Order dispatching method, device, computer equipment and storage medium for power exchange platform

Abstract

The present invention relates to a dispatching method, device, computer equipment and storage medium for a battery swapping platform. The method constructs a battery swapping load and demand forecasting model based on real-time collected data on the use of power from battery swapping stations and electric vehicles, and accurately predicts the load changes of battery swapping stations and regional battery swapping needs in future periods. It uses multi-objective optimization technology to create an optimized allocation model for battery swapping orders, comprehensively considers current loads and future demand forecasts, and realizes intelligent matching of orders with battery swapping station resources. When a battery swapping request is received, the model quickly optimizes the processing, allocates a suitable battery swapping station and generates a dispatching instruction, which improves the response speed, alleviates inventory pressure, realizes balanced resource utilization, and optimizes user experience and platform operation efficiency. The implementation of the present invention breaks through the technical bottleneck of traditional battery swapping platforms and lays a solid foundation for the intelligent upgrade of electric travel support systems.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence and big data analysis technology, and in particular to a method, device, computer equipment and storage medium for dispatching orders on a battery swap platform. Background Art

[0002] With the popularity of electric bicycles and the continuous growth of user demand, the battery swap platform has become increasingly important as an important infrastructure to support electric transportation. However, the existing battery swap platform faces many technical bottlenecks in the dispatching process. First, the accuracy and timeliness of real-time data collection and updating are insufficient, making it difficult for battery swap stations to grasp the dynamic changes of battery inventory, vehicle location and user demand in real time. Traditional methods rely on manual recording and updating of data, resulting in data lag and inaccuracy, which seriously affects the overall operational efficiency of the battery swap platform.

[0003] In addition, existing technologies have limited capabilities in data preprocessing and analysis. Many platforms lack efficient data cleaning and analysis mechanisms, resulting in uneven data quality. The prediction of battery swap demand and load relies solely on simple statistical models, which is difficult to cope with complex and changing actual situations. The inaccuracy of the prediction results directly affects the rationality of scheduling decisions, resulting in unbalanced loads at battery swap stations, chaotic battery inventory management, and poor user experience.

[0004] Finally, the existing dispatching algorithms generally lack intelligence and flexibility, and are unable to dynamically adjust dispatching strategies based on real-time data, making it impossible for battery swap stations to efficiently respond to sudden changes in demand. At the same time, the lack of effective real-time monitoring and feedback mechanisms makes it difficult for the platform to promptly discover and resolve problems in operation, affecting the smoothness and efficiency of the battery swap process. These problems have severely restricted the service capabilities and user satisfaction of the battery swap platform, and an innovative technical solution is urgently needed to solve them. Summary of the invention

[0005] The main purpose of the present invention is to provide a dispatching method, device, computer equipment and storage medium for a battery swap platform, so as to solve the technical problems faced by existing battery swap platforms in the dispatching process, such as untimely real-time data collection and updating, insufficient data preprocessing and analysis capabilities, and lack of intelligent and flexible scheduling algorithms.

[0006] To achieve the above-mentioned objectives, the present invention provides a method for dispatching orders for a battery swap platform, comprising the following steps: collecting first electric energy situation data that at least describes the historical power usage and real-time power usage of a battery swap station and second electric energy situation data that at least describes the historical power usage and real-time power usage of an electric vehicle; constructing a battery swap load prediction model based on the first electric energy situation data of the battery swap station and the second electric energy situation data of the electric vehicle, wherein the battery swap load prediction model is used to predict the load changes of each battery swap station in a future time period; constructing a battery swap demand prediction model based on the first electric energy situation data of the battery swap station and the second electric energy situation data of the electric vehicle, wherein In the method, the battery swap demand prediction model is used to predict the demand for battery swap stations in different time periods and different areas; a multi-objective optimization method is used to construct a battery swap order optimization allocation model based on the battery swap demand prediction model and the battery swap load prediction model, wherein the battery swap order optimization allocation model is used to allocate battery swap stations to battery swap orders according to the load conditions and usage requirements of each battery swap station; in response to a battery swap request order sent by a client, the battery swap order optimization allocation model is used to optimize the allocation processing of the battery swap request order, determine the battery swap station corresponding to each battery swap request order, and generate a battery swap delivery order based on the battery swap station corresponding to the battery swap request order.

[0007] Furthermore, the first power status data includes at least any of the following: battery inventory information, battery swap station load information, battery swap station location and coverage, environmental factor data, equipment status data, historical battery swap data and user reservation data; the second power status data includes at least any of the following: battery status information, real-time location and speed data, user driving habit data, vehicle model information, environmental factor data, vehicle historical battery swap information and user reservation data.

[0008] Furthermore, before constructing a battery swap load prediction model based on the first power condition data of the battery swap station and the second power condition data of the electric vehicle, the dispatching method also includes: cleaning and deduplicating the first power condition data and the second power condition data to obtain the cleaned and deduplicated first power condition data and the cleaned and deduplicated second power condition data; and, moving and translating the variable data in the cleaned and deduplicated first power condition data and the cleaned and deduplicated second power condition data to obtain a smoothed variable data trend line, and determining the mutation value in the variable data based on the variable data trend line; and smoothing the mutation value in the first power condition data and the second power condition data.

[0009] Furthermore, based on the first power condition data of the battery swap station and the second power condition data of the electric vehicle, a battery swap load prediction model is constructed, including: performing feature engineering processing on the first power condition data based on the battery swap station and the second power condition data of the electric vehicle to extract key features; and constructing composite features based on the key features; constructing an initial battery swap load prediction model based on a multivariate time series, wherein the initial battery swap load prediction model based on a multivariate time series is used to capture the time dependence of the load data, and the input of the initial battery swap load prediction model based on the multivariate time series includes multiple key feature data and multiple composite feature data, and the output includes The load forecast values ​​for each future time period are included; based on the key features and the composite features, the initial model for battery swap load prediction based on multivariate time series is trained, and the L2 regularization technology is introduced to optimize the initial model for battery swap load prediction based on multivariate time series to prevent the initial model for battery swap load prediction based on multivariate time series from overfitting during the training process; the hyperparameters of the initial model for battery swap load prediction based on multivariate time series are automatically adjusted using the Bayesian optimization method to optimize the prediction accuracy of the initial model for battery swap load prediction based on multivariate time series, and obtain a basic model for battery swap load prediction based on multivariate time series;

[0010] Based on the key features and the composite features, an influencing factor matrix is ​​constructed, wherein the influencing factor matrix contains multiple external factor data that affect the battery swapping load; the influencing factor matrix is ​​fitted into the basic model for battery swapping load prediction based on multivariate time series, and the basic model for battery swapping load prediction based on multivariate time series is made to fit and correct the prediction results based on the external factor data to obtain the final battery swapping load prediction model based on multivariate time series.

