Parking space allocation result generation method, device, equipment and medium

Abstract

The application discloses a parking lot parking space allocation result generation method and device, equipment and medium, and relates to the technical field of intelligent parking, comprising: collecting parking request information and parking space state data, calculating comprehensive performance index data by using a preset parking space allocation model based on deep learning, and screening a plurality of candidate schemes according to the comprehensive performance index data. Then, a constraint optimization strategy is used for further optimization, so that the requirements of a specific parking space reservation ratio and a waiting time of non-reserved vehicles are met, and finally, a target parking space allocation scheme is obtained, the resource utilization rate of the parking lot is improved, and the waiting time of users is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent parking, and in particular to a parking space allocation result generation method, device, equipment and medium. BACKGROUND

[0002] With the acceleration of urbanization and the continuous growth of motor vehicle ownership, urban parking problems have increasingly become an important factor affecting the efficiency of urban operation and the quality of life of residents. The spatial and temporal distribution of urban parking demand is uneven, leading to great challenges in the allocation of parking resources. On the one hand, the parking demand in peak hours and certain areas far exceeds the supply, causing traffic congestion and parking difficulties; on the other hand, parking spaces in some areas are largely idle during off-peak hours, resulting in low resource utilization. In addition, with the popularity of new energy vehicles and the care for special groups such as the disabled, the demand for specific types of parking spaces (such as new energy vehicle parking spaces and disabled parking spaces) is also increasing. How to efficiently and fairly allocate limited parking resources to meet the needs of different user groups, while improving the operational efficiency of parking lots and user satisfaction, has become a problem to be solved in the field of intelligent transportation.

[0003] Currently, existing parking space allocation algorithms mainly rely on the following two categories: static proportion division method, which divides the available parking resources of reserved and non-reserved users by pre-setting a fixed proportion, such as reserving 30%-50% of the total parking spaces in a parking lot as a reserved area based on historical parking data, and the remaining spaces are available for temporary users. This method combines regional physical isolation or electronic fence technology to achieve resource segmentation, and dynamically adjusts the proportion threshold through a rule engine. There is also a demand response type dynamic adjustment method, which is suitable for target tracking in complex scenarios. However, the calculation complexity of Transformer is high, and it usually needs to manually initialize the target, limiting its application in real-time scenarios. Real-time demand prediction technology is adopted to dynamically adjust the allocation proportion of parking spaces for two types of users through a weighted scoring model.

[0004] However, the main problem of the static proportion division method is that it is essentially an empirical decision based on historical averages, and cannot respond to dynamic demand fluctuations. This leads to a failure to adjust the resource allocation in real time when there is a significant difference in the arrival rate of reserved and non-reserved users during peak hours, resulting in resource mismatch, reducing overall resource utilization, and exacerbating road congestion. Although the demand response type dynamic adjustment method introduces a dynamic adjustment mechanism, the core algorithm relies on manually defined weight parameters (such as temporary user priority coefficients and reserved parking violation penalty factors), which leads to systematic bias between the decision logic and the real scenario. When faced with sudden demand surges or multi-event coupling situations, suboptimal decisions may be made, resulting in the incorrect release of reserved parking spaces and causing resource competition between the two user groups. Therefore, there is an urgent need for an efficient method to allocate limited parking resources to improve the operational efficiency of parking lots and user satisfaction. SUMMARY

[0005] The main purpose of the present application is to provide a parking lot parking space allocation result generation method, device, equipment and medium, aiming at solving the technical problem of how to efficiently and reasonably allocate parking space resources.

[0006] To achieve the above purpose, the present application provides a parking lot parking space allocation result generation method, comprising:

[0007] Obtaining parking request information and parking space state data;

[0008] Inputting the parking request information and the parking space state data into a preset parking space allocation model to obtain comprehensive performance index data and a corresponding candidate allocation scheme, the preset parking space allocation model being a neural network model with a double network structure, and the comprehensive performance index data including parking fee income, parking space utilization rate and user waiting time;

[0009] The candidate allocation scheme is screened by comparing the comprehensive performance index data to obtain a parking space allocation scheme;

[0010] The parking space allocation scheme is optimized by using a constraint optimization strategy to obtain a target parking space allocation scheme.

[0011] In an embodiment, the preset parking space allocation model includes a main network and a target network, the main network being used for real-time updating and learning of an optimal parking space allocation strategy, and the target network being used for stabilizing the learning process, and the weights of the main network being copied to the target network every preset time step or preset learning period.

[0012] In an embodiment, before the parking request information and the parking space state data are inputted into the preset parking space allocation model to obtain the comprehensive performance index data and the corresponding candidate allocation scheme, the method further comprises:

[0013] Obtaining parking request information sample data and parking space state sample data, and establishing an initial parking space allocation model;

[0014] Training the initial parking space allocation model according to the parking request information sample data and the parking space state sample data to obtain the preset parking space allocation model.

[0015] In an embodiment, the step of training the initial parking space allocation model according to the parking request information sample data and the parking space state sample data to obtain the preset parking space allocation model comprises:

[0016] Normalizing the parking request information sample data and the parking space state sample data to construct a training set;

[0017] define state information, action information and a reward function, wherein the reward function is represented as:

[0018]

[0019] wherein T represents a period of time, π1 represents an expected revenue for a reserved parking user, a t and b t represent parameters of the reward function, p t represents a parking fee per unit time, π2 is an expected revenue for a non-reserved parking user, θ is a proportion of reservable parking spaces, k is a total number of parking spaces, is a reservation rate paid by a reserved parking user, is a demand for a reserved parking user, is a random variable of the number of reserved parking users, is a probability density function of a reserved user, is a demand for a non-reserved parking user, is a random variable of the number of non-reserved parking users, is a probability density function of a non-reserved user;

[0020] initialize training parameters, the state information and the action information, the training parameters including a learning rate, a discount factor and an exploration rate;

[0021] obtain an action state value based on the state information and the action information through a preset strategy, the preset strategy formula being:

[0022]

[0023] wherein τ represents an exploration rate, |A| represents a size of the action information, argmax a Q w (s t ,a t ) represents finding an a value that maximizes (s t ,a t ), and (s t ,a t ) is obtained through the state information;

[0024] perform calculation according to the action state value to obtain actual comprehensive performance index data;

[0025] input the state information and the action information into an initial parking space allocation model to obtain predicted comprehensive performance index data;

[0026] obtain an error value between the actual comprehensive performance index data and the predicted comprehensive performance index data according to loss function calculation;

[0027] The training parameters are iteratively updated by an optimization algorithm according to the learning rate until a maximum number of iterations is reached or an error value calculated by a loss function converges to a preset threshold, so as to obtain a preset parking space allocation model.

