Historical-data-free regional battery swap station deployment method based on electric two-wheeled traffic flow

Abstract

The invention discloses an electric two-wheeled vehicle flow-based regional battery swap station deployment method without historical data. The method comprises the steps of obtaining regional feature data and environmental data related to a battery swap station in a city; constructing and training a generation model based on an analysis result of the historical data of the deployed battery swap station area in combination with a de-noising diffusion probability model; constructing a battery swap station deployment optimization problem which enables the correlation between the number of battery swap stations and the two-wheeled vehicle area flow and the coverage area of the battery swap stations to reach the maximum balance; solving a battery swap station deployment optimization problem through a simulated annealing algorithm to obtain an optimal battery swap station deployment scheme; in the step-by-step solving process, calling the trained generation model to synchronously generate the electric two-wheeled traffic flow according to the regional feature data and the environmental data of the target region; and realizing the electrical changing station deployment of the target area through the optimal electrical changing station deployment scheme. According to the method, the accuracy and efficiency of solving the optimization problem of the electric swap station deployment and the practicability of the electric two-wheeled vehicle regional swap station deployment can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of battery replacement for electric two-wheeled vehicles, and in particular to a method for deploying battery replacement stations in areas without historical data based on the flow of electric two-wheeled vehicles. Background Art

[0002] Electric two-wheelers have become an important means of transportation for people's short-distance travel due to their fast, convenient, and green characteristics. However, the traditional charging method of two-wheeled electric vehicles has serious fire safety issues due to long charging time and irregular charging locations. Therefore, an innovative battery sharing and battery swapping model has emerged, which achieves fast and safe energy replenishment through "replacement instead of charging". In recent years, the battery swapping industry has entered a period of rapid development, which has prompted companies providing battery swapping services to expand their battery swapping business to new areas of the city. During the expansion process, it is urgent to solve the deployment problem of new battery swapping stations. A reasonable battery swapping station deployment plan is crucial to ensuring customer experience.

[0003] For battery swap companies, they prefer to place battery swap stations in areas with high user activity. In the battery swap scenario, the regional flow of two-wheeled electric vehicles is a kind of urban data that describes the dynamic flow pattern of users between urban areas. The flow includes inflow and outflow, which respectively represent the number of people entering or leaving a certain area within a given time interval. The regional flow of two-wheeled electric vehicles can reflect the user activity in different areas. Generally, a large regional flow of two-wheeled electric vehicles represents high user activity, and the number of battery swap stations in the area should also be large. However, there is a strong correlation between the size of the regional flow of two-wheeled electric vehicles and the deployment of battery swap stations. Based on a large-scale real battery swap data set provided by a battery swap company in a certain city, the applicant found that the Pearson index of the number of battery swap stations near the region and the size of the electric two-wheeled vehicle flow in the area was 0.715. In other words, when using regional flow to determine the deployment plan of battery swap stations, the impact of different deployment plans on the regional flow itself should be considered. However, before the actual deployment of battery swap stations, regional flow data is difficult to obtain, resulting in a deadlock dilemma in the deployment decision of battery swap stations in areas without historical data.

[0004] In recent years, generative artificial intelligence has developed rapidly. Among them, the denoising diffusion probability model has received widespread attention in various fields, including computer vision, natural language processing, time series, audio processing, graphics generation, and geospatial data generation, providing an opportunity to solve the deadlock dilemma of battery swap station deployment. Battery swap companies have placed battery swap stations in some areas of the city and stored historical traffic data through hardware devices. Based on this part of historical data, the denoising diffusion probability model can be trained to generate regional flows under different deployment schemes for the areas where battery swap stations are to be deployed (i.e., areas without historical data).

[0005] Based on the above background, the applicant focused on the deployment of battery swap stations in areas without historical data. Specifically, the data of the deployed battery swap station area was used to train the denoised diffusion probability model, and the corresponding regional traffic was generated for different battery swap station deployment schemes in areas without historical data. However, the implementation of the above scheme faces the following challenges:

[0006] 1) The generation of electric two-wheeled vehicle traffic is highly complex. The accuracy of regional traffic generation directly affects the decision-making plan for the deployment of battery swap stations. However, regional traffic is not only affected by the combined effects of multiple factors, such as user behavior, regional characteristics, and urban environment, but also has significant spatiotemporal dependence. Therefore, it is difficult to accurately generate electric two-wheeled vehicle traffic, resulting in poor practicality for the subsequent deployment of regional battery swap stations for electric two-wheeled vehicles.

[0007] 2) The optimization problem of battery swap station deployment is non-convex. Since the objective function contains multiple complex factors, such as the correlation between regional traffic and the number of battery swap stations and the trade-off between coverage area, there are multiple local optimal solutions in the solution space, which increases the difficulty of the algorithm to jump out of the local optimum, resulting in poor accuracy and efficiency in solving the battery swap station deployment optimization problem. Summary of the invention

[0008] In view of the deficiencies of the above-mentioned prior art, the technical problem to be solved by the present invention is: how to provide a method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles, and to build and train a generation model based on the analysis results of the historical data of the deployed battery swap station area combined with the denoising diffusion probability model, so as to generate the flow of electric two-wheeled vehicles in the target area and improve the practicality of the deployment of battery swap stations in the electric two-wheeled area. At the same time, a battery swap station deployment optimization problem is constructed for the target area so that the correlation between the number of battery swap stations and the flow of two-wheeled vehicles in the area and the coverage of the battery swap station are maximized, and the optimization problem is solved by the simulated annealing algorithm, so as to improve the accuracy and efficiency of solving the battery swap station deployment optimization problem.

[0009] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0010] The method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles includes:

[0011] S1: Obtain regional characteristic data and environmental data related to battery swap stations in the city;

[0012] S2: Build and train a generative model based on the analysis results of historical data of deployed battery swap stations combined with a denoised diffusion probability model;

[0013] Among them, the trained generation model is used to generate electric two-wheeled vehicle traffic in the target area;

[0014] S3: Construct a battery swap station deployment optimization problem for the target area to maximize the balance between the correlation between the number of battery swap stations and the regional flow of two-wheeled vehicles and the coverage of the battery swap stations;

[0015] S4: Solve the battery swap station deployment optimization problem through simulated annealing algorithm to obtain the optimal battery swap station deployment plan;

[0016] In the process of gradual solution, the trained generation model is called to synchronously generate the flow of electric two-wheeled vehicles according to the regional characteristic data and environmental data of the target area;

[0017] S5: Achieve the deployment of battery swap stations in the target area through the optimal battery swap station deployment plan.

