Parking lot rolling optimization allocation method

Abstract

This invention provides a rolling optimization method for parking lot allocation: It employs a dual-objective optimization model to optimize the parking lot allocation scheme for each vehicle, including objective 1: minimizing the system's generalized cost, and objective 2: maximizing the user's overall utility. Addressing key issues in urban dynamic parking allocation such as insufficient environmental adaptability, lack of user psychological perception mechanisms, and imbalance in multi-objective optimization, this invention constructs a rolling optimization model for parking lot allocation that integrates dynamic psychological perception and dual-objective game theory. Simulation verification shows that this invention can effectively reduce the system's generalized cost, improve user overall utility and parking guidance acceptance rate, and significantly suppress secondary parking. It has advantages over traditional static allocation and free selection methods, and can provide a feasible solution for urban intelligent parking guidance systems.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation technology, and specifically relates to a method for rolling optimization allocation of parking lots. Background Technology

[0002] With the increasing number of motor vehicles, the parking supply and demand imbalance in core urban functional areas (commercial centers, transportation hubs, and densely populated office areas) is becoming increasingly severe. Scarcity of parking resources, low parking space search efficiency, and insufficient effectiveness of guidance systems have become key constraints exacerbating regional traffic congestion and reducing the quality of travel services. Parking space / area allocation, as the core final link in parking demand management, can effectively reduce vehicle waiting time and ineffective cruising by rationally allocating and accurately guiding vehicles to designated parking spaces or areas, thereby improving road network efficiency and transfer convenience. It is an important technical approach to alleviating urban parking difficulties.

[0003] Existing research on parking allocation has yielded multi-dimensional results. Shao et al. established a parking space allocation model with time window constraints, aiming to maximize the benefits of the parking system. Zhang Wenhui et al. constructed a dual-objective parking space allocation model guided by maximizing the utilization rate of shared parking spaces and minimizing walking distance, and used particle swarm optimization (PSO) for multi-objective solution. Wu Ruowei, Zhou Biyang, and others conducted allocation research based on Dijkstra's shortest path algorithm, comprehensively considering attributes such as internal parking lot path distance, road traffic quality, and parking space availability. Lin Xiaowei et al. introduced cooperative game theory into parking lot parking space allocation, effectively reducing system parking costs.

[0004] However, these parking guidance methods rely heavily on static information dissemination and proximity allocation strategies, failing to fully consider real-time road network congestion, differences in road segment preferences, and changes in drivers' subjective psychology. This can easily lead to problems such as parking space overload, user rejection, and secondary patrols, making it difficult to achieve optimal synergy between system cost and user experience. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a rolling optimization allocation method for parking lots, which solves the parking guidance optimization problem under multiple superimposed factors.

[0006] The present invention achieves the above-mentioned technical objectives through the following technical means.

[0007] A rolling optimization method for parking lot allocation: The following bi-objective optimization model is used to optimize the parking lot allocation scheme for each vehicle:

[0008] Objective 1: Minimize the generalized cost of the system.

[0009]

[0010]

[0011] Objective 2: Maximize overall user utility.

[0012]

[0013]

[0014] In the formula, For the system's generalized cost, For the overall benefit of users; For vehicles With parking lot The allocation variable between them, when set to 1, represents a vehicle. Assigned to parking lot When the value is 0, it indicates that the vehicle... Not allocated to parking lots ; This is the time value coefficient; This is the penalty coefficient; for Time vehicle For allocation to parking lot Psychological acceptance; To induce a higher number of failed vehicles, for vehicles If there is no such condition... According to the allocation scheme, the vehicle is included in Statistics; For objective utility weighting coefficients, This is the moderating factor for psychological utility. for Time vehicle Drive to parking lot The duration of the trip, for Time vehicle Users from parking lot The walking time to reach one's destination.

[0015] Furthermore, the bi-objective optimization model is subject to the following constraints:

[0016] Capacity constraint: The number of vehicles allocated to each parking lot shall not exceed the number of parking spaces remaining in that parking lot;

[0017] Uniqueness constraint: For any vehicle, at most one parking lot can be assigned to it;

[0018] Psychological feasibility constraint: only when Only then can it be made The value is 1.

[0019] Furthermore, the psychological acceptability function is:

[0020]

[0021] In the formula, The driver's initial psychological intention. Anxiety enhancement coefficient, For vehicles The waiting time for allocation; for Time vehicle For allocation to parking lot The objective negative effect is obtained by weighted summation of multiple indicators.