[0011] Furthermore, based on the first power situation data of the battery swap station and the second power situation data of the electric vehicle, a battery swap demand prediction model is constructed, including: performing feature engineering processing on the first power situation data based on the battery swap station and the second power situation data of the electric vehicle to extract key features; and constructing composite features based on the key features; constructing an initial battery swap demand prediction model based on a long short-term memory network, wherein the initial battery swap demand prediction model based on a long short-term memory network is used to capture the time dependency and long-term time relationship of the battery swap demand data, and the input layer of the initial battery swap demand prediction model based on the long short-term memory network is used to obtain multiple key feature data and multiple composite feature data, the LSTM layer is used to perform multi-layer stacking processing on the data input by the input layer, and the fully connected layer is used to output the predicted value of the electric vehicle battery swap demand in the future time period; based on the key features and the composite features, the initial battery swap demand prediction model based on the long short-term memory network is iterated. The invention discloses a method for iterative training of the initial model of battery swap demand prediction based on the long short-term memory network, and a method for iterative training of the initial model of battery swap demand prediction based on the long short-term memory network. The method uses a back-propagation algorithm to calculate the gradient value of the loss function of the initial model of battery swap demand prediction based on the long short-term memory network relative to the loss of the parameters of the initial model of battery swap demand prediction based on the long short-term memory network, and uses the Adam optimization algorithm to update the parameters of the initial model of battery swap demand prediction based on the long short-term memory network according to the gradient value, so as to accelerate the convergence of the initial model of battery swap demand prediction based on the long short-term memory network; and, in the process of iteratively training the initial model of battery swap demand prediction based on the long short-term memory network based on the key features and the composite features, a Dropout layer is added after each LSTM layer of the initial model of battery swap demand prediction based on the long short-term memory network, and a preset proportion of neurons are randomly discarded to prevent the initial model of battery swap demand prediction based on the long short-term memory network from overfitting the data training of the key features and the composite features, so as to obtain the final model of battery swap demand prediction based on the long short-term memory network.

[0012] Furthermore, a multi-objective optimization method is adopted to construct a battery swap order optimization allocation model based on the battery swap demand prediction model and the battery swap load prediction model, including: generating a load balancing objective function and an inventory balancing objective function, and combining the battery swap demand prediction model and the battery swap load prediction model to construct an initial battery swap order optimization allocation model based on a multi-objective optimization algorithm.

[0013] Wherein, the load balancing objective function is:

[0014] ,

[0015] For the The load of a battery swap station, For the The capacity of each battery swap station;

[0016] Among them, the inventory balance objective function is:

[0017] ,

[0018] For the The battery inventory of each battery swap station, For the The battery swapping demand of each battery swapping station;

[0019] The genetic algorithm and the particle swarm optimization algorithm are integrated into the solution module of the initial model of the battery swap order optimization allocation based on the multi-objective optimization algorithm to obtain the final battery swap order optimization allocation model based on the multi-objective optimization algorithm, wherein the final battery swap order optimization allocation model based on the multi-objective optimization algorithm adopts the genetic algorithm and the particle swarm optimization algorithm to solve the load balancing objective function and the inventory balance objective function.

[0020] Furthermore, in response to the battery swap request order sent by the client, the battery swap order optimization allocation model is used to optimize the allocation processing of the battery swap request order, determine the battery swap station corresponding to each battery swap request order, and generate a battery swap delivery order based on the battery swap station corresponding to the battery swap request order, including: in response to the battery swap request order sent by the client, the battery swap order optimization allocation model is used to generate a set of initial solutions for the battery swap request order, wherein each solution in the set of initial solutions represents a battery swap station corresponding to the battery swap request order, and a delivery route based on the corresponding battery swap station; the fitness of each initial solution is calculated according to the objective function in the battery swap order optimization allocation model, and a roulette wheel method is used to select multiple initial solutions whose fitness is higher than a first preset threshold. , perform individual reproduction processing on the selected multiple initial solutions to obtain new initial solutions; re-execute the step of "calculating the fitness of each initial solution according to the objective function in the battery swap order optimization allocation model" on the new initial solution until the maximum number of iterations is reached or the fitness improvement is less than the second preset threshold; organize the retained initial solutions into a group of initial particles, and initialize the speed and position of each initial particle, and calculate the speed and position of each initial particle according to the objective function in the battery swap order optimization allocation model to obtain the fitness of each initial solution, wherein each initial solution corresponds to an initial particle; use the first formula and the second formula to calculate the optimal position of each particle itself and the optimal position of the group to update the speed and position of the particle,

[0021] Among them, the first formula is:

[0022] ,

[0023] is the particle velocity, is the particle position, is the preset optimal position of the particle itself, is the preset optimal position of the group, is the inertia weight, and is the acceleration constant, and is a random number;

[0024] Wherein, the second formula is:

[0025] ,

[0026] The step of "calculating and processing the speed and position of each initial particle according to the objective function in the battery swap order optimization allocation model to obtain the fitness of each initial solution" is re-executed for each particle until the maximum number of iterations is reached or the fitness improvement is less than a second preset threshold; the particle with the highest fitness is adopted as the optimal node of the battery swap request order, and the battery swap station corresponding to the battery swap request order is determined according to the particle with the highest fitness, and a battery swap delivery order is generated based on the battery swap station corresponding to the battery swap request order and the delivery route of the battery swap station.