[0018] Preferably, in step S2, the processing steps of training the generation model are as follows:

[0019] S201: Acquire original data x0 as training data, regional feature data and environmental data of deployed battery swap stations;

[0020] S202: inputting the regional characteristic data and the environmental data into the flow estimator to perform flow estimation, and outputting the predicted regional flow; constructing a loss function through the predicted regional flow and the actual regional flow, and optimizing the flow estimator model parameters until the loss function value tends to be stable;

[0021] S203: forward diffusion is performed by adding randomly sampled noise ε to the original data x0 to obtain noise data;

[0022] The noise data is consistent with the predicted regional flow output by the flow estimator. Normal distribution of

[0023] S204: Input the constructed urban knowledge graph and the sampled noise data into the denoising network for denoising and restoration, and predict the noise of the corresponding sampling step.

[0024] S205: Noise through prediction Construct the loss function with noise ε and reversely optimize the model parameters of the denoising network;

[0025] S206: Repeat steps S203 to S205 until the loss function value becomes stable.

[0026] Preferably, in step S202, the flow estimator is composed of a two-layer feedforward neural network, the activation function uses the Leaky ReLU function, and the loss function is

[0027] The output of the flow estimator is the predicted regional flow Combine regional characteristic data with predicted regional flow After being processed by the fully connected layer, they are concatenated and the concatenated features are input into the denoising network as conditions.

[0028] Preferably, in step S203, the steps of forward diffusion are as follows:

[0029] In the original denoising diffusion probability model, for the original data x0, the data after adding noise in the tth step is x t , the noise adding process is expressed as Where: ε t-1 Indicates sampling from a standard normal distribution, i.e., ε~N(0,I); β t represents the variance of the additive noise;

[0030] Defining β t =1-α t , then the noise adding process can be rewritten as in

[0031] Generated noise data x T It conforms to the complete standard normal distribution, that is, x T ~N(0,I).

[0032] Modify the original denoising diffusion probability model so that the noise data generated in each area conforms to the predicted regional flow output by the flow estimator. Normal distribution, that is, noisy data

[0033] Preferably, in step S204, the city knowledge graph is used to model the urban environment, including the following five facts:

[0034] <region BorderBy region> , indicating the adjacency relationship between regions;

[0035] <region NearBy region> , indicating the proximity relationship between regions;

[0036] <region SimilarFunc region> , indicating the functional similarity relationship between regions;

[0037] <POI LocateAt region> , indicating the area to which the POI belongs;

[0038] <POI CateOf category> , indicating the category to which the POI belongs;

[0039] Among them, a fact is a triple, which contains a head entity, a relation and a tail entity.

[0040] Preferably, in step S204, the denoising network is constructed based on the DiffWave model, and the bidirectional dilated convolution block in the DiffWave model is replaced by a knowledge-enhanced spatiotemporal module;

[0041] The knowledge-enhanced spatiotemporal module includes a spatial module and a temporal module. The spatial module uses a 1-layer R-GCN, and the temporal module uses a Transformer layer. The outputs of the spatial module and the temporal module are fused using the attention mechanism. The urban knowledge graph is converted into an urban knowledge graph through TuckER and embedded as the query of the attention mechanism to guide the fusion process.

[0042] Preferably, in step S205, the calculation formula of the loss function is expressed as:

[0043]

[0044] Preferably, in step S2, after the generation model is trained, the flow of electric two-wheeled vehicles is generated through the following steps:

[0045] S211: inputting the regional characteristic data of the target area into the flow estimator to perform flow estimation, and outputting the predicted regional flow^p;

[0047] S212: Based on the predicted regional traffic From obedience Sampling noise data from a normal distribution;

[0048] S213: Input the sampled noise data into the trained denoising network for denoising and restoration, and generate denoised data as the electric two-wheeled vehicle flow.

[0049] Preferably, in step S3, the battery swap station deployment optimization problem is expressed as:

[0050] Objective function:

[0051] Cost constraints:

[0052] Constraints on the number of battery swap stations:

[0053] in:

[0054] r=corr(Flow TR ,X TR );

[0055]

[0056] e j = l j +bss j+m j ;

[0057] Where: α represents the variable used to adjust the importance of the correlation and coverage of the battery swap station service; r represents the correlation between the number of battery swap stations and regional traffic; cov total represents the union of the service coverage of the battery swap station. The service coverage of the battery swap station is calculated as follows: if area r j If there is a battery swap station, then take r j The center point position c j Set the service radius r; e j represents the deployment cost of the battery swap station; B represents the total cost constraint; C represents the upper limit of the number of battery swap stations deployed in each area to avoid unrealistic allocation of the number of battery swap stations; x ij Indicates whether the jth battery swap station is deployed in the i-th area. ij =1 represents deployment, x ij =0 means not deployed; Flow TR Indicates the flow of electric two-wheeled vehicles in the target area; Flow i represents the flow of electric two-wheeled vehicles in the i-th area of ​​the target area; represents the flow of electric two-wheeled vehicles in the ith area during the tth period; u t It represents the influencing factor of the regional flow of electric vehicles in different time periods, which is obtained through the user's battery replacement habits to indicate the importance of regional flow in each time period, u t The closer the value is to 1, the more likely the user is to swap batteries during this period, and the greater the impact on regional traffic during this period; j Indicates land cost; bss j represents the cost of purchasing a battery swap station and batteries; m j Indicates the subsequent maintenance cost.

[0058] Preferably, in step S4, the processing steps of the simulated annealing optimization algorithm include:

[0059] S401: Determine the initial solution S0, initial temperature T0, cooling coefficient ε, total cost constraint B, upper limit C of regional deployment of battery swap stations, maximum number of iterations N, and minimum temperature T of the simulated annealing optimization algorithm min ;

[0060] S402: Initialization: Current solution S current ←S0, optimal solution S * ←S0, temperature T←T0, number of iterations k←0;

[0061] S403: From the current solution current The neighborhood generates a new solution S new , that is, change the current solution S current Generate new solution S from partial regional feature data innew ; The solution refers to the deployment plan of battery swap stations in the target area;

[0062] S404: When the new solution S new When the cost constraint of the battery swap station deployment optimization problem is met: call the generation model to generate the current solution S current And new solution S new The electric two-wheeled vehicle flow is then brought into the objective function of the battery swap station deployment optimization problem to calculate the current solution S current The objective function value of And the new solution S new The objective function value of

[0063] S405: Determine whether the If so, accept the new solution S current ←S new ; Otherwise, execute step S406;

[0064] S406: If the random number is satisfied Then accept the new solution S current ←S new ; Otherwise, the new solution will not be accepted;

[0065] S407: Call the generation model to generate the electric two-wheeled vehicle flow of the optimal solution S*, and then bring it into the objective function of the battery swap station deployment optimization problem to calculate the objective function value of the optimal solution S* When satisfied When , update the optimal solution S * ←S current ;

[0066] S408: Update temperature T←ε·T, iteration count k←k+1;

[0067] S409: Determine whether T>T min Or k<N: If so, return to step S403 and execute the next iteration; otherwise, end the iteration and output the optimal battery swap station deployment plan corresponding to the optimal solution S*.