[0022] Furthermore, the aforementioned objective negative utility The calculation is as follows:

[0023]

[0024]

[0025] In the formula, for Always targeting vehicles Assigned to parking lot The Normalized values ​​of the indicators For vehicles The The weight of each indicator; the indicators include: vehicles Drive to parking lot Trip duration ,vehicle Users from parking lot Walking time to their destination ,vehicle In the parking lot parking fee .

[0026] Furthermore, the weight of the indicator Make dynamic adjustments as follows:

[0027] collection Time Parking Remaining parking spaces The normalized value is calculated as follows. :

[0028]

[0029] Calculate the first Under this indicator, the characteristic proportions of each parking lot :

[0030]

[0031] Calculate the first Information entropy of the indicator:

[0032]

[0033] Calculate the corresponding standardized entropy value:

[0034]

[0035] Calculate the coefficient of difference:

[0036]

[0037] Normalizing the difference coefficients, we get Always targeting vehicles Dynamic weights of each indicator:

[0038]

[0039] In the formula, , This represents the total number of parking lots.

[0040] Furthermore, a genetic algorithm is used to optimize and solve the bi-objective optimization model.

[0041] Furthermore, in the genetic algorithm, after each crossover and mutation operation, the following constraint repair is performed:

[0042] Capacity constraint repair: If parking lot If the number of vehicles allocated exceeds the number of available parking spaces, the excess vehicles will be randomly removed and reassigned to other parking lots that are not full.

[0043] Psychological feasibility constraint repair: In the inspection of the allocation plan, for each group of vehicles The parking lot allocated to it Do they satisfy the condition? ;against vehicles Calculate its relationship with all other parking lots that are not overcrowded. If there is Then select from them The largest parking lot is allocated to this vehicle If not Then the vehicle Parking lots will not be allocated, and will be included in the calculation. .

[0044] Furthermore, in the genetic algorithm:

[0045] The fitness function is:

[0046]

[0047] In the formula For fitness, parameters =10 -6 ;

[0048] The operator is selected as:

[0049] A hybrid selection method combining elite retention and roulette wheel selection is used. The top 10% of fittest individuals are retained, do not participate in crossover mutation, and directly enter the next generation. The remaining individuals are selected based on their fitness probabilities, and these probabilities are used to enter the next generation. The selection probabilities are:

[0050]

[0051] In the formula, Individual The probability of choosing, Individual The fitness of;

[0052] The crossover operator is:

[0053] A single-point crossover method is used, where crossover points are randomly selected and the genes at the crossover points of two parent chromosomes are crossed. After the crossover, two new chromosomes are formed and added to the next generation population as offspring.

[0054] The mutation operator is:

[0055] Translocation mutations were performed on a single chromosome individual, with two gene loci randomly selected for mutation on each chromosome individual.

[0056] Furthermore, the duration of the trip The calculation is as follows:

[0057]

[0058]

[0059] In the formula, For vehicles Drive to parking lot The planned path, For road section In the present The travel time at any given moment For road section Free-flow travel time, For road section In the present Traffic flow at any given time For road section The design capacity of the passageway and This is a key parameter for the BPR function.

[0060] Furthermore, the walking time The calculation is as follows:

[0061]

[0062] In the formula, For vehicles Users from parking lot The distance to its destination, This refers to walking speed.

[0063] The beneficial effects of this invention are as follows:

[0064] (1) This invention provides a rolling optimization allocation method for parking lots. Addressing key issues in urban dynamic parking allocation, such as insufficient environmental adaptability, lack of user psychological perception mechanisms, and imbalance in multi-objective optimization, a rolling optimization allocation model for parking lots integrating dynamic psychological perception and bi-objective game theory is constructed. Simulation verification shows that this invention can effectively reduce the generalized cost of the system, improve the overall utility of users and the acceptance rate of parking guidance, and significantly suppress secondary patrols. It has advantages over traditional static allocation and free selection methods and can provide a feasible solution for urban intelligent parking guidance systems.

