area and the first The potential index for staggered sharing of office functional areas is relatively high, and the first If the remaining capacity of the underground parking lot in an office function area is sufficient, the heuristic information value of the underground parking lot path will increase, making it easier for ants to assign vehicle guidance units to the underground parking lot. If the first If the standardized occupancy rate of on-street parking and the standardized heat value of historical illegal parking are both high in the office functional area, the heuristic information value of on-street parking space path will decrease, and the probability of ants choosing on-street parking space path will be suppressed.

[0077] It should be noted that the heuristic information that integrates collaborative features simultaneously expresses the value of cross-regional sharing, the value of prioritizing the use of underground parking, and the need to mitigate the risks of on-street parking. This enables ant colony search to have a governance orientation from the initial stage, reducing ineffective searches and improving the quality of multi-objective compromise solutions.

[0078] In a specific implementation, S3's multi-target ant colony optimization based on capacity pressure and congestion perception is as follows: When coordinating parking resource scheduling across multiple areas, it is necessary to simultaneously improve parking turnover efficiency, reduce surface parking oversaturation, increase underground parking utilization, and mitigate the risk of on-street parking spillover. Since these objectives conflict, improving underground parking utilization may increase detour distances for some vehicles, while reducing on-street parking spillover may require limiting the allocation ratio of on-street parking spaces. This step constructs an ant scheduling scheme on a hierarchical directed ant colony search graph, introduces a capacity pressure sensing factor and a capacity constraint self-repair mechanism to reduce infeasible solutions, and maintains Pareto solution diversity through a congestion-aware pheromone update mechanism. The specific steps are as follows: S3.1 Construction of an ant scheduling scheme combining capacity constraint self-healing Since the ant colony algorithm is prone to generating overcapacity infeasible solutions during solution construction, this step designs a capacity pressure-aware factor to reduce the probability of selecting parking facilities with insufficient remaining capacity when ants make probabilistic selections. Furthermore, a capacity constraint self-repair mechanism is introduced to re-allocate vehicles exceeding facility capacity in the generated initial scheduling scheme according to predetermined rules, thereby improving the executability of candidate scheduling schemes. The specific steps are as follows: 1) Initialize the ant colony and pheromone concentration. Ant indexing is used. express, Used to distinguish different ants participating in parallel search, with a value range of [value missing]. arrive ; This represents the total number of ants, used to control the number of candidate scheduling schemes generated in each iteration; a value of 50 is acceptable. The iteration index is used... express, Used to identify the search round in the ant colony optimization process, with a value range of [value missing]. arrive ; This represents the maximum number of iterations, which can be 200.

[0079] Specifically, pheromone concentration is denoted as , Indicates the first The scheduling period is the first In the nth iteration, from the... The residential departure area leads to the first Select the first office functional area. The pheromone concentration of parking facility paths is used to record the cumulative preference of historical high-quality scheduling schemes for the current path. In the specific implementation, the pheromone concentration of all feasible paths can be initialized to the same constant (e.g., 1.0); no pheromone concentration is set for impassable nodes and impassable directed edges.

[0080] 2) Calculate the capacity pressure perception factor based on the number of remaining parking spaces and the demand for vehicles to be allocated. The capacity pressure perception factor is denoted as... , Indicates the first The scheduling period starts from the first The residential departure area leads to the first Select the first office functional area. The capacity of a parking facility is considered to reduce the probability that a parking facility with insufficient remaining capacity will be selected.

[0081] In the specific implementation, the first The first office functional area The number of remaining parking spaces for Class 1 parking facilities and Class 2 parking facilities Starting from the residential function area to the first By comparing the demand for vehicles to be allocated in each office functional area, a capacity pressure perception factor is obtained. ,Right now .in, Indicates the first The first office functional area Class II parking facilities The number of remaining berths for each scheduling period; Indicates the first The scheduling period starts from the first The residential departure area leads to the first The number of vehicles to be allocated in each office functional area; This represents the minimum value function. If the number of remaining parking spaces is greater than or equal to the demand for vehicles to be allocated, the capacity pressure perception factor is 1, indicating that the parking facility has sufficient capacity; if the number of remaining parking spaces is significantly less than the demand for vehicles to be allocated, the capacity pressure perception factor decreases, indicating that the parking facility is not suitable for accommodating too many vehicles.

[0082] like Figure 4 As shown in the bar chart, the capacity pressure perception factor of different parking facility types in different office function areas is analyzed, which shows the capacity of underground parking lots, surface parking lots and on-street parking spaces in different office function areas to accommodate new vehicles; the horizontal axis is the index of office function area (dimensionless), and the vertical axis is the capacity pressure perception factor (dimensionless). Figure 4 The experiment used a grouped bar chart to distinguish between three types of facilities: underground parking lots, surface parking lots, and on-street parking spaces. The experiment showed that the capacity pressure perception factor of underground parking lots was generally higher than that of on-street parking spaces, proving that the capacity pressure perception factor can effectively reduce the probability of facilities with insufficient capacity being selected.

[0083] 3) Calculate the state transition probability based on pheromone concentration, heuristic information value, and capacity pressure sensing factor. The state transition probability is denoted as... , Indicates the first Only ants in the first The next iteration, the... The scheduling period will be the first one. Starting from the residential function area to the first The vehicle guidance unit for the office functional area is assigned to the first The probability of a parking facility of this type.

[0084] In one implementation, it can be based on a set of parking facility types. The state transition probability is calculated for each parking facility type, and the calculation method is expressed as follows: ; in, Indicates the first The scheduling period starts from the first The residential departure area leads to the first A set of parking facility types available when choosing an office function area; This represents the index of the candidate parking facility type, used to traverse the set of selectable parking facility types; This represents the pheromone influence index, used to control the strength of the influence of pheromone concentration on the state transition probability, and can be set to 1.0; This represents the heuristic influence index, used to control the strength of the influence of heuristic information values ​​on state transition probabilities, and can be set to 2.0; This represents the capacity pressure impact index, used to control the strength of the influence of the capacity pressure perception factor on the state transition probability, and can be set to 1.5. If the denominator is 0, it means that none of the candidate parking facility types can accommodate vehicles. In this case, the corresponding vehicle guidance unit is marked as an unassigned vehicle guidance unit, and a penalty term is included in the subsequent multi-objective evaluation.

[0085] 4) Select the parking facility type for each vehicle guidance unit in turn using each ant, and generate an initial scheduling plan. The initial scheduling plan is denoted as... , It refers to the first Only ants in the first The next iteration, the... The vehicle allocation results generated for each scheduling period include the number of vehicles allocated to underground parking lots, surface parking lots, and on-street parking spaces corresponding to each combination of residential departure area and office destination area.