[0027] The present invention also provides a dispatching device for a battery swap platform, comprising: a collection unit, used to collect first power situation data describing at least the historical power usage and real-time power usage of a battery swap station and second power situation data describing at least the historical power usage and real-time power usage of an electric vehicle; a first prediction unit, used to construct a battery swap load prediction model based on the first power situation data of the battery swap station and the second power situation data of the electric vehicle, wherein the battery swap load prediction model is used to predict the load changes of each battery swap station in a future time period; a second prediction unit, used to construct a battery swap demand prediction model based on the first power situation data of the battery swap station and the second power situation data of the electric vehicle, wherein The battery swap demand prediction model is used to predict the usage demand of battery swap stations in different time periods and different areas; the allocation unit is used to adopt a multi-objective optimization method to construct a battery swap order optimization allocation model based on the battery swap demand prediction model and the battery swap load prediction model, wherein the battery swap order optimization allocation model is used to allocate battery swap stations to battery swap orders according to the load conditions and usage requirements of each battery swap station; the response unit is used to respond to the battery swap request order sent by the client, adopt the battery swap order optimization allocation model to optimize the allocation processing for the battery swap request order, determine the battery swap station corresponding to each battery swap request order, and generate a battery swap delivery order based on the battery swap station corresponding to the battery swap request order.

[0028] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0029] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned methods are implemented.

[0030] In summary, the technical solution provided by the present invention has comprehensively innovated the dispatching mechanism of the traditional battery swap platform through precise and meticulous data collection, building advanced and reliable prediction models, and implementing intelligent order allocation strategies. This solution effectively solves many problems that existed in the past, such as inaccurate data, distorted predictions, and rigid scheduling, and significantly improves the operational efficiency, resource optimization allocation level, and user experience quality of the battery swap service. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a schematic diagram of the steps of a method for dispatching orders for a battery swap platform in one embodiment of the present invention;

[0032] Figure 2 It is a structural block diagram of a dispatching device of a power swap platform in one embodiment of the present invention;

[0033] Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0034] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0036] Reference Figure 1 This embodiment provides a method for dispatching orders on a battery swap platform, comprising the following steps:

[0037] S1, collecting first power situation data that at least describes the historical power usage and real-time power usage of the battery swap station and second power situation data that at least describes the historical power usage and real-time power usage of the electric vehicle.

[0038] S2. Based on the first power condition data of the battery swap station and the second power condition data of the electric vehicle, a battery swap load prediction model is constructed, wherein the battery swap load prediction model is used to predict the load change conditions of each battery swap station in a future time period.

[0039] S3, constructing a battery swap demand prediction model based on the first power status data of the battery swap station and the second power status data of the electric vehicle, wherein the battery swap demand prediction model is used to predict the usage demand of the battery swap station in different time periods and different areas.

[0040] S4, adopting a multi-objective optimization method, based on the battery swap demand prediction model and the battery swap load prediction model, constructing a battery swap order optimization allocation model, wherein the battery swap order optimization allocation model is used to allocate battery swap stations to battery swap orders according to the load conditions and usage requirements of each battery swap station.

[0041] S5, in response to the battery swap request order sent by the client, the battery swap order optimization allocation model is used to optimize the allocation processing for the battery swap request order, the battery swap station corresponding to each battery swap request order is determined, and based on the battery swap station corresponding to the battery swap request order, a battery swap delivery order is generated.

[0042] That is, through accurate data collection and intelligent model prediction, this technical solution realizes the intelligent dispatch of battery swap services, and through a series of innovative strategies, comprehensively improves the operational efficiency and user satisfaction of the battery swap platform. First, in the data collection link, this solution focuses on improving the timeliness and accuracy of data (S1). This includes real-time tracking of battery usage status and inventory levels at battery swap stations, and close attention to the power demand and driving trajectory of electric vehicle users. This is equivalent to building a real-time monitoring system for the battery swap platform, which can capture the changes of various key indicators in real time, so as to accurately grasp the operating status of the battery swap station and user needs. Secondly, in terms of data processing and analysis, this solution uses advanced prediction models (S2 and S3). On the one hand, through the battery swap load prediction model, the future load change trend of each battery swap station can be predicted, providing strong support for the platform resource allocation. On the other hand, the battery swap demand prediction model is used to combine the time and space dimensions to make a detailed prediction of the battery swap demand at a specific time and place, ensuring the accurate delivery of service resources. On this basis, this solution further introduces a multi-objective optimization method (S4), combines the results of load prediction with demand prediction, and constructs a battery swap order optimization allocation model. The model seeks the optimal solution between multiple goals such as meeting customer needs, balancing the load of battery swap stations, and optimizing battery inventory management, so as to realize the intelligent allocation of battery swap orders. Finally, when the user issues a battery swap request (S5), this solution quickly matches the most suitable battery swap station by optimizing the allocation model and automatically generates a battery swap dispatch instruction. This automated process greatly improves the response speed and dispatch efficiency, effectively shortens the user's waiting time, and improves the operational efficiency of the battery swap station and the utilization rate of battery resources. In summary, this technical solution has successfully broken through the development bottleneck of the traditional battery swap platform through innovative measures such as optimizing the data processing process and introducing an intelligent scheduling mechanism. While improving the accuracy and efficiency of dispatching, it effectively balances the workload of the battery swap station, improves battery inventory management, and provides users with a smoother and more convenient battery swap experience. This series of improvements not only enhances the competitiveness of the battery swap platform, but also injects new vitality into the development of electric bicycle battery swap services.

[0043] In one example, the first power situation data includes at least any of the following: battery inventory information, battery swap station load information, battery swap station location and coverage, environmental factor data, equipment status data, historical battery swap data, and user reservation data.