[0068] Compared with the prior art, the method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles in the present invention has the following beneficial effects:

[0069] The present invention is based on the analysis results of the historical data of the deployed battery swap station area combined with the denoising diffusion probability model to construct and train the generation model. The trained generation model can generate the electric two-wheeled vehicle flow according to the regional characteristic data and environmental data of the target area. First, the denoising diffusion probability model generates high-quality data samples by gradually removing noise. In the generation of electric two-wheeled vehicle flow, it can capture the nonlinear relationship between complex regional characteristic data (such as battery swap station distribution, urban environment and user behavior, etc.) and two-wheeled vehicle flow, so as to accurately generate the electric two-wheeled vehicle flow and improve the practicality of the deployment of electric two-wheeled vehicle regional battery swap stations. Secondly, because the generation model learns the mapping from noise to data during the training process, the model can generate reasonable electric two-wheeled vehicle flow when facing new and unseen regional characteristic data and urban data, and can adapt to the dynamic changes of regional characteristic data and urban data. This generalization ability is particularly important for the deployment of battery swap stations in areas without historical data. Finally, the present invention constructs a generation model based on the denoising diffusion probability model and combines the analysis results of the historical data of the deployed battery swap station area. First, the inverse denoising network of the generative model is composed of spatial modules and temporal modules, with the aim of better capturing the complex spatiotemporal dependencies of electric two-wheeled vehicle traffic, thereby taking into account the impact of battery swap station distribution and user behavior on regional traffic. Second, an urban knowledge graph is constructed to simulate the urban environment and represent the complex relationships between entities. The urban knowledge graph is used as the input of the denoising network of the generative model, and the spatiotemporal modules are further fused using the attention mechanism under the guidance of the urban knowledge graph embedding, so as to better capture the impact of the urban environment on regional traffic. Third, the flow estimator uses regional features to estimate the size of the electric two-wheeled vehicle flow in advance, so as to be better suitable for situations where the flow differences between regions are significant.

[0070] The present invention constructs an optimization problem for battery swap station deployment that maximizes the balance between the correlation between the number of battery swap stations and the regional flow of two-wheeled vehicles and the coverage of the battery swap stations. First, by constructing the optimization problem, it can be ensured that the number and location of battery swap stations are closely related to the regional flow of two-wheeled vehicles, which can avoid excessive construction of battery swap stations and waste of resources, thereby improving the rationality of the deployment of battery swap stations in the electric two-wheeled vehicle area. Secondly, the optimization problem also takes into account the coverage of the battery swap stations to ensure that as many two-wheeled vehicle users as possible can easily access the battery swap stations, which helps to improve user satisfaction and the utilization rate of the battery swap stations. Finally, by balancing the number and coverage of battery swap stations, the optimization problem can reduce the construction and operation costs of battery swap stations while meeting user needs.

[0071] In view of the non-convexity of the optimization problem, the present invention solves the optimization problem through a simulated annealing algorithm. First, the battery swap station deployment optimization problem contains multiple complex factors, such as the correlation between regional traffic and the number of battery swap stations and the trade-off between coverage area. There are multiple local optimal solutions in the solution space, which increases the difficulty of the algorithm to jump out of the local optimum. The simulated annealing algorithm is a heuristic search algorithm that gradually approaches the global optimal solution in large-scale non-convex optimization problems by introducing probability, thereby improving the accuracy and efficiency of solving the battery swap station deployment optimization problem. Secondly, the simulated annealing algorithm has a faster convergence speed and can find a near-optimal battery swap station deployment plan in a shorter time, which helps to shorten the decision-making cycle and improve the efficiency of the deployment of battery swap stations in the area of ​​electric two-wheeled vehicles. Finally, in the process of gradually optimizing the layout of battery swap stations, the synchronous generation of regional traffic is achieved through the trained generation model, thereby improving the practicality of the deployment of battery swap stations in the area of ​​electric two-wheeled vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to make the purpose, technical solution and advantages of the invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings, in which:

[0073] Figure 1 This is a network structure diagram of the battery swap station deployment method in areas without historical data based on electric two-wheeled vehicle traffic.

[0074] Figure 2 Schematic diagram of the battery swap system.

[0075] Figure 3 This is the spatial distribution of electric two-wheeler traffic in a city.

[0076] Figure 4 The temporal distribution of electric two-wheeler traffic at different time scales: (a) is hourly distribution, and (b) is daily distribution.

[0077] Figure 5 The temporal distribution of electric two-wheeler traffic in the office area at different time scales: (a) is hourly distribution, and (b) is daily distribution.

[0078] Figure 6 The temporal distribution of electric two-wheeler traffic in residential areas at different time scales: (a) is hourly distribution, and (b) is daily distribution.

[0079] Figure 7 Architecture diagram when generating a model for training.

[0080] Figure 8 Network structure diagram of the knowledge-enhanced spatiotemporal module. DETAILED DESCRIPTION

[0081] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but only represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0082] The following is a further detailed description through specific implementation methods:

[0083] Example:

[0084] This embodiment discloses a method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles.

[0085] like Figure 1 As shown in the figure, the method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles includes:

[0086] S1: Obtain regional characteristic data and environmental data related to battery swap stations in the city;

[0087] In this embodiment, the city is divided into I non-overlapping areas by the trunk road network.

[0088] Regional feature data includes the number of battery swap stations in the region, the number of battery swap stations in adjacent areas, the number of battery swap stations in neighboring areas, and the density of ten types of POIs in adjacent areas and neighboring areas, including companies, education, entertainment, catering, government departments, residences, hotel accommodation, life services, medical care, and shopping. Environmental data includes road network data and urban knowledge graph data.