[0086] In practical implementation, for each combination of residential departure area and office destination area, ants use a roulette wheel approach to allocate vehicle guidance units one by one according to the state transition probability. If the... Starting from the residential function area to the first The required number of vehicles to be allocated for each office functional area is 80, and the minimum adjustment granularity is 5%. Therefore, each vehicle guidance unit corresponds to 4 vehicles. If 12 of the 20 vehicle guidance units choose underground parking, 6 choose surface parking, and 2 choose on-street parking, then 48 vehicles will be allocated to underground parking, 24 to surface parking, and 8 to on-street parking, corresponding to vehicle guidance ratios of 60%, 30%, and 10%, respectively. Through the vehicle guidance unit selection process, the scheduling scheme can express proportional allocation while maintaining discrete executability.

[0087] 5) Perform capacity constraint self-repair on the initial scheduling scheme. If the initial scheduling scheme makes the first... The first office functional area The number of vehicles allocated to this type of parking facility exceeds If the excess vehicles are recorded as vehicles to be repaired, then a replacement will be selected from other available parking facility types within the same office function area.

[0088] In practice, the number of excess vehicles is first calculated and deducted from the overcapacity parking facility type. Then, among other feasible parking facility types within the same office function area, vehicles to be repaired are allocated based on heuristic information values ​​and the number of remaining parking spaces. For candidate parking facility types... , No. The scheduling period starts from the first The residential departure area leads to the first Select the first office functional area. The heuristic information value for parking facilities is denoted as , No. The scheduling period is the first The first office functional area The number of remaining parking spaces for this type of parking facility is denoted as Calculate the priority value for acceptance. Vehicles awaiting repair will be accommodated in descending order of priority, with the number of available spaces not exceeding the remaining capacity of that parking facility type. If other parking facility types within the same office function area cannot fully accommodate the vehicles awaiting repair, the remaining vehicles will be transferred to adjacent office function area parking facilities with higher time-sharing potential index and that meet travel constraints. If no suitable alternatives are found, these vehicles will be marked as unassigned and penalized in the on-street parking spillover rate evaluation.

[0089] In one embodiment, as an example, if an initial scheduling scheme is directed to the first... If an underground parking garage in an office function area is allocated 90 spaces, but only 70 spaces remain, there are 20 excess vehicles. The system first corrects the allocated number of vehicles in the underground garage to 70, then checks the surface parking garage and on-street parking spaces in the same office function area. If there are 15 spaces remaining in the surface parking garage and 10 spaces remaining in the on-street parking garage, but the spillover risk elasticity index is high, the system prioritizes transferring the 15 vehicles to the surface parking garage. The remaining 5 vehicles are then transferred to the underground parking garage in an adjacent office function area or marked as unallocated vehicles according to the risk mitigation strategy. This self-correcting process reduces overcapacity issues and makes the final scheduling plan closer to an engineering-executable state.

[0090] 6) Perform a multi-objective evaluation on the repaired ant scheduling scheme. The ant scheduling scheme is denoted as... , Refers to the initial scheduling scheme Candidate scheduling schemes are generated after capacity constraint self-repair. In specific implementation, a complete initial scheduling scheme is first constructed based on pheromones, heuristic information, and capacity pressure sensing factors. Then, capacity constraint self-repair is performed on the initial scheduling scheme to obtain candidate scheduling schemes. Finally, a multi-objective evaluation is conducted on the ultimate impact of the candidate scheduling scheme on all parking facilities. The multi-objective evaluation includes parking turnover efficiency, surface parking oversaturation, underground parking utilization, and spillover risk of on-street parking.

[0091] In one implementation, the parking turnover efficiency target can be calculated based on the historical turnover coefficient of the parking facility and the number of vehicles allocated. The first scheduling period The first office functional area The historical turnover coefficient of parking facilities is denoted as , Used to characterize the parking facility in the The capacity of a single berth to complete vehicle entry, exit, and reuse within a given scheduling period can be obtained from historical vehicle entry and exit records. The parking turnover efficiency target for the ant scheduling scheme corresponding to only one ant is denoted as: , This is used to measure the expected parking turnover efficiency of vehicles after the scheduling plan is implemented; the higher the target, the better. The calculation can be based on the number of allocated vehicles and the historical turnover coefficient of the corresponding parking facilities. The calculation method is expressed as follows: ; in, Indicates the first The scheduling period is the first In the ant scheduling scheme, starting from the ant number... The residential departure area leads to the first Each office functional area was assigned to the first The number of vehicles in this type of parking facility; The larger the value, the higher the parking turnover efficiency.

[0092] Furthermore, the first The ground parking oversaturation target for each ant in the ant scheduling scheme is denoted as: , This is used to measure the average severity of the ground parking lots in each office function area exceeding their capacity limit after the ant scheduling scheme is implemented; the goal is to minimize this severity. The calculation can be based on the average percentage of vehicles exceeding their parking capacity in each area's surface parking lots after the plan is implemented. The calculation method is expressed as follows: ; in, Indicates the first The scheduling period is the first In the ant scheduling scheme, starting from the ant number... The residential departure area leads to the first The number of vehicles allocated to each office functional area and the number of vehicles assigned to the ground parking lot; The larger the value, the more severe the oversaturation of the surface parking lot.

[0093] Furthermore, the first The underground parking utilization target for each ant in the ant scheduling scheme is denoted as: , This measure is used to gauge the effectiveness of underground parking facilities, with the goal of maximizing utilization. The calculation can be based on the proportion of the total number of vehicles allocated to the underground parking lot to the total number of remaining parking spaces in the underground parking lot. The calculation method is expressed as follows: ; in, Indicates the first The scheduling period is the first In the ant scheduling scheme, starting from the ant number... The residential departure area leads to the first The number of vehicles allocated to the underground parking lot is determined by the functional purpose of each office area. The larger the value, the more fully underground parking resources are utilized.

[0094] Furthermore, the first The on-street parking spillover risk target for each ant in the ant scheduling scheme is denoted as: , This measure is used to assess the extent to which the ant scheduling scheme causes on-street parking spillover or illegal parking risks, with the goal of minimizing these risks. The calculation can be performed by multiplying the number of vehicles allocated to each on-street parking space by the risk elasticity index of that area, adding the penalty for unallocated vehicles, and then normalizing the total demand. The calculation method is expressed as follows: ; in, Indicates the first The scheduling period is the first In the ant scheduling scheme, starting from the ant number... The residential departure area leads to the first The number of vehicles allocated to on-street parking spaces within each office functional area; Indicates the first The scheduling period is the first The number of vehicles that, after capacity constraint self-repair in the ant scheduling scheme, still cannot find a legal parking facility to take over, is represented by the number of vehicles allocated for on-street parking. By multiplying the number of vehicles allocated for on-street parking by the on-street spillover risk elasticity index, on-street parking allocation in high-risk areas can be subject to stronger penalties.

[0095] 7) Perform fast non-dominated sorting on all ant scheduling schemes to obtain multiple Pareto fronts. The non-dominated level is denoted as... , Indicates the first Only ants in the first The non-dominated levels in the next iteration are used to measure the overall multi-objective superiority of the ant scheduling scheme; the smaller the value, the better the scheme. The crowding distance is calculated for ant scheduling schemes within the same Pareto front, and denoted as [missing information]. , Indicates the first Only ants in the first The sparsity of the target space in the next iteration; a larger value indicates that the ant scheduling scheme is located in a less covered compromise region.