[0028] In an embodiment, the step of inputting the parking request information and the parking space state data into the preset parking space allocation model to obtain comprehensive performance index data and a corresponding candidate allocation scheme comprises:

[0029] Obtaining a ratio of reserved parking spaces to non-reserved parking spaces;

[0030] Inputting the parking request information and the parking space state data into the preset parking space allocation model by dynamically adjusting the ratio to obtain comprehensive performance index data;

[0031] Converting based on the comprehensive performance index data to generate a plurality of candidate parking space allocation schemes.

[0032] In an embodiment, the step of optimizing the parking space allocation scheme by using a constraint optimization strategy to obtain a target parking space allocation scheme comprises:

[0033] Obtaining a constraint condition set according to the constraint optimization strategy, the constraint condition set comprising a specific parking space reservation ratio, a new energy parking space reservation ratio, and a non-reserved vehicle entry waiting time, the specific parking space comprising a disabled parking space or a charging pile parking space;

[0034] When the parking space allocation scheme does not satisfy any one of the specific parking space reservation ratio, the new energy parking space reservation ratio, and the non-reserved vehicle entry waiting time, optimizing the parking space allocation scheme by using a non-dominated sorting genetic algorithm to obtain a target parking space allocation scheme.

[0035] In an embodiment, before the step of obtaining parking request information and parking space state data, comprising:

[0036] Obtaining parking lot basic information and historical parking information;

[0037] Calculating based on the historical parking information to obtain a standard price and a standard parking demand, the formula being:

[0038]

[0039] wherein, T represents a period of time, k is the total number of parking spaces, p t represents the parking fee per unit time, λ t represents the parking space occupancy rate, p0 represents the standard price, and λ0 represents the standard parking demand.

[0040] According to the parking lot basic information, the standard price and the standard parking demand, a probability density function of the reservation user and a probability density function of the non-reservation user are obtained by function fitting, so as to input the parking request information and the parking space state data into a preset parking space allocation model to obtain comprehensive performance index data and a corresponding candidate allocation scheme.

[0041] In addition, to achieve the above object, the application further provides a parking lot parking space allocation result generation device, which comprises:

[0042] An acquisition module is configured to acquire parking request information and parking space state data.

[0043] A processing module is configured to input the parking request information and the parking space state data into a preset parking space allocation model to obtain comprehensive performance index data and a corresponding candidate allocation scheme, wherein the preset parking space allocation model is a neural network model with a double-network structure, and the comprehensive performance index data comprises parking fee income, parking space utilization rate and user waiting time.

[0044] A screening module is configured to screen the candidate allocation scheme by comparing the comprehensive performance index data to obtain a parking space allocation scheme.

[0045] A result module is configured to optimize the parking space allocation scheme by using a constraint optimization strategy to obtain a target parking space allocation scheme.

[0046] In addition, to achieve the above object, the application further provides a medium, which is a computer readable medium, and a computer program is stored on the medium, and the computer program is executed by a processor to implement the steps of the parking lot parking space allocation result generation method described above.

[0047] In addition, to achieve the above object, the application further provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement the steps of the parking lot parking space allocation result generation method described above.

[0048] The application collects parking request information and parking space state data, calculates comprehensive performance index data by using a preset parking space allocation model based on deep learning, and screens a plurality of candidate schemes according to the income. Then, a constraint optimization strategy is used for further optimization to ensure that specific parking space reservation ratios and non-reservation vehicle waiting time requirements are met, and finally a target parking space allocation scheme is obtained, which improves the parking lot resource utilization rate and reduces the user waiting time. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the accompanying drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced. Obviously, for those skilled in the art, based on these drawings, other drawings can also be obtained without creative labor.

[0050] Figure 1 Flowchart of the first embodiment of the parking space allocation result generation method of the parking lot of the present application;

[0051] Figure 2 Flowchart of the second embodiment of the parking space allocation result generation method of the parking lot of the present application;

[0052] Figure 3 Flowchart of the third embodiment of the parking space allocation result generation method of the parking lot of the present application;

[0053] Figure 4 Module structure diagram of the parking space allocation result generation device of the first embodiment of the parking space allocation result generation method of the parking lot of the present application;

[0054] Figure 5 Device structure diagram of the hardware running environment involved in the parking space allocation result generation method of the parking lot in the embodiments of the present application.

[0055] The object implementation, functional features and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0056] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application, and are not used to limit the present application.

[0057] In order to better understand the technical solutions of the present application, the specific embodiments will be described in detail below with reference to the drawings in the specification.

[0058] With the acceleration of global urbanization process, the number of motor vehicles continues to rise, and the urban static traffic management is facing unprecedented challenges. According to relevant statistical data, about 30% of the traffic congestion problems in major cities are directly caused by the difficulty in finding parking spaces, which not only leads to hundreds of billions of economic losses every year, but also increases environmental pollution and energy consumption. The traditional parking management system relies on fixed sensor network and manual scheduling mode, which has significant shortcomings in response speed and resource allocation efficiency, and cannot effectively cope with complex and variable urban traffic conditions.

[0059] Therefore, the application provides a method for efficiently and reasonably allocating parking space resources. The main solution of the embodiment of the application is: obtaining parking request information and parking space state data, the parking request information including reservation parking space information and reservation time length information; inputting the parking request information and the parking space state data into a preset parking space allocation model to obtain comprehensive performance index data and a corresponding candidate allocation scheme, the preset parking space allocation model being a neural network model with a double network structure, and the comprehensive performance index data including parking fee income, parking space utilization rate and user waiting time; screening the candidate allocation scheme by comparing the comprehensive performance index data to obtain a parking space allocation scheme; and optimizing the parking space allocation scheme by using a constraint optimization strategy to obtain a target parking space allocation scheme.