[0044] In this example, "first power situation data" as a comprehensive concept covers multi-dimensional data related to the operation of the battery swap station and the behavior of electric bicycle users, aiming to improve the efficiency and accuracy of the battery swap service with the help of these detailed data information. At least one of the data items involved is explained as follows: First, battery inventory information refers to the number of available batteries currently held by the battery swap station and the specific inventory status of various types of batteries. This data is crucial to ensure that the user's immediate battery swap needs can be met, and can effectively avoid service interruptions caused by insufficient batteries. Secondly, the load information of the battery swap station mainly reflects the current workload status of the battery swap station, including key indicators such as the number of batteries being charged and the number of users waiting for battery swap. Through real-time monitoring of these data, it is possible to clearly understand whether the service capacity of the battery swap station is close to saturation, thereby providing a decision-making basis for resource allocation. Furthermore, the location and coverage data of the battery swap station involve the specific geographical location of the battery swap station and the regional scope it can serve. This data is of great significance for analyzing potential user needs and planning a reasonable distribution network, and helps to improve the coverage and response speed of services. In addition, environmental factor data is also an aspect that cannot be ignored, including external factors such as weather conditions, holiday arrangements, special events, etc. that may affect users' travel choices and battery replacement needs. For example, bad weather may increase the frequency of use of electric bicycles, thereby increasing the demand for battery replacement services. At the same time, equipment status data is also an indispensable part of battery replacement services. It covers the operating conditions of various facilities in the battery replacement station, such as the performance of charging equipment and the health status of the battery. Through real-time monitoring of these data, the smooth progress of the battery replacement process can be ensured, and potential faults can be discovered and handled in advance. Finally, as a valuable experience accumulation, historical battery replacement data can be used to mine valuable information such as users' battery replacement habits and peak hours through in-depth analysis of historical records, providing strong support for the construction of prediction models and thus improving the accuracy of predictions. Combining the above types of data, the battery replacement platform can build a more sophisticated and dynamic operation model, effectively solving the technical bottleneck of real-time data collection and updating. At the same time, with the help of advanced data analysis technology, the ability to predict battery replacement demand and load can be significantly improved, providing a scientific basis for scheduling decisions. By optimizing battery inventory management and ensuring load balancing at battery replacement stations, the quality of battery replacement services and user experience can be improved overall.

[0045] In one example, the second power condition data includes at least any of the following: battery status information, real-time location and speed data, user driving habit data, vehicle model information, environmental factor data, vehicle historical battery replacement information, and user reservation data.

[0046] In this example, the "second power situation data" mentioned involves detailed information at multiple levels, which plays a vital role in optimizing the operational efficiency and service quality of the battery swap platform. The data types involved are explained in detail as follows: First, battery status information includes the remaining battery power, health status assessment (such as the number of cycles, aging assessment), and current charging status. This information is crucial for the platform, which can help it accurately predict which batteries are about to enter the cycle that needs to be replaced or charged, so as to make corresponding preparations in advance to ensure that the battery swap station can continue to provide sufficient and high-performance battery resources to meet the user's usage needs. Secondly, real-time location and speed data involve the location tracking of electric bicycles and the user's driving speed record. These data play a key role in the real-time scheduling link. The platform can use them to predict the estimated time for users to arrive at the nearest battery swap station, thereby achieving more accurate battery replacement service arrangements, effectively reducing user waiting time, and improving service efficiency. In addition, user driving habit data is also one of the important contents of platform analysis. By conducting in-depth analysis of the user's driving pattern (such as average driving speed, acceleration and deceleration frequency, etc.), the platform can more accurately estimate the battery consumption rate, provide users with personalized battery life recommendations, and optimize the battery allocation strategy to achieve reasonable resource allocation. Furthermore, vehicle model information should not be ignored. Different models of electric bicycles may be equipped with batteries of different specifications, so it is crucial for the platform to understand the specific model of the user's vehicle. This helps the platform prepare battery resources of corresponding specifications to ensure the smooth progress of the battery replacement process and improve the efficiency of battery replacement. At the same time, environmental factor data is also one of the important factors affecting the operation of the battery replacement platform. Environmental factors including weather conditions (such as temperature and humidity changes) and road slopes will affect battery performance and electric vehicle energy consumption. For example, in cold weather conditions, battery performance may decrease, so the platform needs to adjust the priority of battery replacement according to such conditions or recommend users to perform battery replacement operations in advance. In addition, vehicle historical battery replacement information is also an important basis for platform optimization operations. By recording and analyzing users' past battery swapping behaviors (such as battery swapping frequency, time preferences, and frequently visited battery swapping stations), the platform can more accurately predict future demand patterns, thereby optimizing inventory management and battery swapping station layout, and improving resource utilization efficiency. Finally, user reservation data is also one of the key factors in improving the operational efficiency of the battery swapping platform. By obtaining users' battery swapping reservation information in advance, the platform can more effectively plan resource allocation and avoid supply and demand imbalances during peak hours, thereby providing users with a better user experience. In summary, the comprehensive use of "second power situation data" helps the battery swapping platform improve its data processing and analysis capabilities and achieve a smarter and more flexible dispatching strategy. This can not only improve the platform's operational efficiency and service quality, but also effectively balance the load of battery swapping stations, optimize battery inventory management, and enhance user experience.

[0047] In one example, before constructing a battery swap load prediction model based on the first power condition data of the battery swap station and the second power condition data of the electric vehicle, the dispatching method also includes: cleaning and deduplicating the first power condition data and the second power condition data to obtain the cleaned and deduplicated first power condition data and the cleaned and deduplicated second power condition data; and, moving and translating the variable data in the cleaned and deduplicated first power condition data and the cleaned and deduplicated second power condition data to obtain a smoothed variable data trend line, and determining the mutation value in the variable data based on the variable data trend line; and smoothing the mutation values ​​in the first power condition data and the second power condition data.

[0048] In this example, in order to overcome the limitations of the existing battery swap platform at the data processing level, especially the problems of inaccurate data and insufficient analysis depth, a series of rigorous data preprocessing measures were taken to improve data quality and enhance the accuracy of subsequent predictive analysis. The specific work steps are as follows: 1. Data cleaning and deduplication: As the first step in data preprocessing, this step aims to eliminate errors, duplicate records and other irrelevant or misleading information in the data. For example, the misreported battery inventory data that may exist in the battery swap station, or multiple records at the same electric vehicle location, will be effectively handled through cleaning and deduplication, ensuring that the data based on the subsequent analysis is accurate and unique, laying a solid foundation for in-depth analysis. 2. Mobile translation processing and trend line construction: After completing the preliminary data cleaning, mobile translation processing technology is further adopted, which is a key technology in time series analysis. By calculating the average trend of data points over time, a smooth trend line is constructed. This helps to identify cyclical fluctuations and long-term trends in the data, weaken the impact of short-term random fluctuations, and make the laws and trends behind the data clearer. 3. Mutation value detection and smoothing adjustment: After the trend line is established, further in-depth analysis of the mutation values ​​in the data is performed. These mutation values ​​appear as abnormal data points that deviate from the normal trend, which may be caused by various factors such as equipment failure and abnormal weather affecting user behavior. By identifying and smoothing these mutation values, it is possible to effectively avoid their adverse effects on the prediction model, ensuring that the model can learn the data pattern under normal conditions, thereby improving the stability and accuracy of the prediction. Through the strict implementation of the above data preprocessing steps, not only the quality and reliability of the data are significantly improved, but also solid data support is provided for the construction of battery swap load prediction models and battery swap demand prediction models. High-quality data input will help the model to more accurately understand and predict the load changes of battery swap stations and user battery swap needs, thereby guiding more reasonable and efficient dispatching decisions, optimizing battery inventory management, improving user experience, and effectively responding to the challenges faced by battery swap platforms.