[0096] like Figure 5 As shown, the parallel coordinates of the normalized values ​​of the Pareto scheduling scheme set are analyzed to show the normalized performance of multiple non-dominated scheduling schemes obtained by ant colony optimization on four optimization objectives. The horizontal axis represents the parking turnover efficiency objective (dimensionless), the surface parking oversaturation objective (dimensionless), the underground parking utilization level objective (dimensionless), and the on-street parking spillover risk objective (dimensionless), while the vertical axis represents the normalized objective value (dimensionless). Figure 5 Different colored broken lines represent different candidate scheduling schemes. Experiments show that there are obvious trade-offs between different schemes for each objective, proving that multi-objective ant colony optimization can provide diverse Pareto compromise schemes.

[0097] It should be noted that the capacity pressure sensing factor and the capacity constraint self-correction mechanism can bring capacity constraints forward to the solution construction stage and perform secondary correction after solution construction. The capacity pressure sensing factor can reduce the probability of ants selecting parking facilities with insufficient capacity, and the capacity constraint self-correction mechanism can transfer any remaining overcapacity allocations to other feasible parking facilities, thereby improving the executability and optimization efficiency of candidate scheduling schemes.

[0098] S3.2, Pheromone Update In multi-objective ant colony optimization, updating pheromones solely based on non-dominated levels can easily concentrate search results in areas where a particular target has an advantage. This may lead to an overemphasis on solutions with high underground parking utilization, while neglecting compromise solutions with lower surface parking oversaturation and on-street parking spillover risks. This step adjusts pheromone release and volatilization intensity using congestion distance and Pareto front distribution entropy. The specific steps are as follows: 1) Regarding the first In the next iteration, all ant scheduling schemes are subjected to non-dominated sorting, and the congestion distance is calculated within each Pareto front. In practice, for each Pareto front, the parking turnover efficiency target is considered separately. Target for oversaturation of ground parking Underground parking utilizes horizontal targets and on-street parking spillover risk target The ant scheduling schemes are sorted; after normalizing each objective, the objective difference between adjacent ant scheduling schemes is calculated, and multiple objective differences are accumulated to form the congestion distance. Ant scheduling schemes located at the objective boundary can be set with a larger congestion distance to preserve the objective spatial boundary scheme.

[0099] In one embodiment, as an example, suppose that after a certain iteration, the first Pareto front has three options: Option 1, Option 2, and Option 3. After normalizing each objective, each objective is sorted and the crowding distance component is calculated. If for objective... The values ​​are sorted from smallest to largest as 1, 2, and 3. After normalization, their values ​​are 0.1, 0.5, and 0.9 respectively. Therefore, Scheme 1... The congestion component is infinite (boundary), and the component of scheme 2 is... The component of scheme 3 is infinite (boundary); if for the target The values, sorted from smallest to largest, are 3, 1, and 2. After normalization, their values ​​are 0.2, 0.6, and 0.8 respectively. Therefore, scheme 3... The congestion component is infinite, and the component of scheme 1 is... The component of scheme 2 is infinite.

[0100] 2) Construct an elite ant set based on the non-dominated hierarchy and crowding distance. The elite ant set is denoted as […]. , Indicates the first In the next iteration, the set of ants used to release pheromones is selected with priority from the first Pareto front where the crowding distance is relatively large, and a small number of ants are randomly selected from the second or third Pareto front to balance convergence and exploration.

[0101] In practical implementation, the top 30% of ants by crowding distance from the first Pareto front can be selected to join the elite ant set, and then 10% of ants can be randomly selected from the second Pareto front to join the elite ant set. If the number of ants in the first Pareto front is insufficient, the number can be expanded to the second Pareto front. In this way, pheromone updates will not rely solely on a single optimal region, but can retain multiple target compromise directions.

[0102] It should be noted that after the fast non-dominated sorting, all scheduling schemes are divided into different non-dominated levels. The first Pareto front refers to the set of schemes where no other scheme is non-inferior to it in all objectives and superior to it in at least one objective; it represents the current optimal compromise solution. After removing the first Pareto front from all schemes, the remaining schemes are non-dominated sorted again, and the resulting set of non-dominated schemes is the second Pareto front. This process continues to obtain subsequent levels.

[0103] 3) Determine the pheromone evaporation coefficient based on the Pareto front entropy. The Pareto front entropy is denoted as... , Indicates the first The uniformity of the Pareto front solution in the target space during each iteration; a larger value indicates a more balanced Pareto front distribution; the ideal uniform distribution entropy is denoted as... , Used to normalize the entropy of the Pareto front distribution.

[0104] In practical implementation, the ant scheduling scheme in the first Pareto front is divided according to the target spatial location. Each interval This indicates the number of intervals for the statistical analysis of the target spatial distribution, which can be 10; the target spatial distribution interval index is used... express, Used to iterate through all partitioned intervals, with values ​​ranging from 1 to... ;No. The proportion of schemes in each interval is denoted as . , Used to characterize falling into the first Pareto front. The proportion of schemes in each interval. The Pareto front distribution entropy is calculated as follows: Furthermore, the entropy of an ideal uniform distribution can be taken as... When the ant scheduling schemes are concentrated in a few intervals, the Pareto front distribution entropy is low, indicating that the search results are clustered; when the ant scheduling schemes are more evenly distributed across multiple intervals, the Pareto front distribution entropy is high, indicating that the search coverage is more comprehensive.

[0105] Furthermore, the pheromone volatile coefficient is denoted as... , Indicates the first The pheromone evaporation coefficient of the next iteration is used to weaken the excessive influence of historical paths on the current search, and is calculated as follows: ; in, This represents the lower limit of the pheromone evaporation coefficient, used to ensure that high-quality path information can be preserved; it can be set to 0.10. This represents the upper limit of the pheromone evaporation coefficient, used to enhance the exploration capability when the Pareto front distribution is uneven, and can be set to 0.45. When the Pareto front distribution entropy is low, the pheromone evaporation coefficient increases, and the system evaporates existing pheromones more quickly to escape local aggregation; when the Pareto front distribution entropy is high, the pheromone evaporation coefficient decreases, and the system retains existing high-quality paths to achieve stable convergence.

[0106] 4) Update the pheromone concentration for each feasible directed edge in the hierarchical directed ant colony search graph. The directed edge index uses... express, Used to identify a feasible transition path in a hierarchical directed ant colony search graph; A feasible directed edge is on the 1st The pheromone concentration in the next iteration is denoted as... . No. Only ants in the first The complete scheduling path constructed in the next iteration is denoted as . , This is used to record all parking allocation paths chosen by the ant.