[0060] Based on the above, the embodiment of the application further provides a parking lot parking space allocation result generation method, which refers to Figure 1 , Figure 1 The flowchart of the first embodiment of the parking lot parking space allocation result generation method of the application is shown in FIG. 1.

[0061] In this embodiment, the parking lot parking space allocation result generation method includes steps S10-S40:

[0062] Step S10, obtaining parking request information and parking space state data.

[0063] It should be noted that this process relies on the deployment of sensing technology and intelligent devices, such as cameras, geomagnetic sensors, and license plate recognition systems, which can monitor the occupancy of each parking space in the parking lot in real time, obtain real-time parking space state (such as idle, occupied, reserved, etc.), and upload these information to the central management system of the parking lot. At the same time, through mobile applications or website platforms, users can submit parking request information, which covers the details of the user's parking needs, such as reservation time, estimated parking time, vehicle type (such as small car, large car, new energy car, etc.), and whether it is a reservation user, etc. These information not only contains the basic information of the user, but also covers the details of the estimated arrival time, departure time and special needs (such as disabled parking space) etc.

[0064] In addition, in order to ensure the accuracy and timeliness of the data, an efficient data collection and processing mechanism is established. On the one hand, the sensors deployed inside the parking lot need to have high sensitivity and stability to ensure the accurate collection of parking space state information; on the other hand, the communication network needs to support fast transmission of a large amount of data so that the central system can update the use of the parking space in real time. In addition, considering privacy protection and data security, all transmitted information should be encrypted to prevent unauthorized access.

[0065] Further, before step S10, it also includes obtaining parking lot basic information and historical parking information, the above basic information includes total number of parking spaces, unit time charging sequence of different charging periods and demand elasticity of parking spaces, and the historical parking information covers parking fee and parking space occupancy rate in a specific time period. According to the historical parking information, the standard price and the standard parking demand are calculated, and the formula is:

[0066]

[0067] Wherein, T represents a period of time, k is the total number of parking spaces, p t represents the unit time parking fee, λ t represents the parking space occupancy rate, p0 represents the standard price, λ0 represents the standard parking demand. According to the parking lot basic information, the standard price and the standard parking demand, the probability density function of the reservation user and the probability density function of the non-reservation user are obtained by function fitting, so as to execute the steps of S20. Specifically, the parking demand distribution of reservation and non-reservation users in different time periods can be estimated by regression analysis, assuming that it follows normal distribution or other suitable distribution model. The purpose of doing so is to more accurately predict the future parking demand mode, so as to provide input data for the preset parking space allocation model. Finally, these processed parking request information and parking space state data are input into the Double DQN-based parking space allocation model to generate a preliminary parking space allocation scheme.

[0068] Step S20, input the parking request information and parking space state data into the preset parking space allocation model to obtain comprehensive performance index data and corresponding candidate allocation scheme.

[0069] It should be noted that the preset parking space allocation model includes a main network and a target network, the main network is used to update and learn the optimal parking space allocation strategy in real time, and the target network is used to stabilize the learning process, and the weights of the main network are copied to the target network every preset time step or preset learning period, specifically, the preset parking space allocation model adopts a neural network model with a double network structure, and through the cooperative work of the main network and the target network, the parking space allocation strategy can be dynamically adjusted. The main network is the core of the model, which is responsible for real-time processing of input parking request information and parking space state data. Through a deep learning algorithm, the main network continuously updates its weights to learn the optimal parking space allocation strategy. It can dynamically adjust the proportion of reserved parking spaces and non-reserved parking spaces, optimize the parking space allocation scheme, and thus maximize the comprehensive performance index data, such as parking fee income, parking space utilization rate, and user waiting time, etc. The target network plays the role of stabilizer. Its structure is the same as that of the main network, but the frequency of weight update is lower. Every preset time step or learning period (for example, every 100 iterations or every hour), the weights of the main network are copied to the target network. This mechanism helps to reduce fluctuations in the learning process and avoid instability caused by frequent updates. The stability of the target network provides a reliable benchmark for the model, enabling the main network to explore and learn in a more stable environment.

[0070] Meanwhile, the preset parking space allocation model includes an input layer, a hidden layer, and an output layer. The input layer includes a first preset number of nodes, the hidden layer uses a ReLU activation function, and the output layer includes a second preset number of nodes. Specifically, the input layer includes a first preset number of nodes for receiving the current state (such as the state of each parking space in the parking lot, user parking request information, etc.) and possible actions (such as the proportion of parking spaces allocated to users with reservations). These information is encoded into a feature vector and input into the network, preparing for subsequent processing. The hidden layer is composed of multiple layers of fully connected networks and uses a ReLU activation function to increase the non-linear expression capability, so that the model can learn the complex patterns in the data. The ReLU activation function helps to speed up the training process and reduce the gradient vanishing problem by setting negative values to zero while preserving positive values. The output layer includes a second preset number of nodes, each node corresponding to a value evaluation (Q value) of an action, i.e., the expected return of selecting a specific action. The neural network model with a double network structure can process complex input data and obtain comprehensive performance indicator data based on the input data. Then, based on the comprehensive performance indicator data, the corresponding candidate allocation scheme can be generated, and each scheme corresponds to a comprehensive performance indicator data. The above comprehensive performance indicator data includes parking fee revenue, parking space utilization rate, and user waiting time. Parking fee revenue is a direct economic benefit, reflecting the total revenue obtained by the parking lot through parking space allocation. Different parking space types and parking durations will generate different revenues. Parking space utilization rate is a measure of the efficiency of parking spaces in the parking lot, i.e., the proportion of time the parking space is occupied. High utilization rate means that resources are fully utilized, thereby indirectly improving economic benefits. User waiting time is the average waiting time for users to find a parking space from entering the parking lot. Shorter waiting time can improve user satisfaction and reduce additional fuel consumption and environmental pollution caused by searching for parking spaces.

[0071] This neural network model based on a double network structure not only maximizes parking fee revenue, but also improves parking space utilization rate while minimizing user waiting time, thereby achieving efficient use of parking lot resources and improving user satisfaction.

[0072] Step S30, the candidate allocation scheme is screened by comparing the comprehensive performance indicator data, and the parking space allocation scheme is obtained.