[0049] In one example, a battery swap load prediction model is constructed based on the first power condition data of the battery swap station and the second power condition data of the electric vehicle, including: performing feature engineering processing on the first power condition data based on the battery swap station and the second power condition data of the electric vehicle to extract key features; and constructing composite features based on the key features; constructing an initial battery swap load prediction model based on a multivariate time series, wherein the initial battery swap load prediction model based on a multivariate time series is used to capture the time dependency of the load data, and the input of the initial battery swap load prediction model based on a multivariate time series includes multiple key feature data and multiple composite feature data, and the output includes load prediction values ​​for future time periods; based on the key features and the composite features, the initial battery swap load prediction model based on the multivariate time series is trained, and the L2 regularization technology is introduced to train the initial battery swap load prediction model based on the multivariate time series. The initial model for battery swap load prediction is optimized to prevent overfitting of the initial model for battery swap load prediction based on multivariate time series during training; the hyperparameters of the initial model for battery swap load prediction based on multivariate time series are automatically adjusted using the Bayesian optimization method to optimize the prediction accuracy of the initial model for battery swap load prediction based on multivariate time series, and obtain a basic model for battery swap load prediction based on multivariate time series; based on the key features and the composite features, an influencing factor matrix is ​​constructed, wherein the influencing factor matrix contains multiple external factor data that affect the battery swap load; the influencing factor matrix is ​​fitted to the basic model for battery swap load prediction based on multivariate time series, and the basic model for battery swap load prediction based on multivariate time series is made to fit and correct the prediction results based on the external factor data, and obtain the final battery swap load prediction model based on multivariate time series.

[0050] In this example, the goal is to build a highly accurate battery swap load prediction model through a series of complex and sophisticated data processing and modeling steps, aiming to solve the shortcomings of existing battery swap platforms in real-time data processing, prediction accuracy, and scheduling efficiency. The following is a detailed description of the technical solution: 1. Feature engineering processing: First, we deeply analyze the power usage data of the battery swap station (i.e., the first power situation data) and the power usage data of the electric vehicle (i.e., the second power situation data), and extract the key features that have a significant impact on the prediction model. These key features include but are not limited to the frequency of battery replacement, the number of battery swap requests in a specific time period, and the user density in a specific area. On this basis, we further use mathematical operations and logical combinations to construct composite features to enhance the expression ability of the original features so that they can more comprehensively reflect complex business scenarios. 2. Multivariate time series model construction: We built a prediction model based on multivariate time series analysis. The model can effectively capture the trend and seasonality of load data over time, while fully considering the interaction between multiple key features and composite features. The input of the model is the historical data of these features, and the output is the load forecast value for each future time period. 3. Prevent overfitting and model optimization: To avoid poor performance on new data due to over-complexity of the model (i.e., overfitting), we adopted L2 regularization technology. By adding penalty terms during model training, the complexity of the model is effectively limited. In addition, we also use Bayesian optimization methods to automatically adjust the model's hyperparameters, such as learning rate and regularization strength, to maximize prediction accuracy while maintaining model simplicity, thereby further optimizing model performance. 4. Consider external influencing factors: An influencing factor matrix is ​​constructed to include data on external factors such as weather conditions, holidays, and special events. These external factors may have an important impact on battery swapping demand. By incorporating the influencing factor matrix into the time series model, the model can fully consider these external factors when predicting, thereby achieving dynamic adjustment and precise correction of the prediction results. In summary, this technical solution significantly improves the accuracy and practicality of battery swap load prediction by comprehensively using advanced data processing techniques, complex model construction and optimization strategies, and the integration of external factors. This will help the battery swap platform to more effectively manage battery inventory and balance the load of each station, ultimately improving user experience and the overall operational efficiency of the platform.

[0051] In one example, a battery swap demand prediction model is constructed based on the first power situation data of the battery swap station and the second power situation data of the electric vehicle, including: performing feature engineering processing on the first power situation data based on the battery swap station and the second power situation data of the electric vehicle to extract key features; and constructing composite features based on the key features; constructing an initial battery swap demand prediction model based on a long short-term memory network, wherein the initial battery swap demand prediction model based on the long short-term memory network is used to capture the time dependency and long-term time relationship of the battery swap demand data, and the input layer of the initial battery swap demand prediction model based on the long short-term memory network is used to obtain multiple key feature data and multiple composite feature data, the LSTM layer is used to perform multi-layer stacking processing on the data input by the input layer, and the fully connected layer is used to output the predicted value of the electric vehicle battery swap demand in the future time period; based on the key features and the composite features, the initial battery swap demand prediction model based on the long short-term memory network is processed. Iterative training processing, and calculating the gradient value of the loss function of the initial model for battery swap demand prediction based on the long short-term memory network relative to the loss of the parameters of the initial model for battery swap demand prediction based on the long short-term memory network through the back-propagation algorithm, and using the Adam optimization algorithm to update the parameters of the initial model for battery swap demand prediction based on the long short-term memory network according to the gradient value, so that the initial model for battery swap demand prediction based on the long short-term memory network converges faster; and, in the process of iterative training processing of the initial model for battery swap demand prediction based on the long short-term memory network based on the key features and the composite features, adding a Dropout layer after each LSTM layer of the initial model for battery swap demand prediction based on the long short-term memory network, and randomly discarding a preset proportion of neurons, so as to prevent the initial model for battery swap demand prediction based on the long short-term memory network from overfitting the data training of the key features and the composite features, so as to obtain the final model for battery swap demand prediction based on the long short-term memory network.