[0089] S2: Build and train a generative model based on the analysis results of historical data of deployed battery swap stations combined with denoising diffusion probabilistic models (DDPMs);

[0090] Among them, the trained generation model is used to generate electric two-wheeled vehicle traffic in the target area;

[0091] S3: Construct a battery swap station deployment optimization problem for the target area to maximize the balance between the correlation between the number of battery swap stations and the regional flow of two-wheeled vehicles and the coverage of the battery swap stations;

[0092] S4: Solve the battery swap station deployment optimization problem through simulated annealing algorithm to obtain the optimal battery swap station deployment plan;

[0093] In the process of gradual solution, the trained generation model is called to synchronously generate the flow of electric two-wheeled vehicles according to the regional characteristic data and environmental data of the target area;

[0094] S5: Achieve the deployment of battery swap stations in the target area through the optimal battery swap station deployment plan.

[0095] The present invention is based on the analysis results of the historical data of the deployed battery swap station area combined with the denoising diffusion probability model to construct and train the generation model. The trained generation model can generate the electric two-wheeled vehicle flow according to the regional characteristic data and environmental data of the target area. First, the denoising diffusion probability model generates high-quality data samples by gradually removing noise. In the generation of electric two-wheeled vehicle flow, it can capture the nonlinear relationship between complex regional characteristic data (such as battery swap station distribution, urban environment and user behavior, etc.) and two-wheeled vehicle flow, so as to accurately generate the electric two-wheeled vehicle flow and improve the practicality of the deployment of electric two-wheeled vehicle regional battery swap stations. Secondly, because the generation model learns the mapping from noise to data during the training process, the model can generate reasonable electric two-wheeled vehicle flow when facing new and unseen regional characteristic data and urban data, and can adapt to the dynamic changes of regional characteristic data and urban data. This generalization ability is particularly important for the deployment of battery swap stations in areas without historical data. Finally, the present invention constructs a generation model based on the denoising diffusion probability model and combines the analysis results of the historical data of the deployed battery swap station area. First, the inverse denoising network of the generative model is composed of spatial modules and temporal modules, with the aim of better capturing the complex spatiotemporal dependencies of electric two-wheeled vehicle traffic, thereby taking into account the impact of battery swap station distribution and user behavior on regional traffic. Second, an urban knowledge graph is constructed to simulate the urban environment and represent the complex relationships between entities. The urban knowledge graph is used as the input of the denoising network of the generative model, and the spatiotemporal modules are further fused using the attention mechanism under the guidance of the urban knowledge graph embedding, so as to better capture the impact of the urban environment on regional traffic. Third, the flow estimator uses regional features to estimate the size of the electric two-wheeled vehicle flow in advance, so as to be better suitable for situations where the flow differences between regions are significant.

[0096] The present invention constructs an optimization problem for battery swap station deployment that maximizes the balance between the correlation between the number of battery swap stations and the regional flow of two-wheeled vehicles and the coverage of the battery swap stations. First, by constructing the optimization problem, it can be ensured that the number and location of battery swap stations are closely related to the regional flow of two-wheeled vehicles, which can avoid excessive construction of battery swap stations and waste of resources, thereby improving the rationality of the deployment of battery swap stations in the electric two-wheeled vehicle area. Secondly, the optimization problem also takes into account the coverage of the battery swap stations to ensure that as many two-wheeled vehicle users as possible can easily access the battery swap stations, which helps to improve user satisfaction and the utilization rate of the battery swap stations. Finally, by balancing the number and coverage of battery swap stations, the optimization problem can reduce the construction and operation costs of battery swap stations while meeting user needs.

[0097] In view of the non-convexity of the optimization problem, the present invention solves the optimization problem through a simulated annealing algorithm. First, the battery swap station deployment optimization problem contains multiple complex factors, such as the correlation between regional traffic and the number of battery swap stations and the trade-off between coverage area. There are multiple local optimal solutions in the solution space, which increases the difficulty of the algorithm to jump out of the local optimum. The simulated annealing algorithm is a heuristic search algorithm that gradually approaches the global optimal solution in large-scale non-convex optimization problems by introducing probability, thereby improving the accuracy and efficiency of solving the battery swap station deployment optimization problem. Secondly, the simulated annealing algorithm has a faster convergence speed and can find a near-optimal battery swap station deployment plan in a shorter time, which helps to shorten the decision-making cycle and improve the efficiency of the deployment of battery swap stations in the area of ​​electric two-wheeled vehicles. Finally, in the process of gradually optimizing the layout of battery swap stations, the synchronous generation of regional traffic is achieved through the trained generation model, thereby improving the practicality of the deployment of battery swap stations in the area of ​​electric two-wheeled vehicles.

[0098] In order to better introduce the technical solution of the present invention, this embodiment is described through the following parts.

[0099] 1. Analysis of Electric Two-wheeled Vehicle Traffic

[0100] Typical battery replacement systems include Figure 2 As shown. The battery swap company deploys battery swap stations in the city. The driver arrives at the battery swap station and exchanges the low-power battery in the car with the high-power battery in the battery swap cabinet. The battery management system transmits the relevant data of the battery to the cloud management platform of the battery swap service provider via 4G / 5G. From this, the trajectory data of the battery can be obtained, and after matching it to the area divided by the urban road network, the flow of electric two-wheeled vehicles can be obtained. The data set of this embodiment comes from a battery swap company in a certain city, and the time span is from March 1 to March 31, 2023. The details of the data are as follows.

[0101] Battery status data: Contains more than 2,000 shared batteries, collected through the battery management system on the shared batteries. The data mainly includes the battery ID, data timestamp, battery GPS, etc. The update granularity of GPS data is 15 minutes.

[0102] Battery swap transaction data: describes the battery swap record ID, user ID, transaction time, location of the battery swap station, battery IDs swapped in and out, etc.

[0103] Battery swap station data: including the battery swap station’s ID, specific location, construction timestamp, etc.

[0104] Based on this real data set, this embodiment analyzes the data characteristics of electric two-wheeled vehicle traffic in order to better utilize the generation model in new areas without historical traffic data. The analysis details are as follows.

[0105] 1) The spatial distribution of regional traffic is highly uneven. Figure 3 The average daily flow of electric two-wheeled vehicles in different areas of Chengdu after road network division is visualized. The flow shows obvious aggregation characteristics between different areas, which is manifested as a high concentration of flow in multiple areas, while the flow difference between different areas is significant. Furthermore, this embodiment analyzes which regional characteristics affect the flow of electric two-wheeled vehicles. The characteristics of each area are standardized using the z-score method to ensure consistency between areas. As shown in Table 1, the size of regional flow is closely related to the number and location of battery swap stations, and is also related to the density and type of POIs. These characteristics respectively reflect the attractiveness of the region to users and its overall prosperity.