[0065] (2) The present invention adopts the improved gray entropy method to realize the dynamic quantification and weight calculation of indicators, constructs a user psychological acceptance function that considers waiting anxiety, establishes a dual-objective optimization model with the minimum system cost and the maximum user utility, and completes the rolling solution through a genetic algorithm with constraint repair. Attached Figure Description

[0066] Figure 1 This is a schematic diagram of the simulation scenario built in the test example;

[0067] Figure 2 This is a heatmap of vehicle-parking lot psychological acceptance calculated in the test case;

[0068] Figure 3 For comparison of the generalized cost of the system in the test case;

[0069] Figure 4 For the comparison of overall user utility in the test cases;

[0070] Figure 5 For comparison of induced acceptance rates in test cases;

[0071] Figure 6 This is a comparison of the secondary cruise rates in the test cases. Detailed Implementation

[0072] The embodiments of the present invention are described in detail below. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0073] I. Technical Solution

[0074] 1. Problem Description

[0075] Suppose at a certain moment The area has A vehicle waiting to park wants to guide the system to issue a parking request. The system can allocate parking lots in the following locations: The system needs to allocate a unique parking space to each vehicle while meeting parking lot capacity constraints, achieving a dual-objective collaborative optimization that minimizes the system's generalized cost and maximizes overall user satisfaction.

[0076] 1. Constraints

[0077] 1) Capacity constraint: The number of vehicles allocated to a single parking lot shall not exceed the number of available parking spaces in real time.

[0078] 2) Uniqueness constraint: Each vehicle is assigned to only one parking lot, without duplication or omission.

[0079] 3) Psychological feasibility constraint: Users' psychological acceptance of the allocated parking spaces shall not be lower than the set threshold.

[0080] 3. Parking lot selection criteria

[0081] This invention selects four key indicators—trip time, walking time, charging standard, and remaining parking spaces—to construct a parking lot selection evaluation system.

[0082] 3.1, Trip Duration

[0083] To accurately reflect parking travel time under real-time congestion conditions in the regional road network, the classic BPR (Bureau of Public Roads) road resistance function is introduced as the basic model to achieve dynamic quantification of travel time for each parking lot. For any road segment on the vehicle's travel path, the actual travel time consists of two parts: free-flow time and congestion correction term, calculated as follows:

[0084]

[0085] In the formula, For road section In the present The travel time at any given moment For road section Free-flow travel time (i.e., travel time without congestion, in seconds). For road section In the present Traffic flow at any given time (unit: pcu / h). For road section Design capacity (unit: pcu / h) and These are the key parameters for the BPR function (set according to the actual urban road network conditions).

[0086] In the present At that moment, the vehicle Drive to parking lot The trip duration is calculated as follows:

[0087]

[0088] In the formula, for Time vehicle Drive to parking lot The duration of the trip, For vehicles Drive to parking lot The planned path.

[0089] 3.2 Walking time

[0090] vehicle Users from parking lot The walking time to its destination is calculated as follows:

[0091]

[0092] In the formula, for Time vehicle Users from parking lot The walking time to their destination. For vehicles Users from parking lot The distance to its destination, This refers to walking speed.

[0093] 3.3, Fee Standard

[0094] vehicle In the parking lot The fee is calculated as follows:

[0095]

[0096] In the formula, and parking lots The basic parking fee and time-limited surcharge; actual parking duration = user's stay at the destination + walking time to and from the parking lot. Since the stay at the destination is fixed, while the walking time changes with the allocation scheme, this invention only considers the impact of walking time on parking fees when optimizing the allocation scheme.

[0097] 3.4 Remaining parking spaces

[0098] Time Parking Real-time remaining parking spaces The number of remaining parking spaces is reported in real time by each parking lot management. Note: The actual number of remaining parking spaces depends on the number of vehicles. This application is not relevant; it is merely for formatting purposes. Add subscript to .

[0099] 4. Indicator Weights

[0100] To objectively reflect the importance of each factor at different times (e.g., the morning rush hour focuses on time, and the evening rush hour focuses on cost), this invention uses an improved gray entropy theory to determine the weight of each indicator.

[0101] 4.1, for any vehicle , build Original evaluation matrix at time step :

[0102]

[0103] 4.2 To eliminate differences in dimensions and orders of magnitude among different indicators, the range standardization method is used to standardize the matrix. The standard evaluation matrix is ​​obtained through processing. :

[0104]

[0105] in:

[0106] 1) Trip duration Walking time ,Parking Fee As a cost-related indicator, following the principle of "the smaller the better," its standardized formula is:

[0107]

[0108] 2) Remaining parking spaces As a benefit-oriented indicator, following the principle of "the larger the better," its standardized formula is:

[0109]

[0110] 4.3 The weighting of the above-mentioned indicator characteristics is calculated based on the improved information entropy theory:

[0111] 1) Calculate the first item( Under the indicator, the characteristic proportions of each parking lot :

[0112]

[0113] like Then let .