[0052] In this example, the goal is to use advanced machine learning techniques, especially long short-term memory networks (LSTMs), to build a battery swap demand prediction model to improve the accuracy of predictions. The following is a detailed description of the key steps of the solution and their functions: 1. Feature engineering processing: First, extract the features that are crucial to the prediction of battery swap demand, namely "key features", from the power usage data of the battery swap station (first power situation data) and the power usage data of electric vehicles (second power situation data). These key features may include battery usage frequency, peak time for battery swaps, user behavior patterns, etc. On this basis, further construct "composite features" to enhance the model's ability to capture complex relationships through mathematical transformations or feature combinations, thereby improving the accuracy of predictions. 2. Construct an LSTM model: As a special recurrent neural network (RNN), LSTM has significant advantages in processing data with time series characteristics. Therefore, this solution uses LSTM to build a battery swap demand prediction model. The model design includes an input layer (receiving key features and composite feature data), an LSTM layer (processing time dependencies and long-term dependencies through a special gating mechanism), and a fully connected layer (outputting the predicted value of battery swap demand in the future time period). By stacking multiple LSTM layers, the learning ability of the model is enhanced, so that it can understand the complex interactions between data at a deeper level, thereby improving the accuracy of prediction. 3. Iterative training and optimization: During the model training process, the back propagation algorithm is used to calculate the gradient of the loss function with respect to the model parameters, and the Adam optimization algorithm is used to iteratively adjust the model parameters. The Adam algorithm can effectively accelerate the model convergence process due to its characteristic of adaptively adjusting the learning rate, and exhibits good performance when processing a large number of parameters. Through multiple iterative training, the model parameters are continuously optimized to improve the accuracy of battery swap demand prediction. 4. Strategy to prevent overfitting: In order to prevent the model from overfitting during training, this solution adds a Dropout layer after the LSTM layer. The Dropout layer achieves a regularization effect by randomly "discarding" a part of the neurons, reducing the model's excessive dependence on specific features or patterns. This helps to improve the generalization ability of the model, so that it can maintain good prediction performance on unseen data. In summary, this technical solution uses the LSTM network in deep learning, combined with feature engineering, iterative training and regularization strategies, to build a model that can accurately predict future battery swap demand. This model not only improves the accuracy and reliability of predictions, but also solves the problems of manual reliance, data lag, and inaccurate predictions in traditional methods through automation and intelligent methods. This will be of great significance to improving the operational efficiency and user satisfaction of the battery swap platform.

[0053] In one example, a multi-objective optimization method is used to construct a battery swap order optimization allocation model based on the battery swap demand prediction model and the battery swap load prediction model, including: generating a load balancing objective function and an inventory balance objective function, and combining the battery swap demand prediction model and the battery swap load prediction model to construct an initial model for optimizing the allocation of battery swap orders based on a multi-objective optimization algorithm, wherein the load balancing objective function is:

[0054] ,

[0055] is the load of the i-th battery swap station, is the capacity of the i-th battery swap station;

[0056] The inventory balance objective function is:

[0057] ,

[0058] For the The battery inventory of each battery swap station, For the The battery swapping demand of each battery swapping station; integrating the genetic algorithm and the particle swarm optimization algorithm into the solution module of the initial model of the battery swapping order optimization allocation based on the multi-objective optimization algorithm, and obtaining the final battery swapping order optimization allocation model based on the multi-objective optimization algorithm, wherein the final battery swapping order optimization allocation model based on the multi-objective optimization algorithm adopts the genetic algorithm and the particle swarm optimization algorithm to solve the load balancing objective function and the inventory balance objective function.

[0059] In this example, we mainly focus on the data real-time, prediction accuracy and scheduling efficiency problems encountered by the battery swap platform in the dispatching process, and propose a battery swap order optimization allocation model based on a multi-objective optimization method, aiming to improve the load balancing and inventory management of battery swap stations, thereby improving the overall operational efficiency and user experience. The following is a detailed explanation of the core part of this technical solution: 1. Multi-objective optimization: In this scenario, there is more than one optimization goal, but two key factors are considered at the same time-load balancing and inventory balance. Multi-objective optimization aims to find a solution that makes all objective functions reach the optimal state as much as possible or achieve a satisfactory trade-off between the various objectives. 2. Load balancing objective function ( ): This function measures the minimization of the sum of the squares of the loads of each battery swap station and their capacities. represents the instantaneous load of the ith battery swap station, is its maximum capacity. By minimizing this value, we can ensure that no single battery swap station is overloaded and achieve load balancing of the entire network. 3. Inventory balance objective function ( ): This function focuses on maintaining the battery inventory at each battery swap station ( ) and current battery replacement demand ( ) is as small as possible to achieve inventory balance. Similarly, by minimizing this function value, inventory can be effectively managed to avoid resource waste or insufficient supply caused by excessive or insufficient inventory. 4. Integration of genetic algorithm and particle swarm optimization algorithm: In order to solve the above complex multi-objective problems, the model adopts two efficient optimization algorithms-genetic algorithm and particle swarm optimization algorithm. Genetic algorithm simulates natural selection and genetic mechanism, and searches for the optimal solution in the solution space through operations such as crossover and mutation; while particle swarm optimization algorithm imitates the foraging behavior of bird flocks, and approaches the optimal solution by exploring and adjusting the flight direction and speed of individuals (particles) in the solution space. The integration of these two algorithms aims to combine the advantages of both, improve the speed and quality of finding the optimal solution, especially when dealing with multi-objective optimization problems, and be able to more flexibly balance between different objectives. Through this comprehensive optimization strategy, the battery swap platform can more intelligently consider the real-time status of each battery swap station when dispatching orders, ensure that the battery swap demand is efficiently met, optimize resource allocation, reduce unnecessary transportation costs, and improve the service quality and response speed of the battery swap station. Ultimately, this will bring a better user experience, enhance users' trust and reliance on battery swap services, and promote a healthier and more sustainable development of the electric bicycle battery swap market.