[0073] It is worth noting that by comparing the Q values of all possible actions, the model can determine the optimal action strategy to achieve the optimization goal, such as maximizing parking lot revenue or minimizing user waiting time. Next, according to the set target (such as maximizing parking lot revenue or minimizing user waiting time), the preliminary parking space allocation scheme will be obtained. This process involves complex calculations and optimization, not just simple numerical comparisons, but also detailed analysis in combination with actual situations. For example, although a certain scheme may perform outstandingly in direct economic benefits, if it leads to excessive waiting time for non-reserved users, it may reduce overall user experience and service satisfaction. Therefore, in the screening process, in addition to economic benefits, other key indicators such as the reservation ratio of specific parking spaces (such as disabled parking spaces or charging pile parking spaces), the reservation ratio of new energy parking spaces, and the waiting time of non-reserved vehicles entering the parking lot, etc. are also considered to ensure balance and optimization in all aspects.

[0074] Step S40, the parking space allocation scheme is optimized using a constraint optimization strategy to obtain a target parking space allocation scheme.

[0075] It is worth noting that when optimizing the parking space allocation scheme, the constraint optimization strategy ensures that the final scheme not only maximizes the utilization of parking lot resources, but also meets the specific needs of various users. First, based on the preliminary parking space allocation scheme, the system will evaluate and adjust according to a series of preset constraints. These constraints include but are not limited to: the reservation ratio of specific parking spaces (such as disabled parking spaces or charging pile parking spaces), the reservation ratio of new energy parking spaces, and the waiting time of non-reserved vehicles entering the parking lot.

[0076] Further, step S40 further comprises obtaining a constraint condition set according to a constraint optimization strategy, the constraint condition set comprising a specific parking space reservation ratio, a new energy parking space reservation ratio, and a non-reserved vehicle entry waiting time length, the specific parking space comprising a disabled person parking space or a charging pile parking space. Specifically, based on the preliminarily generated parking space allocation scheme, it is checked whether the specific parking space and the new energy parking space reservation ratio requirements are met. For example, according to relevant regulations, a certain proportion of special parking spaces are provided for disabled people in the parking lot, and with the increase in the number of electric vehicles, enough charging pile parking spaces also need to be reserved to meet the needs of new energy vehicle users. In addition, in order to ensure service quality, a reasonable non-reserved vehicle entry waiting time length limit also needs to be set to avoid the decline in user experience caused by long waiting time. When the parking space allocation scheme does not meet any of the specific parking space reservation ratio, the new energy parking space reservation ratio, and the non-reserved vehicle entry waiting time length, a non-dominated sorting genetic algorithm is used to optimize the parking space allocation scheme to obtain a target parking space allocation scheme. Specifically, in actual operation, there may be a case that the parking space allocation scheme does not completely meet the above constraint conditions. At this time, it is particularly crucial to further optimize the parking space allocation scheme using the non-dominated sorting genetic algorithm (NSGA). NSGA is a multi-objective optimization algorithm that searches for the optimal solution set of a problem by simulating the natural selection process. Specifically, the algorithm can find a balance point between multiple conflicting objectives, such as maximizing parking space utilization while minimizing non-reserved vehicle waiting time, or optimizing overall parking space allocation while ensuring specific parking space reservation ratio. When applying NSGA for optimization, an fitness function needs to be defined first, which considers multiple objectives such as specific parking space reservation ratio, new energy parking space reservation ratio, and non-reserved vehicle entry waiting time length. Then, through an iterative process, individuals (i.e., different parking space allocation schemes) in the population are constantly adjusted, and genetic operators such as selection, crossover, and mutation are used to generate new generations of solutions. Each generation evaluates the performance of individuals and selects outstanding individuals based on non-dominated relationship to enter the next generation until the preset stopping condition (such as maximum iteration number or convergence standard) is reached. Finally, the parking space allocation scheme obtained after NSGA optimization not only meets all constraint conditions, but also achieves a good balance between multiple optimization objectives. This not only improves the efficiency of parking lot resources, but also enhances user experience and service satisfaction.

[0077] In addition, to further enhance reliability and adaptability, the target parking space allocation scheme will also be subjected to simulation testing. By simulating different parking demand scenarios (such as peak hours, holidays, etc.), the performance of the scheme under various conditions is evaluated. This step helps to discover potential problems and make timely adjustments, so as to ensure efficient operation even in the face of unexpected situations. Once the final parking space allocation scheme is determined, it will be deployed to the actual operating environment. This means that the parking lot management system will automatically adjust the parking space allocation strategy according to the new scheme, responding to user parking request information in real time. In addition, considering possible future changes, the ability to continuously learn and self-optimize is also required. For example, a Double DQN-based model can continue to learn from new data, constantly improving the decision-making process, so that parking space allocation is more accurate and effective.

[0078] The embodiment collects parking request information and parking space status data, calculates comprehensive performance index data using a pre-set deep learning-based parking space allocation model, and screens multiple candidate schemes. Then, a constraint optimization strategy is used for further optimization to ensure that specific parking space reservation ratios and non-reserved vehicle waiting times are met, and finally a target parking space allocation scheme is obtained, improving parking lot resource utilization and reducing user waiting time.

[0079] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as the above embodiment one can refer to the above introduction, and the subsequent will not be repeated. On this basis, please refer to Figure 2 , the parking lot parking space allocation result generation method step S20 further includes steps S201-S202 before it:

[0080] Step S201, obtaining parking request information sample data and parking space status sample data, and establishing an initial parking space allocation model.

[0081] It should be noted that the parking request information sample data includes detailed information such as the reservation time, the estimated arrival time and the departure time of the user. By collecting these data, the parking behavior patterns of different types of users (such as reservation users and non-reservation users) can be understood. At the same time, the parking space state sample data reflects the real-time occupancy of each parking space in the parking lot, which is particularly crucial for dynamically adjusting the parking space allocation strategy. Based on these comprehensive sample data, the initial parking space allocation model is established next. The model design needs to consider the reasonable configuration of the input layer, the hidden layer and the output layer of the neural network. The deep learning technology, especially the Double DQN-based framework, is used to initialize the training network and the target network. Both networks use a multi-layer fully connected structure, and the double network structure includes a main network and a target network. The main network is used to update and learn the optimal parking space allocation strategy in real time, and the target network is used to stabilize the learning process. Every preset time step or learning period, the weights of the main network are copied to the target network to reduce fluctuations in the learning process and improve the convergence speed of the model. The ReLU activation function is used to enhance the expression ability of the model, and the ReLU function is defined as ReLU(x) = max(0, x), which sets the part of the input value less than 0 to 0 and retains the part greater than 0. This nonlinear activation method not only effectively alleviates the gradient vanishing problem, improves the model training speed, but also increases the nonlinear fitting ability of the model, so that it can better capture the complex relationship between the parking request and the parking space state data, and thus more accurately generate the parking space allocation strategy.