[0107] In one implementation, the congestion distance is first normalized to... The interval is used to obtain the normalized congestion distance. ; Used to characterize the in In the nth iteration The relative sparsity of the ant scheduling scheme in the target space. Then, the update method for pheromone concentration is expressed as: ; in, Indicates the first In the nth iteration The pheromone concentration of each feasible directed edge; The path indicates that it contains an indicator variable, when the first... A feasible directed edge belongs to The value is 1 if the condition is met, otherwise it is 0. This represents the base pheromone strength, used to control the extent to which a single elite ant enhances the path pheromone, and can be set to 1.0. Through this update method, ant scheduling schemes with superior non-dominant levels and located in sparsely populated areas of the target space will release more pheromones, making it easier for the next round of search to explore under-covered compromise areas.

[0108] It should be noted that, The term is the pheromone retention term, which represents the portion of pheromone that remains after evaporation in the previous iteration, preserving the experience of historical high-quality paths; The term is the pheromone enhancement term, which represents the pheromone released by elite ants along the paths they traverse in this iteration. The intensity of the release is related to the non-dominated level and the crowding distance of the ant's scheme. Schemes with a better non-dominated level (smaller value) and located in a sparser region of the target space (larger crowding distance) will receive stronger pheromone enhancement, thereby guiding subsequent searches towards higher quality and more diverse directions.

[0109] 5) Apply boundary restrictions to the updated pheromone concentration. The lower limit of pheromone concentration is denoted as... , To prevent certain feasible paths from being completely lost due to excessively low pheromone levels, a value of 0.1 can be used; the upper limit of pheromone concentration is denoted as... , To prevent premature convergence of the search due to excessively high pheromone levels on certain feasible paths, a value of 10 can be used. If the updated pheromone concentration is lower than... Then set it to If the updated pheromone concentration is higher than Then set it to .

[0110] It should be noted that the crowding-aware pheromone update mechanism can prevent multi-target ant colony search from converging to only a single preference region. By incorporating the non-dominated level, crowding distance, and Pareto front distribution entropy into the pheromone update process, it is possible to maintain the diversity of compromise solutions while ensuring the quality of the scheme, thereby improving the adaptability of the final scheduling scheme to different governance scenarios.

[0111] S3.3 Selection of Pareto Compromise After ant colony optimization, multiple Pareto scheduling schemes are obtained. These schemes make different trade-offs regarding parking turnover efficiency, surface parking oversaturation, underground parking utilization, and the risk of spillover from on-street parking. Since management objectives change with time of day and events, morning rush hour focuses more on mitigating surface congestion and suppressing spillover from on-street parking, while off-peak hours focus more on underground parking utilization and parking turnover efficiency. This step selects the optimal compromise scheme from the Pareto scheduling scheme set based on dynamic governance preferences. The specific steps are as follows: 1) When the ant colony reaches its maximum number of iterations. If, after several consecutive rounds, the Pareto scheduling scheme set no longer shows significant improvement, ant colony optimization is stopped, and a Pareto scheduling scheme set is collected. The Pareto scheduling scheme set is denoted as... , It refers to the first Each scheduling period is a set of non-dominated scheduling schemes obtained by ant colony optimization; the candidate Pareto scheduling scheme index is used for... express, Used to iterate through the different scheduling schemes in the Pareto scheduling scheme set.

[0112] In practical implementation, the external non-dominated archives saved throughout the entire iteration process are used as the Pareto scheduling scheme set. The external non-dominated archives are used to store high-quality non-dominated scheduling schemes that have appeared in each iteration, and to delete dominated scheduling schemes when new scheduling schemes are added, thereby improving the candidate coverage of the final compromise decision.

[0113] 2) Convert the 4-objective values ​​of each candidate Pareto scheduling scheme into fuzzy membership degrees. The target index uses... express, Used to distinguish the four scheduling optimization objectives, with a value range of [value range missing]. arrive ;No. The candidate Pareto scheduling schemes are in the... The fuzzy membership degree on the project label is denoted as , This is used to map target values ​​with different dimensions and optimization directions to a unified evaluation scale, with a value range of [value range missing]. .

[0114] In practical implementation, for benefit-oriented goals such as parking turnover efficiency and underground parking utilization level, the larger the target value, the higher the fuzzy membership degree; for cost-oriented goals such as surface parking oversaturation and on-street parking spillover risk, the smaller the target value, the higher the fuzzy membership degree. For the first... The project objective is a benefit-oriented objective, and the fuzzy membership degree is calculated as follows: ; Or, for the first The project objective is a cost-based objective, and the fuzzy membership degree is calculated as follows: .in, Indicates the first The candidate Pareto scheduling schemes are in the... The target value on the project label; Indicates the first Pareto scheduling scheme in the set of Pareto schemes. The maximum target value of the project; Indicates the first Pareto scheduling scheme in the set of Pareto schemes. The minimum target value of the project. If and If they are the same, the fuzzy membership degree of the target is uniformly set to 1, indicating that the candidate Pareto scheduling schemes are indistinguishable on the target.

[0115] 3) Set the dynamic preference weights corresponding to the current governance scenario. The dynamic preference weight of the project target is denoted as , Used to characterize the manager's attitude towards the first [unclear] in the current governance scenario. The importance attached to the project target, meeting the requirements and Dynamic preference weights can be determined by time-based strategies, event-based strategies, or manual control strategies.

[0116] In one embodiment, for example, during the morning peak hours, the target weights for surface parking oversaturation can be set to 0.35, on-street parking spillover risk to 0.30, underground parking utilization to 0.20, and parking turnover efficiency to 0.15, enabling the system to prioritize scheduling schemes that alleviate surface parking pressure and suppress on-street parking spillover. During off-peak hours, the target weights for underground parking utilization and parking turnover efficiency can be increased, enabling the system to guide more vehicles into underground parking lots and improve the utilization efficiency of existing parking resources. During the closing hours of large events, the target weights for parking turnover efficiency and on-street parking spillover risk can be increased, enabling the system to prioritize rapid evacuation and on-street order.

[0117] 4) Calculate the overall satisfaction level based on fuzzy membership degree and dynamic preference weights, and select the optimal compromise solution. The overall satisfaction of the candidate Pareto scheduling schemes is denoted as . , This is used to measure the degree of fit between the candidate Pareto scheduling scheme and the current dynamic governance preferences, and is calculated as follows: The index of the optimal compromise solution is denoted as , The Pareto scheduling scheme used to identify the final selected and executed plan is calculated as follows: .in, This represents the maximum value indexing operator, used to select the scheduling scheme with the highest overall satisfaction from the Pareto scheduling scheme set. If multiple candidate Pareto scheduling schemes have the same overall satisfaction, the candidate Pareto scheduling scheme with the lower target value of on-street parking spillover risk is selected first; if the target value of on-street parking spillover risk is still the same, the candidate Pareto scheduling scheme with the lower target value of ground parking oversaturation is further selected to meet the safety priority principle of parking management.