[0114] 2) Calculate the first item( Information entropy of the indicator:

[0115]

[0116] The agreement is as follows hour, ;

[0117] The corresponding standardized entropy value is:

[0118]

[0119] 3) Calculate the coefficient of variation:

[0120]

[0121] The coefficient of variation reflects the degree of dispersion of the indicator; the larger the value, the greater the information content and the higher the weight.

[0122] 4) Normalize the difference coefficients to obtain Always targeting vehicles Dynamic weights of each indicator:

[0123]

[0124] 5. Dynamic psychological acceptance

[0125] To quantify drivers' subjective willingness to choose, a vehicle-based system was constructed. Parking lot Psychological acceptability function :

[0126]

[0127]

[0128] In the formula, The driver's initial psychological intention. Anxiety enhancement coefficient, For vehicles The waiting time for allocation (i.e., the time that a vehicle has been waiting up to the present moment because there is no allocation scheme that can meet its parking needs after it has made a parking request). It is a natural constant; The calculation measures anxiety tolerance, which varies with waiting time. With the increase in the number of drivers, drivers will be more willing to accept longer travel times, longer walking distances, or higher parking fees in order to stop as soon as possible; for( Time vehicle For allocation to parking lot The objective negative effect is obtained by weighted summation of the three previously selected indicators (trip duration, walking time, and parking fees).

[0129] 6. Bi-objective game optimization model

[0130] Using the rolling time window as the basic optimization unit, a dual-objective optimization model is established to achieve a balance between system interests and user interests. Within each rolling time segment, the following dual-objective function is constructed:

[0131] 1) Objective 1: Minimize the generalized cost of the system

[0132] System generalized cost Defined as:

[0133]

[0134] In the formula, For vehicles With parking lot The allocation variable between them takes the value 0 or 1, with 1 indicating a vehicle. Assigned to parking lot When the value is 0, it indicates that the vehicle... Not allocated to parking lots ; This is the time value coefficient; This is a penalty coefficient used to quantify the additional system cost incurred by each "failed inducement vehicle"; To induce a lower number of failed vehicles, a psychological acceptance function is used. Perform calculations for a given allocation scheme, when At that time, the vehicle Accept the parking lot assigned to it ,on the contrary At that time, it was stated that the allocation plan was not accepted, regarding the vehicles. If there is no such condition... According to the allocation scheme, the vehicle is included in Statistics.

[0135] Objective 1 is represented as:

[0136]

[0137] Objective 1 encompasses the network driving costs and time value costs, and imposes penalties on vehicles that fail to be induced, in order to reduce network congestion and ineffective cruising.

[0138] 2) Objective 2: Maximize overall user utility

[0139] Overall user utility Defined as:

[0140]

[0141] In the formula, For objective utility weighting coefficients, This is the psychological utility moderating coefficient.

[0142] Objective 2 is represented as:

[0143]

[0144] Objective 2: Integration of objective indicators for evaluation scores With subjective psychological feelings This will comprehensively enhance the user parking experience.

[0145] 3) Constraints

[0146] Capacity constraints:

[0147]

[0148] This indicates that the number of vehicles allocated to each parking lot does not exceed the number of parking spaces remaining in that lot.

[0149] Uniqueness constraint:

[0150]

[0151] This means for any vehicle Only one parking lot will be allocated to it at most;

[0152] Psychological feasibility constraints:

[0153]

[0154] Indicates only when Only then can it be made A value of 1 means that vehicles are allowed. Assigned to parking lot .

[0155] 7. Model Solving

[0156] For the bi-objective game optimization model constructed above, this embodiment specifically uses a genetic algorithm to optimize and solve the problem in order to obtain the optimal allocation scheme.

[0157] 7.1 Chromosome Coding

[0158] The chromosomes of the population are constructed using real-number encoding, and each chromosome is represented by a P×2 matrix (where P is the chromosome length, determined by the number of vehicles to be allocated). An example is shown in Table 1 below:

[0159] Table 1: Examples of Chromosome Coding

[0160]

[0161] In the example above, vehicles 1 and 6 are assigned to parking lot 1, vehicles 2 and 4 are assigned to parking lot 3, vehicle 3 is assigned to parking lot 2, and vehicle 5 is assigned to parking lot 4.