[0082] In step S202, the initial parking space allocation model is trained according to the parking request information sample data and the parking space state sample data to obtain a preset parking space allocation model.

[0083] It should be understood that the initial preset parking space allocation model is trained, and the parking request information sample data and the parking space state sample data are first normalized to construct a training set. The state information, the action information and the reward function are defined, the state information includes the occupancy of the current parking lot, the time period, etc.; the action information relates to the parking space allocation strategies available for selection; and the reward function is used to quantify the benefits brought by a certain action. For this embodiment, the reward function is the expected income of the reservation and non-reservation users, and the reward function is represented as:

[0084]

[0085] wherein T represents a period of time, π1 represents the expected income for the reservation parking user, a t and b t represent the parameters of the reward function, p t represents the parking fee per unit time, π2 is the expected income for the non-reservation parking user, θ is the proportion of the bookable parking space, k is the total number of parking spaces, is the reservation rate that the reservation parking user needs to pay, is the demand of the reservation parking user, is the random variable of the number of reservation parking users, is the probability density function of the reservation user, is the demand of the non-reservation parking user, is the random variable of the number of non-reservation parking users, is the probability density function of the non-reservation user, the reservation user can see the price online, the demand is related to the price, so it is obtained by linear relationship, and the non-reservation user does not know the price in advance, which is represented as a random variable.

[0086] Initialize training parameters, state information and action information, the above training parameters include learning rate, discount factor and exploration rate, the learning rate determines the amplitude of updating the weight each time, the discount factor affects the importance of future rewards, and the exploration rate controls the degree of model exploring new strategies. The action state value is obtained based on the state information and the action information through a preset strategy, in this embodiment, the preset strategy is a greedy strategy, and the preset strategy formula is:

[0087]

[0088] wherein, τ represents the exploration rate, | A | represents the space size of the action information, argmax a Q w (s t ,a t ) represents finding the a value that makes (s t ,a t ) maximum, (s t ,a t) The action that can bring the maximum expected return is selected according to the current state through state information acquisition. The actual comprehensive performance index data is calculated according to the action state value. The state information and action information are input into the initial parking space allocation model to obtain the predicted comprehensive performance index data. Specifically, the actual comprehensive performance index data is calculated based on the action state value. This process involves applying the currently selected action (such as a specific parking space allocation strategy) to a specific state (such as the current parking lot occupancy and user requests), and calculating the direct return generated therefrom according to the reward function. The expected return for reservation and non-reservation users needs to be evaluated comprehensively in combination with their respective probability density functions. For example, by analyzing the random variables of the number of reservation parking users and the number of non-reservation parking users in different time periods, as well as their corresponding probability density functions, the corresponding expected return is calculated. Next, the state information (such as the real-time occupancy rate of the parking lot, peak hours, etc.) and the action information (such as the adjusted parking space allocation ratio) are input into the initial parking space allocation model to obtain the predicted comprehensive performance index data. This model is usually based on deep learning algorithms (such as Double DQN) and can learn the best mapping from state to action, thereby maximizing long-term returns. In this stage, the model will predict the return that can be obtained after taking a certain action based on historical data and the current state, providing a basis for subsequent decision-making. According to the loss function calculation, the error value between the actual comprehensive performance index data and the predicted comprehensive performance index data is obtained. By comparing the error value between the actual comprehensive performance index data and the predicted comprehensive performance index data, the accuracy and reliability of the model can be evaluated, and the specific formula is:

[0089]

[0090] where N represents the total number of samples, q a is the target Q value. The loss function is a key indicator that measures the difference between the model's predicted value and the true value, and is used to update the gradient of the model's training parameters w through the backpropagation algorithm. Specifically, according to the error value of the loss function, the error gradient of the output layer is calculated. Then, using the chain rule, the error gradient of the output layer is combined with the derivative of the activation function of the previous layer to calculate the error gradient of the previous layer. This process continues until the gradient of the model parameters of each layer of the entire network is calculated. These gradient information is used to update the network parameters to minimize the loss function. If the error value is large, it indicates that the model needs to further adjust the training parameters to optimize the model. According to the learning rate, the training parameters are iteratively updated by the optimization algorithm until the maximum number of iterations is reached or the error value calculated by the loss function converges to the preset threshold, obtaining the preset parking space allocation model. By means of cross-validation and other means, the model hyperparameters (such as learning rate) are constantly adjusted to optimize the model performance.

[0091] The embodiment utilizes parking request information and parking space state sample data, establishes and trains an initial parking space allocation model through deep learning, and further obtains an optimized preset parking space allocation model, thereby improving parking space utilization rate and rationality.

[0092] Based on the first embodiment of the present application, the same or similar contents as the above embodiment one can be referred to the above introduction, and will not be described in detail hereinafter. On this basis, please refer to Figure 3 , the parking lot parking space allocation result generation step S20 further comprises steps S301-S303:

[0093] Step S301, the proportion of reserved parking spaces and non-reserved parking spaces is obtained.

[0094] It should be noted that before obtaining, parking request information sample data and parking space state sample data need to be collected, including parking demand distribution of reserved and non-reserved users in different time periods, total number of parking spaces of the parking lot, and occupancy of each period and other information. Based on these data, the ideal proportion of reserved parking spaces and non-reserved parking spaces is determined by analyzing the historical parking mode. For example, during the peak period of weekdays, due to the tendency of commuters to reserve parking spaces in advance, a higher proportion of parking spaces needs to be reserved for reserved users; while on weekends or holidays, the temporary demand of non-reserved users will increase, so the proportion of non-reserved parking spaces needs to be adjusted accordingly. In addition, considering the needs of special user groups (such as disabled people or electric vehicle users), the reservation proportion of specific parking spaces also needs to be further subdivided.