[0118] like Figure 6As shown, the scatter plot of the multi-objective Pareto front (parking turnover efficiency and ground parking oversaturation) is analyzed, showing the distribution of all scheduling schemes on the parking turnover efficiency objective and the ground parking oversaturation objective; the horizontal axis is the parking turnover efficiency objective (dimensionless), and the vertical axis is the ground parking oversaturation objective (dimensionless). Figure 6 In the diagram, gray dots represent all scheduling schemes, blue dots represent the first Pareto front scheme, and red asterisks represent the selected optimal compromise scheme. Experiments show that the Pareto front presents a compromise relationship between two objectives, and improving parking turnover efficiency may be accompanied by an increase in the degree of ground parking oversaturation.

[0119] like Figure 7 As shown, the scatter plot of the multi-objective Pareto front (underground parking utilization level and on-street parking spillover risk) shows the distribution of all scheduling schemes on the underground parking utilization level objective and the on-street parking spillover risk objective; the horizontal axis is the underground parking utilization level objective (dimensionless), and the vertical axis is the on-street parking spillover risk objective (dimensionless). Figure 7 In the diagram, gray dots represent all scheduling schemes, green dots represent the first Pareto front scheme, and red asterisks represent the selected optimal compromise scheme. Experiments show that the Pareto front presents a compromise relationship between the two objectives, and improving the utilization level of underground parking may be accompanied by an increase in the risk of spillover from on-street parking.

[0120] It should be noted that the dynamic preference-driven Pareto compromise selection mechanism can connect multi-objective optimization results with actual governance strategies. By unifying different objective dimensions through fuzzy membership degrees and expressing governance priorities under different time periods and events through dynamic preference weights, the scheduling system can select the most feasible solution that best meets the current management needs from multiple non-dominated scheduling schemes.

[0121] In a specific implementation, the generation of the S4 parking guidance instruction and the rolling feedback update are as follows: After selecting the optimal compromise solution, the scheduling system needs to convert it into an executable parking guidance instruction and continuously update the input data and scheduling parameters for the next scheduling period based on the execution results. This step involves the coordinated execution of parking scheduling through guidance screens, mobile terminals, the parking management system, and the price adjustment system. The specific steps are as follows: 1) Decode the optimal compromise solution into a vehicle allocation ratio instruction. The vehicle allocation ratio instruction specifies the target ratio for vehicles entering underground parking lots, surface parking lots, and on-street parking spaces for each combination of residential departure areas and office destination areas. The scheduling period starts from the first The residential departure area leads to the first The target percentage of vehicles from each office functional area that are guided to the underground parking lot is denoted as follows: , No. The scheduling period starts from the first The residential departure area leads to the first The target percentage of vehicles from each office functional area are guided to the surface parking lot. The surface parking lot vehicle guidance percentage is recorded as follows: , No. The scheduling period starts from the first The residential departure area leads to the first The target percentage of vehicles in each office functional area that are guided to on-street parking spaces is denoted as: And satisfy .

[0122] In practical implementation, the scheduling system calculates the corresponding vehicle guidance ratio based on the number of vehicles allocated to each parking facility type in the optimal trade-off solution and the demand for vehicles to be allocated. For example, the... Starting from the residential function area to the first The required number of vehicles to be allocated for each office functional area is 80. The optimal compromise is to allocate 48 vehicles to the underground parking lot, 24 vehicles to the surface parking lot, and 8 vehicles to the on-street parking space. Therefore, the vehicle allocation ratio is 60% for the underground parking lot, 30% for the surface parking lot, and 10% for the on-street parking space.

[0123] 2) Generate cross-regional sharing matching instructions based on vehicle allocation ratio instructions. Cross-regional sharing matching instructions refer to recommending residential departure areas or adjacent parking areas with available parking spaces to vehicles when the off-peak sharing potential index meets the preset sharing trigger threshold, and providing available parking time periods, estimated travel time, estimated walking distance, and shared parking rules.

[0124] In the specific implementation, the shared trigger threshold is denoted as... , A value of 0.6 can be used to determine whether a cross-regional shared combination has sufficient execution value. If... If the number of remaining parking spaces in the corresponding parking facility meets the allocation demand, the dispatch system generates a cross-regional sharing matching instruction and displays the shared parking area, navigation entrance, permitted parking time period, and departure requirements to the user via mobile terminal. If the user accepts the shared parking recommendation, the vehicle will be included in the reservation or access quota of the target parking facility; if the user rejects the shared parking recommendation, the user acceptance rate parameter for that cross-regional combination will be reduced in the next round of feedback updates.

[0125] 3) Generate parking facility flow restriction instructions and access control parameters based on the optimal compromise solution. Parking facility flow restriction instructions refer to lowering the recommendation level, restricting new reservations, or tightening gate access for parking facilities approaching their capacity limits or with high risk. Access control parameters refer to the parameters that the parking management system generates in the [specific context needed for a complete translation]. The number of target vehicles allowed to enter the corresponding parking facility during each scheduling period.

[0126] In practical implementation, if the predicted number of occupied parking spaces in a surface parking lot is close to or exceeds its capacity, the dispatch system will lower the recommendation level of that surface parking lot on guidance screens and mobile terminals, and restrict new reservation requests. If the spillover risk elasticity index of an area where an on-street parking space is located is high, the dispatch system will reduce the target guidance ratio for on-street parking spaces and prompt "On-street parking spaces are scarce, please go to the underground parking lot first" through guidance screens. If an underground parking lot has a sufficient number of remaining parking spaces and the underground priority access intensity is high, the dispatch system will increase the recommendation priority of the underground parking lot and can link with the price adjustment system to provide parking discounts.

[0127] 4) Dispatch instructions are sent to guidance screens, mobile terminals, and the parking management system. Guidance screens are used to display recommended parking facilities, the number of remaining parking spaces, and driving directions at road entrances, parking lot entrances, and key diversion nodes; mobile terminals are used to push parking facility recommendations, navigation routes, estimated walking distances, price discounts, and reservation entrances to users; the parking management system is used to execute gate access, reservation locking, deduction of remaining parking spaces, and update of parking space release.

[0128] In practice, the scheduling system converts the target number of vehicles to be guided for each parking facility into a real-time recommended quota. When the recommended quota is not used up, the guidance screens and mobile terminals continuously recommend the parking facility; when the recommended quota is nearly exhausted, the recommendation level is lowered; when the recommended quota is exhausted or there are insufficient remaining parking spaces, the parking facility is removed from the candidate recommendation list. Through recommended quota control, secondary congestion caused by a large number of vehicles simultaneously flocking to the same parking facility can be avoided.

[0129] 5) Collect dispatch execution feedback and update the input for the next dispatch period. Dispatch execution feedback includes the actual number of vehicles entering, the actual number of vehicles leaving, parking facility occupancy status, user acceptance rate, on-street parking overflow events, illegal parking events, and road network operation status. User acceptance rate is denoted as... , Used to characterize the During the scheduling period, the user receives from the first... Starting from the residential function area to the first The first office functional area Recommended proportion of parking facilities of this type.