[0060] In one example, in response to a battery swap request order sent by a client, the battery swap order optimization allocation model is used to optimize and allocate the battery swap request order, determine the battery swap station corresponding to each battery swap request order, and generate a battery swap delivery order based on the battery swap station corresponding to the battery swap request order, including: in response to the battery swap request order sent by the client, a set of initial solutions is generated for the battery swap request order using the battery swap order optimization allocation model, wherein each solution in the set of initial solutions represents a battery swap station corresponding to the battery swap request order, and a delivery route based on the corresponding battery swap station; the fitness of each initial solution is calculated according to the objective function in the battery swap order optimization allocation model, and a plurality of initial solutions whose fitness is higher than a first preset threshold are selected by a roulette wheel method, and the selected solutions are sorted; The selected multiple initial solutions are individually propagated to obtain new initial solutions; the step of "calculating the fitness of each initial solution according to the objective function in the battery swap order optimization allocation model" is re-executed on the new initial solutions until the maximum number of iterations is reached or the fitness improvement is less than the second preset threshold; the retained initial solutions are organized into a group of initial particles, and the speed and position of each initial particle are initialized, and the speed and position of each initial particle are calculated according to the objective function in the battery swap order optimization allocation model to obtain the fitness of each initial solution, wherein each initial solution corresponds to an initial particle; the first formula and the second formula are used to calculate the optimal position of each particle itself and the optimal position of the group to update the speed and position of the particle, wherein the first formula is:

[0061] ,

[0062] is the particle velocity, is the particle position, is the preset optimal position of the particle itself, is the preset optimal position of the group, is the inertia weight, and is the acceleration constant, and is a random number;

[0063] The second formula is:

[0064] ,

[0065] The step of "calculating and processing the speed and position of each initial particle according to the objective function in the battery swap order optimization allocation model to obtain the fitness of each initial solution" is re-executed for each particle until the maximum number of iterations is reached or the fitness improvement is less than a second preset threshold; the particle with the highest fitness is adopted as the optimal node of the battery swap request order, and the battery swap station corresponding to the battery swap request order is determined according to the particle with the highest fitness, and a battery swap delivery order is generated based on the battery swap station corresponding to the battery swap request order and the delivery route of the battery swap station.

[0066] In this example, a process of processing battery swap request orders through an optimization algorithm is elaborated in detail, aiming to solve the efficiency and accuracy problems encountered by the battery swap platform in the dispatching process. A hybrid optimization strategy that combines the ideas of genetic algorithm and particle swarm optimization algorithm (PSO) is adopted here to improve the load balancing, inventory management and dispatch efficiency of the battery swap station. The following are the key steps and principle explanations of the process: Initial solution generation and screening: After receiving the battery swap request order, the system first generates a set of initial solutions using the battery swap order optimization allocation model. Each solution represents a battery swap station to which a battery swap request may be assigned and its corresponding dispatch route. - Then, the fitness of each solution is calculated according to the objective function (such as the load balancing and inventory balancing objectives mentioned above), that is, how the solution performs comprehensively under the multi-objective optimization framework. The roulette wheel selection method is used to retain those solutions with fitness higher than the preset threshold, which are regarded as better solutions. Then new solutions are generated through "individual reproduction processing" (similar to crossover and mutation in genetic algorithms), aiming to explore more possibilities in the solution space. Particle swarm optimization loop: Convert the selected solutions into particles in the particle swarm optimization algorithm. Each particle represents a potential swap station allocation solution, including position (the solution itself) and speed (the adjustment direction and rate of the solution). The speed of each particle is dynamically adjusted using the first formula (speed update formula), which combines the current speed of the particle, the individual historical best position ( ), the best historical position of the group ( ) and random factors ( , ), which aims to balance the ability to explore new areas and utilize the known best solution. The second formula (position update formula) moves the particle position according to the updated speed, that is, actually adjusts the configuration of the solution. The above process is continuously iterated, and the fitness of each particle's new position (solution) is evaluated according to the objective function until the predetermined stop condition is reached (such as the maximum number of iterations or the fitness improvement is not enough to reach the set threshold). Result determination and delivery order generation: Finally, the particle with the highest fitness is selected from all particles as the optimal solution, which means that the most suitable battery swap station allocation plan has been found. According to this optimal solution, the battery swap station corresponding to the battery swap request order is determined, and the battery swap delivery order is generated in combination with the specific delivery route information of the battery swap station to guide the actual battery swap service operation. This technical solution effectively solves the problems of real-time data processing, low prediction accuracy and unreasonable scheduling through advanced optimization algorithms, realizes the intelligent allocation of battery swap requests, and thus improves the operational efficiency, inventory management efficiency and user experience of the battery swap platform. Through continuous iterative optimization, the system can flexibly respond to complex and changing actual needs and ensure the efficiency, balance and reliability of battery swap services.

[0067] Reference Figure 2 This embodiment provides a dispatching device for a battery swap platform, including:

[0068] A collecting unit 1, used to collect first power situation data describing at least the historical power usage and real-time power usage of the battery swap station and second power situation data describing at least the historical power usage and real-time power usage of the electric vehicle;

[0069] A first prediction unit 2 is used to construct a power-swap load prediction model based on the first power condition data of the power-swap station and the second power condition data of the electric vehicle, wherein the power-swap load prediction model is used to predict the load transformation of each power-swap station in a future time period;

[0070] A second prediction unit 3 is used to construct a battery swap demand prediction model based on the first power situation data of the battery swap station and the second power situation data of the electric vehicle, wherein the battery swap demand prediction model is used to predict the usage demand of the battery swap station in different time periods and different areas;

[0071] The allocation unit 4 is used to adopt a multi-objective optimization method to construct a battery swap order optimization allocation model based on the battery swap demand prediction model and the battery swap load prediction model, wherein the battery swap order optimization allocation model is used to allocate battery swap stations to battery swap orders according to the load conditions and usage requirements of each battery swap station;

[0072] The response unit 5 is used to respond to the battery swap request order sent by the client, adopt the battery swap order optimization allocation model to optimize the allocation processing for the battery swap request order, determine the battery swap station corresponding to each battery swap request order, and generate a battery swap delivery order based on the battery swap station corresponding to the battery swap request order.