[0095] Step S302, the parking request information and parking space state data are input into the preset parking space allocation model by dynamically adjusting the proportion, and the comprehensive performance index data is obtained.

[0096] It should be noted that by real-time monitoring and analyzing parking request information sample data and parking space state sample data, the proportion of the two types of parking spaces can be dynamically adjusted to adapt to the changing parking demand.

[0097] Further, the above data is input into the preset Double DQN-based parking space allocation model, which can obtain comprehensive performance index data according to the current state of the parking lot (such as parking occupancy rate, peak period, etc.) and the parking request information of the user (including reservation and non-reservation). The above comprehensive performance index data comprehensively considers multiple factors such as parking fee income of the parking lot, parking space utilization rate and user waiting time. Assuming that a scheme reserves 60% of the parking spaces for reserved users, the system will calculate whether the total income of the parking lot under this allocation scheme is maximized, while also considering whether the average waiting time of non-reserved users remains within an acceptable range.

[0098] Step S303, based on the comprehensive performance index data, conversion is performed to generate a plurality of candidate parking space allocation schemes.

[0099] It should be noted that the comprehensive performance index data is used to generate a plurality of different parking space allocation schemes, i.e., candidate parking space allocation schemes. Each candidate parking space allocation scheme represents a specific proportion of reserved parking spaces and non-reserved parking spaces. For example, during the morning peak period of a working day, a plurality of schemes are generated, one of which may be more inclined to provide more parking spaces for reserved users, and the other may be inclined to balance the needs of reserved and non-reserved users. Each scheme is evaluated according to its expected comprehensive performance index data, which takes into account factors such as total parking lot revenue, parking lot turnover rate, and user satisfaction.

[0100] Specifically, for each scheme, the effect after implementation is simulated, and the performance of the scheme in different scenarios is evaluated by a deep reinforcement learning (Double DQN) model. For example, during the morning peak period of a working day, the system may generate a plurality of schemes, one of which may be more inclined to provide more parking spaces for reserved users, and the other may be inclined to balance the needs of reserved and non-reserved users. Each scheme is evaluated according to its expected comprehensive performance index data, which calculates whether the total parking lot revenue is maximized under this allocation scheme, while also considering whether the average waiting time of non-reserved users remains within an acceptable range.

[0101] The embodiment dynamically adjusts the proportion of reserved and non-reserved parking spaces, generates a plurality of comprehensive performance index data and candidate schemes using a preset model, provides a basis for obtaining the optimal parking space allocation scheme, improves resource utilization, optimizes user experience, and enhances system flexibility, effectively alleviating the parking problem.

[0102] Based on the first embodiment of the present application, the present application also provides a parking lot parking space allocation result generation device, please refer to Figure 4 , the device comprises:

[0103] The acquisition module 10 is configured to acquire parking request information and parking space state data.

[0104] The processing module 20 is configured to input the parking request information and the parking space state data into a preset parking space allocation model to obtain comprehensive performance index data and corresponding candidate allocation schemes.

[0105] The screening module 30 is configured to screen the candidate allocation schemes by comparing the comprehensive performance index data to obtain a parking space allocation scheme.

[0106] The result module 40 is configured to optimize the parking space allocation scheme using a constraint optimization strategy to obtain a target parking space allocation scheme.

[0107] The parking lot parking space allocation result generation device provided by the application adopts the parking lot parking space allocation result generation method in the above embodiments, and can solve the technical problem of how to efficiently and reasonably allocate parking space resources. Compared with the prior art, the parking lot parking space allocation result generation device provided by the application has the same beneficial effects as the parking lot parking space allocation result generation method provided by the above embodiments, and other technical features in the parking lot parking space allocation result generation device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0108] In an embodiment, the processing module 20 is further configured to obtain parking request information sample data and parking space state sample data, and establish an initial parking space allocation model; train the initial parking space allocation model according to the parking request information sample data and the parking space state sample data, and obtain a preset parking space allocation model.

[0109] In an embodiment, the processing module 20 is further configured to normalize the parking request information sample data and the parking space state sample data, and construct a training set; define state information, action information and a reward function; initialize the training parameters, the state information and the action information; obtain an action state value based on the state information and the action information through a preset strategy; calculate the actual comprehensive performance index data according to the action state value; input the state information and the action information into the initial parking space allocation model to obtain predicted comprehensive performance index data; calculate the error value between the actual comprehensive performance index data and the predicted comprehensive performance index data according to a loss function; and update the training parameters through an optimization algorithm according to a learning rate until a maximum iteration number is reached or the error value calculated by the loss function converges to a preset threshold, thereby obtaining the preset parking space allocation model.

[0110] In an embodiment, the screening module 30 is further configured to obtain the proportion of reserved parking spaces and non-reserved parking spaces; input the parking request information and the parking space state data into the preset parking space allocation model through dynamic adjustment of the proportion to generate a plurality of candidate parking space allocation schemes; and calculate the comprehensive performance index data of the plurality of candidate parking space allocation schemes.

[0111] In an embodiment, the result module 40 is further configured to obtain the specific parking space reservation proportion, the new energy parking space reservation proportion and the non-reserved vehicle entry waiting time length according to a constraint optimization strategy; and when the parking space allocation scheme does not satisfy any one of the specific parking space reservation proportion, the new energy parking space reservation proportion and the non-reserved vehicle entry waiting time length, the non-dominated sorting genetic algorithm is used to optimize the parking space allocation scheme to obtain a target parking space allocation scheme.

[0112] In an embodiment, the acquisition module 10 is further configured to acquire parking lot basic information and historical parking information; calculate a standard price and a standard parking demand according to the historical parking information; and perform function fitting according to the parking lot basic information, the standard price and the standard parking demand to obtain a probability density function of the reservation users and a probability density function of the non-reservation users, so as to input the parking request information and the parking space state data into the preset parking space allocation model to obtain the comprehensive performance index data.

[0113] The present application provides a parking lot parking space allocation result generation device, which comprises at least one processor and a memory in communication connection with the at least one processor; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the parking lot parking space allocation result generation method in the above-mentioned embodiment one.