[0130] In practical implementation, if the user acceptance rate is lower than the set threshold (e.g., 0.4), it indicates that the recommended parking facility may have problems such as excessive detour distance, unreasonable price, or excessive walking distance. In the next scheduling period, the heuristic information weight of the corresponding path can be reduced. If the actual occupancy of a parking facility increases faster than the predicted value, the number of remaining parking spaces will be updated in the next scheduling period, and the impact index of the capacity pressure perception factor will be increased. If a new on-street parking spillover event occurs in an office function area, the historical standardized heat value of illegal parking and the on-street spillover risk elasticity index will be updated, thereby strengthening on-street parking suppression in subsequent scheduling.

[0131] 6) Implement rolling optimization closed loop. At the end of each scheduling period or when real-time data changes significantly, the scheduling system reconstructs the parking collaborative monitoring samples, recalculates the staggered-time sharing potential index, underground priority access intensity, and on-street spillover risk elasticity index, reinitializes or incrementally updates the hierarchical directed ant colony search graph, and executes the next round of multi-objective ant colony optimization. Through the rolling optimization closed loop, the scheduling system can continuously adjust the parking guidance plan according to changes in parking demand, parking space status, road network operation status, and user acceptance.

[0132] It should be noted that this invention forms a complete technical chain from the construction of parking collaborative monitoring samples, collaborative feature mining, hierarchical directed ant colony search graph initialization, multi-target ant colony optimization, Pareto compromise scheme selection to parking guidance execution feedback. Thus, it can simultaneously achieve staggered sharing, underground priority and on-street spillover management on the basis of capacity feasibility, making multi-area parking resource collaborative scheduling more suitable for complex urban parking scenarios.

[0133] In one embodiment, as an example, for multi-area parking coordination scheduling during the morning rush hour (8:00-8:15) in a city, if there are 12 residential departure areas and 8 office destination areas, during the current scheduling period, the system collects the time-series curve of standardized occupancy rate for each area, the number of remaining parking spaces in each parking facility, the demand for vehicles to be allocated, the standardized heat value of historical illegal parking, and the standardized value of road network service level. First, the staggered-time sharing potential index is calculated; the average standardized occupancy rate of a certain residential departure area in the current and adjacent periods is 0.25 (vacancy rate 0.75), and the current standardized occupancy rate of a certain office destination area is 0.92 (high demand intensity). After calculation, the staggered-time sharing potential index is 0.69, indicating that this cross-area combination has strong staggered-time sharing potential. Secondly, the underground priority access intensity and the spillover risk elasticity index between on-street and off-street parking were calculated. The remaining parking space ratio in the underground parking lot of this office-purpose area is 0.65, and the oversaturation ratio of the surface parking lot is 0.20. The calculated underground priority access intensity is 0.47. The standardized occupancy rate of on-street parking is 0.90, the standardized heat map value of historical illegal parking is 0.80, and the standardized value of road network service level is 0.40. The calculated on-street spillover risk elasticity index is 0.805, indicating a high spillover risk in this area. Then, the ant colony algorithm was used for multi-objective optimization on a hierarchical directed ant colony search graph. After 200 iterations, the Pareto scheduling scheme set contained multiple non-dominated schemes. In the morning peak management scenario, the target weights for surface parking oversaturation, on-street parking spillover risk, underground parking utilization level, and parking turnover efficiency were 0.15. The overall satisfaction of each candidate scheme was calculated, and the highest satisfaction was selected as the optimal compromise scheme. In the optimal compromise, of the 80 vehicles awaiting allocation from the residential area to the office area, 48 (60%) were directed to the underground parking lot, 24 (30%) to the surface parking lot, and 8 (10%) to on-street parking spaces. Based on this, the system generated vehicle allocation ratio instructions and pushed cross-regional sharing matching instructions to users. Simultaneously, due to the high on-street spillover risk elasticity index of 0.805 in this area, the system lowered the recommended level for on-street parking spaces and displayed a message on the guidance screen stating "On-street parking spaces are scarce; please prioritize underground parking." In the next scheduling period, the system continuously updates the parking coordination monitoring samples and various indices based on feedback information such as the actual number of vehicles entering and user acceptance rates, entering a new round of optimization loop. Final execution feedback showed that the oversaturation of surface parking lots decreased by 15%, and on-street illegal parking incidents decreased by 10%, verifying the effectiveness of the strategy.

[0134] Example 2 A parking resource integrated scheduling strategy optimization system based on multi-regional collaboration includes a module for executing processing instructions for each step in a parking resource integrated scheduling strategy optimization method based on multi-regional collaboration. Data collection sample construction module: Collects raw data on parking facilities, vehicle demand, illegal parking risks, and road network operation in multiple time domains, and generates standardized occupancy rate time series, parking space inventory, unallocated demand, violation heat map, and road network service level indicators through standardization and normalization operations, and outputs parking collaborative monitoring samples for each time period. Collaborative Feature and Ant Colony Graph Modeling Module: Based on monitoring samples, the module calculates the staggered sharing potential index, underground priority access intensity, and spillover risk elasticity index to generate collaborative governance factor vectors; it encodes areas and parking facilities to generate search graph nodes, divides vehicle guidance units, calculates path heuristic information values, and writes them into a hierarchical directed ant colony search graph. Multi-objective ant colony optimization solution module: Initialize ant colony pheromones, introduce capacity pressure sensing factors and capacity constraint self-repair mechanism to generate feasible scheduling schemes; rely on crowding sensing pheromone update mechanism to iteratively optimize, and obtain Pareto solution set through non-dominated sorting; calculate comprehensive satisfaction by multi-objective fuzzy membership degree combined with dynamic preference weight, and output the optimal compromise scheduling scheme. The closed-loop iteration module for guidance instructions converts the optimal scheduling scheme into parking guidance instructions for execution, collects landing feedback data, updates the input data and scheduling parameters for the next time period, and realizes dynamic iterative optimization of the scheduling strategy.

[0135] Although the specific embodiments of the invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the invention. Based on the technical solutions of the invention, various modifications or variations that can be made by those skilled in the art without creative effort are still within the scope of protection of the invention.

Claims

1. A method for optimizing the comprehensive scheduling strategy of parking resources based on multi-regional collaboration, characterized in that, Includes the following steps: S1. Collect parking resources, vehicle demand, on-street risks and road network operation data in multiple areas according to the divided scheduling time periods. Based on the collected data, determine the standardized occupancy rate time series curve, parking facility space capacity, number of parking facilities occupied spaces, number of parking facilities remaining spaces, demand for vehicles to be allocated, historical standardized heat value of illegal parking and standardized value of road network service level, and construct parking collaborative monitoring samples. S2. Based on the parking collaborative monitoring samples, calculate the staggered sharing potential index, underground priority access intensity and road spillover risk elasticity index, and then calculate the heuristic information value, and then write the heuristic information value into the hierarchical directed ant colony search graph. S3. Construct an ant scheduling scheme on a hierarchical directed ant colony search graph, introduce a capacity pressure sensing factor and a capacity constraint self-repair mechanism to reduce infeasible schemes, and then maintain the diversity of Pareto schemes through a crowding-sensing pheromone update mechanism to obtain the optimal compromise scheme. S4. Convert the optimal compromise solution into an executable parking guidance instruction, and continuously update the input data and scheduling parameters for the next scheduling period based on the execution results.