[0106] Table 1 Characteristics of the top five regions ranked by Pearson correlation coefficient

[0107]

[0108] 2) User-oriented time distribution. The user group of battery swap services is mainly instant delivery personnel who frequently use electric two-wheelers. Therefore, the time distribution of regional traffic is closely related to people’s behavior patterns of ordering takeout. Figure 4 As shown in (a), traffic is concentrated during the day, and there are two peaks a day: 12-13 noon and 18-19 pm. 12-13 noon is the highest peak, and it is also the peak time for people to order takeout. Figure 4 As shown in (b), the weekly time pattern is similar to the daily time pattern, and the daily average traffic distribution shows the same peak period.

[0109] 3) The spatiotemporal dependence of regional traffic is highly complex and varies significantly. On the one hand, the regional traffic of two-wheelers is affected by neighboring areas and shows spatial dependence; on the other hand, the regional traffic with similar urban functions shows similar temporal patterns, which indicates that regional traffic is also significantly affected by the urban environment. Figure 5 (a) Figure 5 (b) represent the daily and weekly traffic patterns in the office area, respectively. Figure 6 (a) Figure 6 (b) shows the daily and weekly traffic patterns in residential areas. In office areas, the ratio of the midday peak to the evening peak is higher, and people tend to order food at noon on weekdays. The number of people ordering takeouts decreases from Friday onwards, and the decrease is more obvious on weekends. Compared with office areas, the proportion of people ordering takeouts in the morning and evening in residential areas increases. At the same time, Saturday noon is the peak of takeout orders in the week.

[0110] 2. Electric Two-wheeler Traffic Generation Model

[0111] In order to better utilize the denoising diffusion probability model, this embodiment constructs a generation model based on the above analysis results of the historical data of the deployed battery swap station area. The denoising network structure design of the generation model is closely combined with the characteristics of the data to be generated. The overall architecture of the model is as follows Figure 7 The detailed introduction of the model is as follows.

[0112] 1) Denoising Diffusion Probability Model. The electric two-wheeler traffic generation model is based on the denoising diffusion probability model. As one of the most advanced generative models, it consists of two interrelated processes: a predefined forward process that maps the original data x0 to a simpler prior distribution (standard normal distribution) by adding noise ε; and a denoising inference reverse process that uses the denoising network to predict the noise Reverse the effects of the forward process and recover the data from a standard normal distribution.

[0113] In the forward denoising process, the diffusion model is divided into multiple time steps, and the initial data is linearly combined with randomly sampled Gaussian noise. In the original denoising diffusion probability model, for the original data x0, the data after t-th step of denoising is x t , the noise adding process is expressed as Where: ε t-1 Indicates sampling from a standard normal distribution, i.e., ε~N(0,I), where I represents the unit matrix; β t Represents the variance of the additional noise. Generally speaking, there is a scheduling method to define the value of β in advance.

[0114] Defining β t =1-α t , then the noise adding process can be rewritten as in

[0115] Generated completely noisy data x T It conforms to the complete standard normal distribution, that is, x T ~N(0,I).

[0116] In order to make the generated traffic more accurate, the original denoising diffusion probability model is modified so that the completely noise data generated in each area conforms to the predicted regional traffic output by the traffic estimator. Normal distribution, that is, noisy data

[0117] In the backward denoising process, x0 is restored from the complete standard normal distribution. Considering q(x t-1 |x t ) is difficult to solve directly, this embodiment uses a neural network to learn this distribution, that is, p θ (x t-1 |x t ). Finally, this process can be expressed as In fact, it can be regarded as learning noise ε θ (x t ,t). The core of training is to construct the loss function Minimize the error between the model prediction noise and the actual noise, specifically: During actual sampling, denoising is performed step by step starting from t=T, and x0 is finally generated.

[0118] 2) Traffic estimator based on regional features. The basic denoising diffusion probability model uniformly maps the data distribution to the standard normal distribution. However, this does not apply to situations where the traffic volume varies significantly between regions. Therefore, a traffic estimator is used to estimate the size of the electric two-wheeled vehicle traffic volume in advance using regional features. The estimated result is used as the mean of the Gaussian distribution to be mapped in the forward process to distinguish different regions. Specifically, the data formed in the last step of noise addition obeys the distribution x T ~N(p,I), where p is the preset electric vehicle area flow output by the trained flow estimator for different area features.

[0119] The traffic estimator consists of a two-layer feedforward neural network, the activation function uses the Leaky ReLU function, and the loss function is The output of the flow estimator is the predicted regional flow Combine regional characteristic data with predicted regional flow After being processed by the fully connected layer, they are concatenated and the concatenated features are input into the denoising network as conditions.

[0120] 3) Construction of urban knowledge graph. In order to better capture the complex spatiotemporal dependencies of electric two-wheeled vehicle traffic, an urban knowledge graph is constructed to model the urban environment and capture the complex relationships between entities. The urban knowledge graph consists of five facts. A fact is a triple containing a head entity, a relationship, and a tail entity. The five facts are as follows:

[0121] <region BorderBy region> , indicating the adjacency relationship between regions;

[0122] <region NearBy region> , indicating the proximity relationship between regions;

[0123] <region SimilarFunc region> , indicating the functional similarity relationship between regions;

[0124] <POI LocateAt region> , indicating the area to which the POI belongs;

[0125] <POI CateOf category> , indicating the category to which the POI belongs;

[0126] The city knowledge graph represents the adjacency, proximity, and functional area similarity relationships between regions, and indicates the region and category to which the POI belongs.

[0127] 4) Knowledge-enhanced denoising network. The core of the denoising probability model is the structure of the inverse denoising network. The denoising network is built based on the DiffWave model, and the bidirectional dilated convolution block in the DiffWave model is replaced by a knowledge-enhanced spatiotemporal module;

[0128] like Figure 8 As shown in the figure, the knowledge-enhanced spatiotemporal module includes a spatial module and a temporal module; the spatial module adopts a 1-layer R-GCN, and the temporal module adopts a Transformer layer, and the outputs of the spatial module and the temporal module are fused using the attention mechanism; the urban knowledge graph is converted into an urban knowledge graph through TuckER and embedded as the query of the attention mechanism to guide the fusion process.

[0129] In this embodiment, the knowledge-enhanced spatiotemporal module can better capture the spatiotemporal dependencies of data in this scenario, thereby generating electric two-wheeled vehicle traffic.

[0130] In this embodiment, the DiffWave model is based on a diffusion probability model and uses the fixed steps of the Markov chain to convert the white noise signal into a structured waveform. The DiffWave network model mainly includes the following modules:

[0131] Initial noise sampling module: Sample the initial noise from the standard Gaussian distribution as the starting point for waveform generation.