[0162] 7.2 Fitness Function

[0163] Based on the system's generalized cost and the user's comprehensive utility, the following fitness function is designed:

[0164]

[0165] In the formula For fitness, parameters =10 -6 The specifics can be adjusted according to the actual situation.

[0166] 7.3 Genetic manipulation design

[0167] (1) Selection Operator

[0168] To preserve optimal allocation schemes and avoid premature convergence, a hybrid selection method combining elite retention and roulette wheel selection is employed. Specifically, the top 10% of fittest individuals are retained, do not participate in crossover or mutation, and directly enter the next generation. The remaining individuals are selected based on their fitness probabilities, and these probabilities are used to enter the next generation.

[0169]

[0170] In the formula, Individual The probability of selection (i.e., the probability of being selected). Individual The degree of adaptability.

[0171] (2) Crossover operator

[0172] The selection operation yields a set of candidate individuals to be passed on to the next generation. Then, a crossover operation is used to exchange some genes from the selected parent chromosomes to form offspring chromosomes, thus achieving the inheritance of superior genes. The crossover operation determines the global search capability of the genetic algorithm. This embodiment uses a single-point crossover method, randomly selecting a crossover point. When the crossover probability is satisfied (calculating the crossover probability is a standard technique in genetic algorithms), the genes at the crossover point of the two parent chromosomes are crossed, forming two new chromosomes that are added to the next generation population as offspring.

[0173] (3) Mutation operator

[0174] The method of translocation mutation is adopted by individual chromosomes. Two gene loci are randomly selected for mutation in each chromosome individual. When the mutation probability is satisfied (the calculation of mutation probability is a conventional method in genetic algorithms), the two gene loci are swapped to achieve mutation.

[0175] 7.4, Layered Progressive Constraint Repair Mechanism

[0176] Because the crossover and mutation operations in genetic algorithms are random, they can easily generate infeasible solutions that violate uniqueness constraints, parking lot capacity constraints, and user psychological feasibility constraints during iterative optimization, directly leading to the inability to implement the allocation scheme. To ensure the legality, effectiveness, and engineering feasibility of the optimized solution, this invention adopts the following hierarchical progressive constraint repair mechanism: following the execution order of capacity constraints first and psychological feasibility constraints as a fallback, the individuals after crossover and mutation are standardized and repaired to ensure that all individuals in the population meet the model constraints, while preserving the optimization performance of the algorithm to the greatest extent.

[0177] (1) Capacity constraint repair

[0178] If parking lot The number of vehicles allocated exceeds the number of available parking spaces. Then, excess vehicles will be randomly removed and reassigned to other parking lots that are not full. Specifically:

[0179] Step 1: Count the number of vehicles currently allocated to each parking lot and compare it with the real-time remaining parking spaces to identify overloaded parking lots;

[0180] Step 2: For overcapacity parking lots, remove vehicles exceeding the capacity limit from the parking lot allocation set according to the principle of "randomly removing excess vehicles";

[0181] Step 3: The vehicles to be removed are reassigned (randomly) to parking lots that are not overloaded, until the number of vehicles assigned to any parking lot does not exceed its capacity limit.

[0182] (2) Psychological feasibility constraint repair

[0183] In the inspection and allocation plan, each group of vehicles The parking lot allocated to it Do they satisfy the condition? ;against vehicles Calculate its relationship with all other parking lots that are not overcrowded. If there is Then select from them The largest parking lot is allocated to this vehicle If not Then the vehicle Parking lots will not be allocated, and will be included in the calculation. .

[0184] Constraint repair is forcibly triggered after each crossover and mutation operation, employing a hierarchical repair sequence of "capacity → psychological feasibility." The results of previous constraint repairs serve as the input basis for subsequent repairs, avoiding redundant repairs and constraint conflicts. This mechanism can transform all infeasible solutions into legal solutions that satisfy all model constraints without changing the algorithm's optimization direction, ensuring that the allocation scheme output by the improved genetic algorithm is real-time, feasible, and practical, adapting to the engineering requirements of dynamic rolling parking allocation.

[0185] II. Testing and Verification

[0186] like Figure 1 The simulation scenario shown includes three parking lots (P1, P2, and P3), with five vehicles simultaneously requesting parking to verify the allocation performance of this invention under peak supply and demand conditions.