2. The method for optimizing parking resource scheduling strategy based on multi-regional collaboration as described in claim 1, characterized in that, S1 is as follows: The data collection area includes parking facilities in the functional departure area and the office functional purpose area. The types of parking facilities include underground parking lots, surface parking lots, and on-street parking spaces. Parking resource data includes vehicle entry time, vehicle exit time, parking space occupancy status, and parking space capacity. Vehicle entry time, vehicle exit time, and parking space occupancy status in underground and surface parking lots are collected through barrier gate systems, video detection systems, geomagnetic detection systems, or parking fee collection systems. The occupancy status of on-street parking spaces is collected through roadside video detection equipment, geomagnetic detection equipment, inspection terminals, or parking fee collection terminals. The standardized occupancy rate is calculated based on the vehicle entry time and vehicle exit time. The parking facility capacity, the number of parking facilities occupied, and the number of parking facilities remaining parking spaces are calculated based on the parking space occupancy status and parking space capacity. The vehicle demand data is the parking demand data from the residential area to the office area. The number of vehicles to be allocated is obtained by integrating parking reservation data, mobile terminal navigation destination data, historical commuting traffic prediction results and real-time parking guidance requests. The on-street risk data consists of historical illegal parking incident data, which is obtained by collecting illegal parking capture records, law enforcement records, and manual inspection records. The number of historical illegal parking incidents is obtained by statistical analysis by region and dispatch time period, and the standardized heat map value of historical illegal parking is obtained by normalizing the maximum value. Road network operation data includes surrounding road network operation status data, and the standardized value of road network service level is calculated by comprehensively considering the average vehicle speed, queue length, congestion index and road segment capacity. This results in a parking collaborative monitoring sample that includes standardized occupancy rate time-series curves, parking facility capacity, number of occupied parking spaces, number of remaining parking spaces, demand for vehicles to be allocated, standardized heatmap values ​​of historical illegal parking, and standardized values ​​of road network service levels. , Indicates the first Parking coordination monitoring samples for each scheduling period, The index represents the scheduling period, and its value range is... arrive , This indicates the total number of scheduling periods per day.

3. The method for optimizing parking resource scheduling strategy based on multi-regional collaboration as described in claim 2, characterized in that, S2 is as follows: S2.1 Set a time offset window near the current scheduling period, and perform weighted matching of the vacancy supply capacity of the residential function departure area and the parking demand intensity of the office function destination area to obtain the staggered sharing potential index; S2.2 Calculate the underground priority access intensity based on the parking capacity and remaining number of parking spaces in the underground parking lot and the parking capacity and occupied number of parking spaces in the surface parking lot in the office functional area. The spillover risk elasticity index is calculated based on the parking capacity, number of occupied parking spaces, historical standardized heat value of illegal parking, and standardized value of road network service level of the on-street parking area for office function purposes. The underground priority access intensity and the spillover risk elasticity index between roads and outside the road are combined into a collaborative governance factor vector; S2.

3. The residential function departure area, the office function destination area, and the parking facility type are jointly encoded as nodes in a hierarchical directed ant colony search graph. The vehicle guidance unit, which is divided according to the demand for vehicles to be allocated, is used as the minimum allocation object for ants to build a scheduling scheme, and a hierarchical directed ant colony search graph is established.

4. The method for optimizing parking resource scheduling strategy based on multi-regional collaboration as described in claim 3, characterized in that, S2.1 is as follows: A time offset window is selected with the current scheduling period t as the center, and the radius of the time offset window is set to be . The time offset index is h, and the value of h ranges from 1 to 2. arrive If t+h is less than 1 or greater than 1 Instead, the occupancy rate of the boundary scheduling period is used; The standardized occupancy rate of the residential function departure area in the t+h scheduling period is converted into a free parking rate. The standardized occupancy rate of the office function destination area in the t scheduling period is taken as the current parking demand intensity. By multiplying the standardized free parking rate with the current parking demand intensity, the shared matching intensity between the residential function departure area and the office function destination area at the time offset h is obtained. A decay weight is set for the basic sharing matching strength under different time offsets, and a weighted average is performed within the time offset window to obtain the time-sharing potential index.

5. The method for optimizing parking resource scheduling strategy based on multi-regional collaboration as described in claim 3, characterized in that, S2.2 is as follows: The remaining number of parking spaces and their capacities in multiple underground parking garages within the office function area are summed, and the ratio of these sums yields the proportion of remaining parking spaces in the underground garages. The occupied number of parking spaces and their capacities in the multiple underground parking garages within the office function area are summed, and the ratio of these sums yields the oversaturation rate of the surface parking garages. Based on the proportion of remaining parking spaces in the underground garages and the oversaturation rate of the surface parking garages, the underground priority access intensity is calculated using the following formula: ， in, This represents the underground priority access intensity of the j-th office function destination area during the t-th scheduling period, with a value range of [0,1], where j represents the index of the office function destination area; clip( (,0,1) denotes the truncation function; Indicates the weight of remaining underground capacity; Indicates the ground oversaturation weight; This represents the number of remaining parking spaces in the underground parking lot of the j-th office function destination area during the t-th scheduling period; This represents the parking capacity of the underground parking lot in the j-th office functional area; Represents the minimum stability constant; This represents the number of parking spaces occupied in the ground parking lot of the j-th office function destination area during the t-th scheduling period; This represents the parking capacity of the ground-level parking lot in the j-th office functional area; max( () represents the function that takes the maximum value; For the j-th office function area, read the number and capacity of parking spaces in on-street parking lots, the historical standardized heat value of illegal parking, and the standardized value of road network service level for the t-th scheduling period. Obtain the standardized occupancy rate of on-street parking based on the ratio of the number of parking spaces to the capacity of on-street parking lots. Add one to the negative of the standardized value of road network service level as the road operation pressure. Sum the standardized occupancy rate of on-street parking, the historical standardized heat value of illegal parking, and the road operation pressure with weights to obtain the spillover risk elasticity index between on-street and off-street areas. Finally, the underground priority access intensity and the spillover risk elasticity index between roads and outside the road are merged into a collaborative governance factor vector.

6. The method for optimizing parking resource scheduling strategy based on multi-regional collaboration according to claim 3, characterized in that, S2.3 is as follows: Establish a hierarchical directed ant colony search graph , This represents the directed graph structure used for ant colony search during the t-th scheduling period, including the set of nodes and the set of feasible directed edges; First, generate underground parking nodes, surface parking nodes, and on-street parking space nodes by combining each residential function departure area and office function destination area. Then, screen the nodes for feasibility based on the remaining capacity of parking facilities, gate access, reservation rules, traffic restriction strategies, and management strategies, and retain candidate routes. The demand for vehicles to be allocated is divided into vehicle guidance units according to the proportional granularity, which are used as the smallest vehicle groups to construct parking allocation schemes in ant colony search. Heuristic information values ​​are calculated based on the time-sharing potential index, underground priority access intensity, and spillover risk elasticity index. These heuristic information values ​​are then written into the feasible directed edges of the hierarchical directed ant colony search graph.