[0132] Denoising module: Through a series of denoising steps, the noise is gradually converted to the target audio. Each denoising step is completed by a neural network model, which receives the noise signal and time step information at the current moment as input and outputs the audio signal estimate at the corresponding moment to guide the next denoising step.

[0133] Bidirectional dilated convolution module: This is a unique module in the DiffWave model. Different from the WaveNet architecture, it avoids the unidirectional error and error accumulation problems of the autoregressive model. The bidirectional dilated convolution module enables the network to process input signals in parallel, increasing the generation speed.

[0134] Local Conditioner Module (optional): In a speech synthesis system, a neural vocoder can synthesize a corresponding waveform based on local condition information (such as aligned semantic features or spectrograms). DiffWave supports synthesizing the corresponding sound based on the Mel-spectrogram, which is implemented through the local conditioner module.

[0135] Combination Figure 7 As shown in Figure 2, the processing steps for training the generative model are as follows:

[0136] S201: Acquire original data x0 as training data, regional feature data and environmental data of deployed battery swap stations;

[0137] S202: inputting the regional characteristic data and the environmental data into the flow estimator to perform flow estimation, and outputting the predicted regional flow; constructing a loss function through the predicted regional flow and the actual regional flow, and optimizing the flow estimator model parameters until the loss function value tends to be stable;

[0138] S203: forward diffusion is performed by adding randomly sampled noise ε to the original data x0 to obtain noise data;

[0139] The noise data is consistent with the predicted regional flow output by the flow estimator. Normal distribution of

[0140] S204: Input the constructed urban knowledge graph and the sampled noise data into the denoising network for denoising and restoration, and predict the noise of the corresponding sampling step.

[0141] S205: Noise through prediction Construct the loss function with noise ε and reversely optimize the model parameters of the denoising network;

[0142] S206: Repeat steps S203 to S205 until the loss function value becomes stable.

[0143] Specifically, after the generation model is trained, the electric two-wheeled vehicle traffic is generated through the following steps:

[0144] S211: Input the regional feature data of the target area into the flow estimator to perform flow estimation, and output the predicted regional flow

[0145] S212: Based on the predicted regional traffic From obedience Sampling noise data from a normal distribution;

[0146] S213: Input the sampled noise data into the trained denoising network for denoising and restoration, and generate denoised data as the electric two-wheeled vehicle flow.

[0147] The present invention constructs a generative model based on the denoising diffusion probability model and the analysis results of historical data in areas where battery swap stations have been deployed. First, the inverse denoising network of the generative model is composed of a spatial module and a temporal module, with the aim of better capturing the complex spatiotemporal dependencies of electric two-wheeled vehicle traffic, thereby taking into account the impact of battery swap station distribution and user behavior on regional traffic. Second, an urban knowledge graph is constructed to simulate the urban environment and represent the complex relationships between entities. The urban knowledge graph is used as the input of the denoising network of the generative model, and the spatiotemporal modules are further fused using the attention mechanism under the guidance of the embedding of the urban knowledge graph, so as to better capture the impact of the urban environment on regional traffic. Third, the flow estimator uses regional features to estimate the size of the electric two-wheeled vehicle flow in advance, so as to be better suitable for situations where the flow differences between regions are significant.

[0148] 3. Modeling of battery swap station deployment optimization problem

[0149] After the trained electric two-wheeled vehicle flow generation model is obtained, the model is used for optimizing the actual deployment of battery swap stations. To this end, this embodiment establishes an optimization model for battery swap station deployment, aiming to comprehensively maximize the correlation between the number of battery swap stations and regional flow and the coverage of battery swap stations.

[0150] 1. Define urban areas. The city is divided into I non-overlapping areas by the main road network. Furthermore, R = {r1, r2, ..., r I} represents the set of all areas of the city, R TR ={r1,r2,...,r m}∈R represents the target area where m battery swap stations are to be deployed, R SR = {r m+1 ,r m+2 ,...,r I}∈R represents the remaining source areas where battery swap stations have been deployed.

[0151] 2. Define the flow of electric two-wheeled vehicles. i =[f i 1 ,f i 2 ,...f i T ] represents the traffic in the ith area. Among them, f i t represents the flow of electric two-wheeled vehicles in the tth time interval of a day (i.e., the average of inflow and outflow), reflecting the user activity level in the area during that period. For convenience, F SR Represents R SR Aggregation of electric two-wheeler traffic in the source area.

[0152] 3. Define the battery swap station. Assume that the number of battery swap stations waiting to be deployed is J, and the tuple w j =(g j ,e j ) represents the jth battery swap station.

[0153] Among them, g j ∈R represents the location of the battery swap station, e j represents the deployment cost of the battery swap station. In particular, e j It consists of three parts:

[0154] e j = l j +bss j +m j ;

[0155] Among them, l j is the land cost, bss j is the cost of purchasing battery swap stations and batteries, m j For later maintenance costs.

[0156] 4. Define the deployment plan of the battery swap station. Matrix X = [x ij ] I×J represents the deployment plan of the battery swap station, where the variable x ij represents whether the jth battery swap station is deployed in the ith area. ij =1 represents deployment, x ij = 0 means not deployed. For convenience, X TR Represents the target area R TR Deployment plan, X SR Represents the source region R SR The actual deployment plan.

[0157] 5. Define the generation of electric two-wheeled vehicle traffic. f(·) represents the generation model. For region rj ∈R TR , traffic generation can be expressed as:

[0158]

[0159] Among them, D SR and D TR They respectively represent the environmental information and regional characteristics of their respective areas.

[0160] 6. Define the battery swap station deployment optimization problem. This problem aims to comprehensively realize the correlation between the number of battery swap stations and regional traffic and maximize the coverage of battery swap station services. Among them, the correlation between the number of battery swap stations and regional traffic is defined as:

[0161] r=corr(Flow TR , X TR );

[0162] Flow TR =[Flow1, Flow2,...,Flow m ] represents the flow vector of the target area. Further, the flow size of the i-th area is defined as:

[0163]

[0164] u t is the influencing factor of the regional flow of electric vehicles in different time periods, which is obtained through the user's battery replacement habits to indicate the importance of regional flow in each time period. t The closer the value of is to 1, the more likely the user is to swap batteries during this period, and the greater the impact on regional traffic during this period. The calculation method for the service coverage of the battery swap station is as follows: j If there is a battery swap station, then take r j The center point position c j , set the service radius r, for all areas where battery swap stations are deployed, calculate the union of the battery swap station network service coverage area and express it as CoV total .