[0114] Reference will be made to the following description Figure 5 which shows a structural diagram of a parking lot parking space allocation result generation device suitable for use to implement the embodiments of the present application. The parking lot parking space allocation result generation device in the embodiments of the present application can include, but is not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (Personal Digital Assistant), PADs (Portable Application Description), PMPs (Portable MediaPlayer), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), and the like, and fixed terminals such as digital TVs, desktop computers, and the like. Figure 5 The parking lot parking space allocation result generation device shown is merely an example and should not bring any limitation to the functions and use range of the embodiments of the present application.

[0115] As Figure 5As shown, the parking lot parking space allocation result generation device can include a processing device 1001 (for example, a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 1002 or programs loaded from a storage device 1003 to a random access memory (RAM) 1004. In the RAM 1004, various programs and data required for the parking lot parking space allocation result generation device operation are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; the storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the parking lot parking space allocation result generation device to communicate with other devices wirelessly or wired to exchange data. Although a parking lot parking space allocation result generation device with various functions is shown in the figure, it should be understood that it is not required to implement or have all the shown. More or less can be implemented or provided instead.

[0116] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing program codes for executing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network through a communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the methods of the embodiments disclosed in the present application are performed.

[0117] The parking lot parking space allocation result generation device provided by the present application adopts the parking lot parking space allocation result generation method in the above-mentioned embodiments, and can solve the technical problem of how to efficiently and reasonably allocate parking space resources. Compared with the prior art, the parking lot parking space allocation result generation device provided by the present application has the same beneficial effects as the parking lot parking space allocation result generation method provided by the above-mentioned embodiments, and other technical features in the parking lot parking space allocation result generation device are the same as the features disclosed in the previous embodiment method, which will not be repeated here.

[0118] It should be understood that various parts of the present application can be realized with hardware, software, firmware, or a combination thereof. In the above description of embodiments, specific functional configurations, structures, materials or characteristics can be combined in any appropriate manner in one or more embodiments or examples.

[0119] The above description is merely a specific implementation of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, and all such changes or replacements should be encompassed within the scope of protection of the present application. Therefore, the scope of protection of the present application should be subject to the scope of protection of the claims.

[0120] The present application provides a computer readable medium having stored thereon computer readable program instructions (i.e., computer programs) for performing the parking space allocation result generation method in the above-described embodiments.

[0121] The computer readable medium provided by the present application may, for example, be a U disk, but is not limited to an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor, or device, or any combination thereof. More specific examples of the computer readable medium can include, but are not limited to, an electrical connection having one or more conductive wires, a portable computer disk, a hard disk, a random access memory (RAM), a read only memory (ROM), an erasable programmable read only memory (EPROM or flash memory), an optical fiber, a portable compact disk read only memory (CD-ROM), an optical storage device, a magnetic storage device, or any appropriate combination thereof. In the present embodiment, the computer readable medium can be any tangible medium containing or storing a program that can be executed by an instruction execution device or used in conjunction therewith. The program code contained on the computer readable medium can be transmitted by any appropriate medium, including but not limited to an electrical wire, an optical cable, an RF (Radio Frequency), or the like, or any appropriate combination thereof.

[0122] The above-described computer readable medium can be contained in the parking space allocation result generation device; or can exist separately without being assembled into the parking space allocation result generation device.

[0123] The computer-readable medium can be a machine-readable storage medium, including but not limited to diskette, RAM, ROM, EEPROM, solid state drive (SSD) or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. When the one or more programs are implemented by hardware, the one or more programs can be specially configured and implemented hardware. In some embodiments, the one or more programs can be implemented by one or more of the following: a processor, a controller, a microcontroller, a microprocessor, a central processing unit (CPU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or any other suitable programmable logic device. When the one or more programs are implemented by software, the one or more programs can be stored in the computer-readable medium and executed by the processor. The computer-readable medium can be a machine-readable storage medium, including but not limited to diskette, RAM, ROM, EEPROM, solid state drive (SSD) or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer.

[0124] The flowcharts and block diagrams in the drawings illustrate the possible implementation architectures, functions and operations of the methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowcharts or block diagrams can represent a module, a program segment, or a part of code containing one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in different order than that shown in the drawings. For example, two blocks that are shown in succession can actually be executed substantially in parallel, and sometimes in reverse order, depending on the involved functions. It should also be noted that each block in the block diagrams and / or flowcharts, and the combination of blocks in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0125] The modules involved in the embodiments of the present application can be implemented by software or by hardware. In some cases, the name of the module does not constitute a limitation on the unit itself.

[0126] The readable medium provided by the application is a computer readable medium, which stores computer readable program instructions (i.e. computer program) for executing the parking lot parking space allocation result generation method described above, and can solve the technical problem of how to efficiently and reasonably allocate parking space resources. Compared with the prior art, the computer readable medium provided by the application has the same beneficial effects as the parking lot parking space allocation result generation method provided by the above-mentioned embodiments, and will not be repeated here.

[0127] The application also provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of the parking lot parking space allocation result generation method as described above.

[0128] The computer program product provided by the application can solve the technical problem of how to efficiently and reasonably allocate parking space resources. Compared with the prior art, the computer program product provided by the application has the same beneficial effects as the parking lot parking space allocation result generation method provided by the above-mentioned embodiments, and will not be repeated here.

[0129] The above-mentioned is only part of the embodiments of the application, and does not limit the patent scope of the application, and any equivalent structural transformation, direct / indirect application in other related technical fields within the technical concept of the application, using the content of the application specification and drawings, are included in the patent protection scope of the application.