[0187] The relevant parameter settings for this test are shown in Tables 2 and 3 below:

[0188] Table 2: Model Parameters

[0189]

[0190] Table 3: Basic Information on Parking Lots

[0191]

[0192] Taking one of the five vehicles waiting to park as an example, the original evaluation matrix is ​​calculated for it. :

[0193]

[0194] Corresponding standard evaluation matrix :

[0195]

[0196] Calculate the characteristic proportions of each parking lot And summarized into a matrix form as follows:

[0197]

[0198] Furthermore, the standardized entropy values ​​and difference coefficients for each vehicle are obtained as follows:

[0199]

[0200] Ultimately, the dynamic weights of each indicator for this vehicle are obtained. :

[0201]

[0202] Based on the psychological acceptance function, calculate the psychological acceptance between vehicles and parking lots. ,like Figure 2 The image shows a heatmap of vehicle-parking lot psychological acceptance, which clearly illustrates the heterogeneous distribution of driver preferences.

[0203] Based on the constructed bi-objective game optimization model, a genetic algorithm was used to optimize the solution, and the allocation schemes are shown in Table 4:

[0204] Table 4: Allocation Scheme

[0205]

[0206] To verify the advantages of this invention, the following comparative optimization scheme was set up: ① Set up a static weight allocation scheme (weights of the four indicators). The values ​​are taken in sequence as follows: ① Free choice of scheme. The system's generalized cost, overall user utility, induced acceptance rate, and secondary visit rate are selected to evaluate the effectiveness of the scheme. A rolling time window of 30 seconds is chosen, with a total time of 10 minutes (corresponding to a total of 20 time segments). The comparison results are as follows: Figures 3-6 As shown.

[0207] Figure 3 The diagram shows a comparison of the generalized system costs. During the dynamic evolution over 20 time slices, the system cost of the "free choice" strategy remained high and fluctuated dramatically throughout, reflecting the severity of local congestion in the absence of scheduling. The static model's cost was moderate but struggled to adapt to environmental changes. In contrast, the model proposed in this invention maintained its lowest value throughout the entire process with stable fluctuations.

[0208] Figure 4The diagram shows a comparison of overall user utility. The static model achieves high single-user utility in some time slices, but its overall performance fluctuates wildly and lacks robustness. The user utility in the free choice mode remains low throughout, resulting in an extremely unstable experience. The model proposed in this invention maintains a consistently high utility level, with a small difference from the static model and significantly higher than the free choice mode. Furthermore, its curve is smooth and exhibits strong resistance to interference.

[0209] Figure 5 The chart shows a comparison of user acceptance rates to inducements. The acceptance rate of the model in this invention remained stable above 80% throughout, exceeding 85% in most time slices, significantly higher than the static model (72%~82%) and the free choice mode (average less than 60%). The static model had a moderate acceptance rate but was easily affected by the scenario, while the free choice mode showed a low level and significant fluctuations.

[0210] Figure 6 The diagram shows a comparison of secondary cruising rates. In the free-choice mode, the cruising rate remained above 30% throughout, reflecting the disorderly nature of traffic flow and low location-finding efficiency without scheduling guidance. The static model had a moderate cruising rate, but still exhibited a high rate of ineffective empty runs. The model proposed in this invention significantly reduced the secondary cruising rate, keeping it below 25% overall.

[0211] Simulation results show that, compared with static weight allocation schemes and free selection strategies, the dynamic optimization model proposed in this invention exhibits significant advantages in reducing the generalized system cost, improving the overall user utility, increasing the induced acceptance rate, and suppressing the secondary patrol rate. This verifies the effectiveness and practicality of the model in dynamic parking guidance scenarios, effectively solving the problem of balancing user experience and system efficiency in traditional parking lot allocation methods. It can provide theoretical support and practical reference for intelligent parking management in dynamic and complex scenarios.

[0212] This invention is not limited to the above-described embodiments. Any obvious improvements, substitutions, or modifications that can be made by those skilled in the art without departing from the essence of this invention are within the scope of protection of this invention.