[0165] Therefore, the optimization problem of battery swap station deployment is expressed as:

[0166] Objective function:

[0167] Cost constraints:

[0168] Constraints on the number of battery swap stations:

[0169] in:

[0170] r=corr(Flow TR , XTR );

[0171]

[0172] e j = l j +bss j +m j ;

[0173] Where: α represents the variable used to adjust the importance of the correlation and coverage of the battery swap station service; r represents the correlation between the number of battery swap stations and regional traffic; Cov total represents the union of the service coverage of the battery swap station (in order to eliminate the influence of different dimensions, cov total The calculation method of the service coverage of the battery swap station is: if the area r j If there is a battery swap station, then take r j The center point position c j Set the service radius r; e j represents the deployment cost of the battery swap station; B represents the total cost constraint; C represents the upper limit of the number of battery swap stations deployed in each area to avoid unrealistic allocation of the number of battery swap stations; x ij Indicates whether the jth battery swap station is deployed in the i-th area. ij =1 represents deployment, x ij =0 means not deployed; Flow TR Indicates the flow of electric two-wheeled vehicles in the target area; Flow i represents the flow of electric two-wheeled vehicles in the i-th area of ​​the target area; represents the flow of electric two-wheeled vehicles in the ith area during the tth period; u t It represents the influencing factor of the regional flow of electric vehicles in different time periods, which is obtained through the user's battery replacement habits to indicate the importance of regional flow in each time period, u t The closer the value is to 1, the more likely the user is to swap batteries during this period, and the greater the impact on regional traffic during this period; j Indicates land cost; bss j represents the cost of purchasing a battery swap station and batteries; m j Indicates the subsequent maintenance cost.

[0174] 4. Simulated Annealing Optimization Algorithm

[0175] In this embodiment, a simulated annealing algorithm is used to solve the non-convex optimization problem of battery swap station deployment optimization, and the synchronous generation of electric two-wheeled vehicle traffic is achieved in the process of gradually approaching the global optimal solution.

[0176] Since the battery swap station site selection problem is a complex non-convex optimization problem, and is subject to multiple constraints such as deployment cost and regional upper limit, it is difficult to find a global optimal solution within polynomial time. In order to effectively solve this problem, this embodiment adopts an optimization algorithm based on simulated annealing. In order to achieve the synchronous generation of electric two-wheeled vehicle traffic, for each possible battery swap station deployment plan, the regional characteristics are calculated and the regional feature vector is generated, and input into the traffic estimator of the generation model. Specifically, the regional characteristics are changed: 1) The number of battery swap stations in the area 2) The number of battery swap stations in adjacent areas 3) The number of battery swap stations in neighboring areas. At this time, the generation model generates regional traffic for the battery swap station deployment plan, and then calculates the plan. In the iterative process of simulated annealing, the model is continuously called to generate the flow of electric two-wheeled vehicles and calculate To obtain the optimal battery swap station deployment plan.

[0177] Specifically, the processing steps of the simulated annealing optimization algorithm include:

[0178] S401: Determine the initial solution S0, initial temperature T0, cooling coefficient ε, total cost constraint B, upper limit C of regional deployment of battery swap stations, maximum number of iterations N, and minimum temperature T of the simulated annealing optimization algorithm min ;

[0179] S402: Initialization: Current solution S current ←S0, optimal solution S * ←S0, temperature T←T0, number of iterations k←0;

[0180] S403: From the current solution current The neighborhood generates a new solution S new , that is, change the current solution S current Generate new solution S from partial regional feature data in new ; The solution refers to the deployment plan of battery swap stations in the target area;

[0181] In this embodiment, the following regional characteristics are changed: 1) the number of battery swap stations in the region; 2) the number of battery swap stations in adjacent regions; 3) the number of battery swap stations in neighboring regions.

[0182] S404: When the new solution S new When the cost constraint of the battery swap station deployment optimization problem is met: call the generation model to generate the current solution S current And new solution S new The electric two-wheeled vehicle flow is then brought into the objective function of the battery swap station deployment optimization problem to calculate the current solution S current The objective function value of And the new solution S new The objective function value of

[0183] S405: Determine whether the If so, accept the new solution S current ←S new ; Otherwise, execute step S406;

[0184] S406: Determine whether to accept the new solution based on the Metropolis acceptance criteria. Accept the new solution: If the random number is satisfied Then accept the new solution S current ←S new ; Otherwise, the new solution will not be accepted;

[0185] S407: Call the generation model to generate the electric two-wheeled vehicle flow of the optimal solution S*, and then bring it into the objective function of the battery swap station deployment optimization problem to calculate the objective function value of the optimal solution S* When satisfied When , update the optimal solution S * ←S current ;

[0186] S408: Update temperature T←ε·T, iteration count k←k+1;

[0187] S409: Determine whether T>T min Or k<N: If so, return to step S403 and execute the next iteration; otherwise, end the iteration and output the optimal battery swap station deployment plan corresponding to the optimal solution S*.

[0188] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit the technical solution. Those skilled in the art should understand that those modifications or equivalent substitutions of the technical solution of the present invention that do not depart from the purpose and scope of the technical solution should be included in the scope of the claims of the present invention.

Claims

1. A method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles, characterized in that: include: S1: Obtain regional characteristic data and environmental data related to battery swap stations in the city; S2: Build and train a generative model based on the analysis results of historical data of deployed battery swap stations combined with a denoised diffusion probability model; Among them, the trained generation model is used to generate electric two-wheeled vehicle traffic in the target area; S3: Construct a battery swap station deployment optimization problem for the target area to maximize the balance between the correlation between the number of battery swap stations and the regional flow of two-wheeled vehicles and the coverage of the battery swap stations; S4: Solve the battery swap station deployment optimization problem through simulated annealing algorithm to obtain the optimal battery swap station deployment plan; In the process of gradual solution, the trained generation model is called to synchronously generate the flow of electric two-wheeled vehicles according to the regional characteristic data and environmental data of the target area; S5: Achieve the deployment of battery swap stations in the target area through the optimal battery swap station deployment plan.

2. The method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles as claimed in claim 1, characterized in that: In step S2, the processing steps for training the generated model are as follows: S201: Acquire original data x0 as training data, regional feature data and environmental data of deployed battery swap stations; S202: inputting the regional characteristic data and the environmental data into the flow estimator to perform flow estimation, and outputting the predicted regional flow; constructing a loss function through the predicted regional flow and the actual regional flow, and optimizing the flow estimator model parameters until the loss function value tends to be stable; S203: forward diffusion is performed by adding randomly sampled noise ε to the original data x0 to obtain noise data; The noise data is consistent with the predicted regional flow output by the flow estimator. Normal distribution of S204: Input the constructed urban knowledge graph and the sampled noise data into the denoising network for denoising and restoration, and predict the noise of the corresponding sampling step. S205: Noise through prediction Construct the loss function with noise ε and reversely optimize the model parameters of the denoising network; S206: Repeat steps S203 to S205 until the loss function value becomes stable.