Claims

1. A method for generating a parking space allocation result in a parking lot, characterized by, The method comprises: acquiring parking request information and parking space state data; inputting the parking request information and the parking space state data into a preset parking space allocation model to obtain comprehensive performance index data and a corresponding candidate allocation scheme, the preset parking space allocation model being a neural network model with a double-network structure, and the comprehensive performance index data including parking fee income, parking space utilization rate and user waiting time; screening the candidate allocation scheme by comparing the comprehensive performance index data to obtain a parking space allocation scheme; optimizing the parking space allocation scheme using a constraint optimization strategy to obtain a target parking space allocation scheme; Before the step of inputting the parking request information and the parking space state data into the preset parking space allocation model to obtain the comprehensive performance index data and the corresponding candidate allocation scheme, the method comprises: acquiring parking request information sample data and parking space state sample data, and establishing an initial parking space allocation model; training the initial parking space allocation model according to the parking request information sample data and the parking space state sample data to obtain the preset parking space allocation model; The step of training the initial parking space allocation model according to the parking request information sample data and the parking space state sample data to obtain the preset parking space allocation model comprises: normalizing the parking request information sample data and the parking space state sample data to construct a training set; defining state information, action information and a reward function, wherein the reward function is expressed as: wherein, denotes a period of time, denotes the expected revenue for the reserved parking users, and denotes a parameter of the reward function, denotes the parking fee per unit time, is the expected revenue for the non-reserved parking users, is the proportion of reserved parking spaces, is the total number of parking spaces, is the reservation rate paid by the reserved parking users, is the demand for the reserved parking users, is a random variable of the number of reserved parking users, is the probability density function of the reserved users, is the demand for the non-reserved parking users, is a random variable of the number of non-reserved parking users, is the probability density function of the non-reserved users; initializing training parameters, the state information and the action information, the training parameters including a learning rate, a discount factor and an exploration rate; obtaining an action state value based on the state information and the action information through a preset strategy, the preset strategy formula being: wherein, represents an exploration rate, represents a spatial size of action information, represents finding a value maximizing a value, by the state information acquisition, represents predicted comprehensive performance index data, represents state information, represents an action state value; calculating according to the action state value to obtain actual comprehensive performance index data; inputting the state information and the action information into the initial parking space allocation model to obtain predicted comprehensive performance index data; calculating an error value between the actual comprehensive performance index data and the predicted comprehensive performance index data according to a loss function; iteratively updating the training parameters according to the learning rate through an optimization algorithm until a maximum iteration number is reached or the error value calculated by the loss function converges to a preset threshold value, to obtain the preset parking space allocation model.

2. The method of claim 1, wherein, The preset parking space allocation model comprises a main network and a target network, the main network being used to update and learn an optimal parking space allocation strategy in real time, and the target network being used to stabilize the learning process, the weights of the main network being copied to the target network every preset time step or preset learning period.

3. The method of claim 1, wherein, The step of inputting the parking request information and the parking space state data into the preset parking space allocation model to obtain the comprehensive performance index data and the corresponding candidate allocation scheme comprises: acquiring a ratio of reserved parking spaces to non-reserved parking spaces; inputting the parking request information and the parking space state data into the preset parking space allocation model by dynamically adjusting the ratio to obtain comprehensive performance index data; generating a plurality of candidate parking space allocation schemes based on the comprehensive performance index data.

4. The method of claim 1, wherein, The step of optimizing the parking space allocation scheme by using a constraint optimization strategy to obtain a target parking space allocation scheme comprises: A constraint condition set is obtained according to a constraint optimization strategy, and the constraint condition set comprises a specific parking space reservation ratio, a new energy parking space reservation ratio and a non-reservation vehicle entry waiting time length, and the specific parking space comprises a disabled person parking space or a charging pile parking space; When the parking space allocation scheme does not satisfy any one of the specific parking space reservation ratio, the new energy parking space reservation ratio and the non-reservation vehicle entry waiting time length, a non-dominated sorting genetic algorithm is used to optimize the parking space allocation scheme to obtain a target parking space allocation scheme.

5. The method of claim 1, wherein, Before the steps of obtaining parking request information and parking space state data, the method comprises: obtaining parking lot basic information and historical parking information; standard price and standard parking demand are obtained by calculation according to the historical parking information, and the formula is: wherein, denotes a period of time, is the total number of parking spaces, denotes the parking fee per time unit, denotes the occupancy rate of the parking spaces, denotes the standard price, denotes the standard parking demand; The probability density function of the reservation user and the probability density function of the non-reservation user are obtained by function fitting according to the parking lot basic information, the standard price and the standard parking demand, so as to execute the steps of inputting the parking request information and the parking space state data into a preset parking space allocation model to obtain comprehensive performance index data and a corresponding candidate allocation scheme.

6. A parking lot parking space allocation result generating apparatus characterized by comprising: The device comprises: an acquisition module configured to acquire parking request information and parking space state data; a processing module configured to input the parking request information and the parking space state data into a preset parking space allocation model to obtain comprehensive performance index data and a corresponding candidate allocation scheme, wherein the preset parking space allocation model is a neural network model with a double-network structure, and the comprehensive performance index data comprises parking fee income, parking space utilization rate and user waiting time; the processing module is further configured to acquire parking request information sample data and parking space state sample data, and establish an initial parking space allocation model; train the initial parking space allocation model according to the parking request information sample data and the parking space state sample data to obtain the preset parking space allocation model; normalize the parking request information sample data and the parking space state sample data to construct a training set; define state information, action information and a reward function; initialize the training parameters, the state information and the action information, wherein the training parameters comprise a learning rate, a discount factor and an exploration rate; obtain an action state value based on the state information and the action information through a preset strategy; calculate an actual comprehensive performance index data according to the action state value; input the state information and the action information into the initial parking space allocation model to obtain a predicted comprehensive performance index data; calculate an error value between the actual comprehensive performance index data and the predicted comprehensive performance index data according to a loss function; and update the training parameters through an optimization algorithm according to the learning rate until a maximum iteration number is reached or the error value calculated by the loss function converges to a preset threshold value, so as to obtain the preset parking space allocation model; a screening module configured to screen the candidate allocation scheme by comparing the comprehensive performance index data to obtain a parking space allocation scheme. A result module is configured to optimize the parking space allocation scheme by using a constraint optimization strategy to obtain a target parking space allocation scheme.

7. A parking lot parking space allocation result generating apparatus characterized by comprising: The device comprises a memory, a processor, and a parking lot parking space allocation result generation program stored in the memory and running on the processor, and the parking lot parking space allocation result generation program is configured to implement the steps of the parking lot parking space allocation result generation method according to any one of claims 1-5.

8. A medium characterized by, The medium stores a parking lot parking space allocation result generation program, and the parking lot parking space allocation result generation program implements the steps of the parking lot parking space allocation result generation method according to any one of claims 1-5 when executed by a processor.

Images