Claims

1. A method for optimizing parking lot allocation, characterized in that: The following bi-objective optimization model is used to optimize the parking lot allocation scheme for each vehicle: Objective 1: Minimize the generalized cost of the system. Objective 2: Maximize overall user utility. In the formula, For the system's generalized cost, For the overall benefit of users; For vehicles With parking lot The allocation variable between them, when set to 1, represents a vehicle. Assigned to parking lot When the value is 0, it indicates that the vehicle... Not allocated to parking lots ; This is the time value coefficient; This is the penalty coefficient; for Time vehicle For allocation to parking lot Psychological acceptance; To induce a higher number of failed vehicles, for vehicles If there is no such condition... According to the allocation scheme, the vehicle is included in Statistics; For objective utility weighting coefficients, This is the moderating factor for psychological utility. for Time vehicle Drive to parking lot The duration of the trip, for Time vehicle Users from parking lot The walking time to reach one's destination.

2. The parking lot rolling optimization allocation method according to claim 1, characterized in that: The dual-objective optimization model is subject to the following constraints: Capacity constraint: The number of vehicles allocated to each parking lot shall not exceed the number of parking spaces remaining in that parking lot; Uniqueness constraint: For any vehicle, at most one parking lot can be assigned to it; Psychological feasibility constraint: only when Only then can it be made The value is 1.

3. The parking lot rolling optimization allocation method according to claim 1, characterized in that: The psychological acceptability function is: In the formula, The driver's initial psychological intention. Anxiety enhancement coefficient, For vehicles The waiting time for allocation; for Time vehicle For allocation to parking lot The objective negative effect is obtained by weighted summation of multiple indicators.

4. The parking lot rolling optimization allocation method according to claim 3, characterized in that: The objective negative effect The calculation is as follows: In the formula, for Always targeting vehicles Assigned to parking lot The Normalized values ​​of the indicators For vehicles The The weight of each indicator; the indicators include: vehicles Drive to parking lot Trip duration ,vehicle Users from parking lot Walking time to their destination ,vehicle In the parking lot parking fee .

5. The parking lot rolling optimization allocation method according to claim 4, characterized in that: The weight of the indicator Make dynamic adjustments as follows: collection Time Parking Remaining parking spaces The normalized value is calculated as follows. : Calculate the first Under this indicator, the characteristic proportions of each parking lot : Calculate the first Information entropy of the indicator: Calculate the corresponding standardized entropy value: Calculate the coefficient of difference: Normalizing the difference coefficients, we get Always targeting vehicles Dynamic weights of each indicator: In the formula, , This represents the total number of parking lots.

6. The parking lot rolling optimization allocation method according to claim 1, characterized in that: The bi-objective optimization model is optimized and solved using a genetic algorithm.

7. The parking lot rolling optimization allocation method according to claim 6, characterized in that: In the genetic algorithm, after each crossover and mutation operation, the following constraint repair is performed: Capacity constraint repair: If parking lot If the number of vehicles allocated exceeds the number of available parking spaces, the excess vehicles will be randomly removed and reassigned to other parking lots that are not full. Psychological feasibility constraint repair: In the allocation plan, for each group of vehicles The parking lot allocated to it Do they satisfy the condition? ;against vehicles Calculate its relationship with all other parking lots that are not overcrowded. If there is Then select from them The largest parking lot is allocated to this vehicle If not Then the vehicle Parking lots will not be allocated, and will be included in the calculation. .

8. The parking lot rolling optimization allocation method according to claim 6, characterized in that: In the genetic algorithm: The fitness function is: In the formula For fitness, parameters =10 -6 ; The operator is selected as: A hybrid selection method combining elite retention and roulette wheel selection is used. The top 10% of fittest individuals are retained, do not participate in crossover mutation, and directly enter the next generation. The remaining individuals are selected based on their fitness probabilities, and these probabilities are used to enter the next generation. The selection probabilities are: In the formula, Individual The probability of choosing, Individual The fitness of; The crossover operator is: A single-point crossover method is used, where crossover points are randomly selected and the crossover genes of two parent chromosomes are crossed. The resulting crossover forms two new chromosomes, which are then added to the next generation of the population as offspring. The mutation operator is: Translocation mutations were performed on a single chromosome individual, with two gene loci randomly selected for mutation on each chromosome individual.

9. The parking lot rolling optimization allocation method according to claim 1, characterized in that: The duration of the trip The calculation is as follows: In the formula, For vehicles Drive to parking lot The planned path, For road section In the present The travel time at any given moment For road section Free-flow travel time, For road section In the present Traffic flow at any given time For road section The design capacity of the passageway and This is a key parameter for the BPR function.

10. The parking lot rolling optimization allocation method according to claim 1, characterized in that: walking time The calculation is as follows: In the formula, For vehicles Users from parking lot The distance to its destination, This refers to walking speed.

Images