3. The method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles as claimed in claim 2, characterized in that: In step S202, the flow estimator is composed of a two-layer feedforward neural network, the activation function uses the Leaky ReLU function, and the loss function is The output of the flow estimator is the predicted regional flow Combine regional characteristic data with predicted regional flow After being processed by the fully connected layer, they are concatenated and the concatenated features are input into the denoising network as conditions.

4. The method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles as claimed in claim 3, characterized in that: In step S203, the steps of forward diffusion are as follows: In the original denoising diffusion probability model, for the original data x0, the data after adding noise in the tth step is x t , the noise adding process is expressed as Where: ε t-1 Indicates sampling from a standard normal distribution, i.e., ε~N(0,I); β t represents the variance of the additive noise; Defining β t =1-α t , then the noise adding process can be rewritten as in Generated noise data x T It conforms to the complete standard normal distribution, that is, x T ~N(0,I). Modify the original denoising diffusion probability model so that the noise data generated in each area conforms to the predicted regional flow output by the flow estimator. Normal distribution, that is, noisy data 5. The method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles as claimed in claim 2, characterized in that: In step S204, the city knowledge graph is used to model the urban environment, including the following five facts: <region BorderBy region> , indicating the adjacency relationship between regions; <region NearBy region> , indicating the proximity relationship between regions; <region SimilarFunc region> , indicating the functional similarity relationship between regions; <POI LocateAt region> , indicating the area to which the POI belongs; <POI CateOf category> , indicating the category to which the POI belongs; Among them, a fact is a triple, which contains a head entity, a relation and a tail entity.

6. The method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles as claimed in claim 2, characterized in that: In step S204, the denoising network is constructed based on the DiffWave model, and the bidirectional dilated convolution block in the DiffWave model is replaced by a knowledge-enhanced spatiotemporal module; The knowledge-enhanced spatiotemporal module includes a spatial module and a temporal module. The spatial module adopts a 1-layer R-GCN, and the temporal module adopts a Transformer layer. The outputs of the spatial module and the temporal module are fused using the attention mechanism. The urban knowledge graph is converted into an urban knowledge graph through TuckER and embedded as the query of the attention mechanism to guide the fusion process.

7. The method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles as claimed in claim 2, characterized in that: In step S205, the calculation formula of the loss function is expressed as:

8. The method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles as claimed in claim 2, characterized in that: In step S2, after the generation model is trained, the electric two-wheeled vehicle flow is generated through the following steps: S211: Input the regional feature data of the target area into the flow estimator to perform flow estimation, and output the predicted regional flow S212: Based on the predicted regional traffic From obedience Sampling noise data from a normal distribution; S213: Input the sampled noise data into the trained denoising network for denoising and restoration, and generate denoised data as the electric two-wheeled vehicle flow.

9. The method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles as claimed in claim 1, characterized in that: In step S3, the battery swap station deployment optimization problem is expressed as: Objective function: Cost constraints: Constraints on the number of battery swap stations: in: r=corr(Flow TR ,X TR ); Flow TR =[Flow1,Flow2,...,Flow m ], e j =l j +bss j +m j ; Where: α represents the variable used to adjust the importance of the correlation and coverage of the battery swap station service; r represents the correlation between the number of battery swap stations and regional traffic; cov total represents the union of the service coverage of the battery swap station. The service coverage of the battery swap station is calculated as follows: if area r j If there is a battery swap station, then take r j The center point position c j Set the service radius r; e j represents the deployment cost of the battery swap station; B represents the total cost constraint; C represents the upper limit of the number of battery swap stations deployed in each area to avoid unrealistic allocation of the number of battery swap stations; x ij Indicates whether the jth battery swap station is deployed in the i-th area. ij =1 represents deployment, x ij =0 means not deployed; Flow TR Indicates the flow of electric two-wheeled vehicles in the target area; Flow i represents the flow of electric two-wheeled vehicles in the i-th area of ​​the target area; represents the flow of electric two-wheeled vehicles in the ith area during the tth period; u t It represents the influencing factor of the regional flow of electric vehicles in different time periods, which is obtained through the user's battery replacement habits to indicate the importance of regional flow in each time period, u t The closer the value is to 1, the more likely the user is to swap batteries during this period, and the greater the impact on regional traffic during this period; j Indicates land cost; bss j represents the cost of purchasing a battery swap station and batteries; m j Indicates the subsequent maintenance cost.

10. The method for deploying battery swap stations in areas without historical data based on the flow of electric two-wheeled vehicles as claimed in claim 9, characterized in that: In step S4, the processing steps of the simulated annealing optimization algorithm include: S401: Determine the initial solution S0, initial temperature T0, cooling coefficient ε, total cost constraint B, upper limit C of regional deployment of battery swap stations, maximum number of iterations N, and minimum temperature T of the simulated annealing optimization algorithm. min ; S402: Initialization: Current solution S current ←S0, optimal solution S * ←S0, temperature T←T0, number of iterations k←0; S403: From the current solution current The neighborhood generates a new solution S new , that is, change the current solution S current Generate new solution S from partial regional feature data in new ; The solution refers to the deployment plan of battery swap stations in the target area; S404: When the new solution S new When the cost constraint of the battery swap station deployment optimization problem is met: call the generation model to generate the current solution S current And new solution S new The electric two-wheeled vehicle flow is then brought into the objective function of the battery swap station deployment optimization problem to calculate the current solution S current The objective function value of And the new solution S new The objective function value of S405: Determine whether the If so, accept the new solution S current ←S new ; Otherwise, execute step S406; S406: If the random number is satisfied Then accept the new solution S current ←S new ; Otherwise, the new solution will not be accepted; S407: Calling the generation model to generate the optimal solution S * The electric two-wheeled vehicle flow is then brought into the objective function of the battery swap station deployment optimization problem to calculate the optimal solution S * The objective function value of When satisfied When , update the optimal solution S * ←S current ; S408: Update temperature T←ε·T, iteration count k←k+1; S409: Determine whether T > T is satisfied min or k < N: If so, return to step S403 to perform the next iteration; otherwise, end the iteration and output the optimal solution S * The corresponding optimal deployment plan for battery swapping stations

Images