7. The method for optimizing parking resource scheduling strategy based on multi-regional collaboration according to claim 1, characterized in that, The ant scheduling scheme in S3, which introduces a capacity pressure-aware factor and a capacity constraint self-repair mechanism, is as follows: Initialize the ant population and pheromone concentration. Let the ant index be 'a', ranging from 1 to A, where A represents the total number of ants. Let the iteration index be 'r', ranging from 1 to... , Indicates the maximum number of iterations; In the t-th scheduling period and the r-th iteration, the pheromone concentration of the route from the i-th residential function departure area to the j-th office function destination area, selecting the k-th type of parking facility, is represented as: The pheromone concentration of all feasible paths is initialized to the same constant, and no pheromone concentration is set for impassable nodes and impassable directed edges. The remaining number of parking spaces for various types of parking facilities in each office functional area is compared with the demand for vehicles to be allocated from each residential departure area to each office functional area to obtain the capacity pressure perception factor. The state transition probability is calculated based on pheromone concentration, heuristic information value, and capacity pressure sensing factor. Each ant selects the parking facility type for the vehicle guidance unit in turn and generates an initial scheduling plan. For each combination of residential function departure area and office function destination area, the ants allocate vehicle guidance units one by one according to the state transition probability using a roulette wheel method. The initial scheduling plan is self-repaired based on capacity constraints. If the initial scheduling plan causes the number of vehicles allocated to a certain type of parking facility in the office function area to exceed the number of remaining parking spaces, the excess vehicles are recorded as vehicles to be repaired, and alternative parking facilities are selected from other feasible parking facility types in the same office function area. Then, a multi-objective evaluation is performed on the repaired ant scheduling plan. The multi-objective evaluation includes parking turnover efficiency, surface parking oversaturation, underground parking utilization, and on-street parking spillover risk. Finally, based on the multi-objective evaluation results, a fast non-dominated sorting was performed on all ant scheduling schemes to obtain multiple Pareto fronts.

8. The method for optimizing parking resource scheduling strategy based on multi-regional collaboration according to claim 7, characterized in that, S3 The specific mechanism for updating pheromones in medium density perception is as follows: For all ant scheduling schemes generated in the r-th iteration, perform non-dominated sorting and calculate the congestion distance within each Pareto front. For each Pareto front, sort according to the parking turnover efficiency target, the surface parking oversaturation target, the underground parking utilization level target, and the on-street parking spillover risk target. After normalizing each target, calculate the target difference between adjacent ant scheduling schemes and sum the multiple target differences to form the congestion distance. Construct an elite ant set based on the non-dominated level and the congestion distance, prioritizing the selection of ants with large congestion distances in the first Pareto front, and randomly selecting a small number of ants from the second or third Pareto front. The pheromone evaporation coefficient is determined based on the Pareto front distribution entropy. The ant scheduling schemes in the first Pareto front are then divided according to the target spatial location. There are several intervals, where b represents the index of the target spatial distribution interval. The first Pareto front will fall into the [missing information - likely a specific range or region]. The proportion of schemes in each interval is denoted as Iterate through all the partitioned intervals, based on Calculate the Pareto front entropy, while the ideal uniform distribution entropy is expressed as... The pheromone evaporation coefficient is calculated based on the lower limit of the pheromone evaporation coefficient, the difference between the upper and lower limits of the pheromone evaporation coefficient, the Pareto front distribution entropy, and the ideal uniform distribution entropy; then the pheromone concentration is updated for each feasible directed edge in the hierarchical directed ant colony search graph. Boundary limits are imposed on the updated pheromone concentration, setting a lower and upper limit for the pheromone concentration.

9. The method for optimizing parking resource scheduling strategy based on multi-regional collaboration as described in claim 8, characterized in that, The optimal compromise solution in S3 is selected as follows: When the ant colony reaches the maximum number of iterations, or when the Pareto scheduling scheme set no longer shows significant improvement after several consecutive rounds, stop ant colony optimization and collect the Pareto scheduling scheme set; convert the four-item target values ​​of each candidate Pareto scheduling scheme into fuzzy membership degrees, set the dynamic preference weights corresponding to the current governance scenario, calculate the overall satisfaction based on the fuzzy membership degrees and dynamic preference weights, and select the optimal compromise scheme. The fuzzy membership calculation process is as follows: The index of the four-objective multi-objective evaluation, including the parking turnover efficiency objective, the surface parking oversaturation objective, the underground parking utilization level objective, and the on-street parking spillover risk objective, is represented by m, with a value range of 1 to 4; Let p be the index of a candidate Pareto scheduler, and let denot p be the fuzzy membership degree of the p-th candidate Pareto scheduler on the m-th objective. The target values ​​include benefit-based targets and cost-based targets, and the calculation formula is as follows: For benefit-oriented goals: ; For cost-related objectives: ; in, This represents the maximum objective value of the m-th item in the Pareto scheduling scheme set; Let m represent the minimum objective value of the m-th item in the Pareto scheduling scheme set; This represents the target value of the p-th candidate Pareto scheduling scheme on the m-th target; Denotes the minimum stability constant; if and If they are the same, then the fuzzy membership degree of the target is uniformly set to 1.

10. A parking resource comprehensive scheduling strategy optimization system based on multi-regional collaboration, executing the parking resource comprehensive scheduling strategy optimization method based on multi-regional collaboration as described in any one of claims 1-9, characterized in that, It includes a data collection sample construction module, a collaborative feature and ant colony graph modeling module, a multi-objective ant colony optimization solution module, and an induced instruction closed-loop iteration module; Data collection sample construction module: Collects raw data on parking facilities, vehicle demand, illegal parking risks, and road network operation in multiple time domains, and generates standardized occupancy rate time series, parking space inventory, unallocated demand, violation heat map, and road network service level indicators through standardization and normalization operations, and outputs parking collaborative monitoring samples for each time period. Collaborative Feature and Ant Colony Graph Modeling Module: Based on monitoring samples, the module calculates the staggered sharing potential index, underground priority access intensity, and spillover risk elasticity index to generate collaborative governance factor vectors; it encodes areas and parking facilities to generate search graph nodes, divides vehicle guidance units, calculates path heuristic information values, and writes them into a hierarchical directed ant colony search graph. Multi-objective ant colony optimization solution module: Initialize ant colony pheromones, introduce capacity pressure sensing factors and capacity constraint self-repair mechanism to generate feasible scheduling schemes; rely on crowding sensing pheromone update mechanism to iteratively optimize, and obtain Pareto solution set through non-dominated sorting; calculate comprehensive satisfaction by multi-objective fuzzy membership degree combined with dynamic preference weight, and output the optimal compromise scheduling scheme. The closed-loop iteration module for guidance instructions converts the optimal scheduling scheme into parking guidance instructions for execution, collects landing feedback data, updates the input data and scheduling parameters for the next time period, and realizes dynamic iterative optimization of the scheduling strategy.

Images