[0073] In this embodiment, for the specific implementation of each unit in the above device embodiment, please refer to the above method embodiment, which will not be repeated here.

[0074] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0075] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0076] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0077] In summary, an intelligent dispatching method for a battery swap platform is proposed in the embodiment of the present invention, aiming to optimize the resource allocation of battery swap stations through accurate data analysis and advanced algorithm models, thereby improving the efficiency and user experience of battery swap services. The following are the core points of the technical solution: 1. Data integration and preprocessing: This solution first comprehensively collects the power situation data related to the battery swap platform, covering multi-dimensional data such as historical and real-time power consumption information of the battery swap station and the use status of electric vehicles. Subsequently, the data quality is ensured by cleaning, deduplication, translation processing and mutation value smoothing of the data, providing a reliable data foundation for subsequent model construction. 2. Prediction model construction: Battery swap load prediction model: This solution adopts a multivariate time series analysis method, combined with feature engineering and composite features, to construct a battery swap load prediction model. At the same time, by introducing L2 regularization, Bayesian optimization and other means, the model can be effectively prevented from overfitting, ensuring accurate prediction of the future load of the battery swap station. Battery swap demand prediction model: In view of the battery swap demand of electric vehicles, this solution uses long short-term memory network (LSTM) technology to capture time series characteristics. At the same time, by introducing the Dropout layer, the model is prevented from overfitting, and the prediction accuracy is continuously optimized through iterative training. 3. Multi-objective optimization allocation model: This solution comprehensively considers the two major goals of load balancing and inventory balance, and adopts a multi-objective optimization strategy that integrates genetic algorithm and particle swarm optimization algorithm. By intelligently allocating battery swap orders, the workload of each battery swap station is balanced, the user waiting time is reduced, and inventory management is optimized. 4. Order processing and delivery: After receiving the battery swap request from the client, this model can quickly generate a set of initial solutions. Subsequently, the roulette wheel selection and particle swarm optimization algorithm are continuously iterated to find the best battery swap station allocation plan. Based on this solution, a specific battery swap delivery order is generated to ensure the timeliness and efficiency of the service. 5. System implementation: This technical solution can be designed as a dispatching device including a collection unit, a prediction unit, a distribution unit and a response unit, and can also be deployed on a computer device and executed through a specific computer program. In addition, the solution can also be implemented as a software product on a computer-readable storage medium, which is convenient for flexible application in different scenarios. In summary, this technical solution realizes the intelligence and efficiency of the battery swap platform dispatching process through highly integrated data analysis, precise prediction models and optimization algorithms.

[0078] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0079] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0080] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for dispatching orders on a battery swap platform. It is characterized in that The following steps are involved: Collecting first power situation data describing at least historical power usage and real-time power usage of the battery swap station and second power situation data describing at least historical power usage and real-time power usage of the electric vehicle; Based on the first power situation data of the power swap station and the second power situation data of the electric vehicle, a power swap load prediction model is constructed, wherein the power swap load prediction model is used to predict the load change of each power swap station in a future time period; Based on the first power situation data of the battery swap station and the second power situation data of the electric vehicle, a battery swap demand prediction model is constructed, wherein the battery swap demand prediction model is used to predict the usage demand of the battery swap station in different time periods and different regions; A multi-objective optimization method is adopted to construct a battery swap order optimization allocation model based on the battery swap demand prediction model and the battery swap load prediction model, wherein the battery swap order optimization allocation model is used to allocate battery swap stations to battery swap orders according to the load conditions and usage requirements of each battery swap station; In response to the battery swap request order sent by the client, the battery swap order optimization allocation model is used to optimize the allocation processing for the battery swap request order, determine the battery swap station corresponding to each battery swap request order, and generate a battery swap delivery order based on the battery swap station corresponding to the battery swap request order; Among them, a multi-objective optimization method is adopted to construct a battery swap order optimization allocation model based on the battery swap demand prediction model and the battery swap load prediction model, including: generating a load balancing objective function and an inventory balancing objective function, and combining the battery swap demand prediction model and the battery swap load prediction model to construct an initial model for battery swap order optimization allocation based on a multi-objective optimization algorithm, The load balancing objective function is: , is the load of the i-th battery swap station, is the capacity of the i-th battery swap station; The inventory balance objective function is: , For the The battery inventory of each battery swap station, For the The battery replacement demand of each battery replacement station; The genetic algorithm and the particle swarm optimization algorithm are integrated into the solution module of the initial model of the battery swap order optimization allocation based on the multi-objective optimization algorithm to obtain the final battery swap order optimization allocation model based on the multi-objective optimization algorithm, wherein the final battery swap order optimization allocation model based on the multi-objective optimization algorithm adopts the genetic algorithm and the particle swarm optimization algorithm to solve the load balancing objective function and the inventory balance objective function.

2. The dispatching method according to claim 1, It is characterized in that The first power situation data includes at least any of the following: battery inventory information, battery swap station load information, battery swap station location and coverage, environmental factor data, equipment status data, historical battery swap data, and user reservation data; The second power status data includes at least any of the following: battery status information, real-time location and speed data, user driving habit data, vehicle model information, environmental factor data, vehicle historical battery replacement information and user reservation data.

3. The dispatching method according to claim 1, It is characterized in that Before building a power swap load prediction model based on the first power situation data of the power swap station and the second power situation data of the electric vehicle, the dispatching method further includes: Cleaning and deduplicating the first power situation data and the second power situation data to obtain cleaned and deduplicated first power situation data and cleaned and deduplicated second power situation data; and, The variable data in the first electric energy condition data that has been cleaned and deduplicated and the second electric energy condition data that has been cleaned and deduplicated are moved and translated to obtain a smoothed variable data trend line, and the mutation value in the variable data is determined based on the variable data trend line; and the mutation value in the first electric energy condition data and the second electric energy condition data is smoothed.

4. The dispatching method according to claim 1, It is characterized in that Based on the first power situation data of the power swap station and the second power situation data of the electric vehicle, a power swap load prediction model is constructed, including: Performing feature engineering processing on the first power situation data based on the battery swap station and the second power situation data of the electric vehicle to extract key features; and constructing composite features based on the key features; Constructing an initial model for battery swap load prediction based on multivariate time series, wherein the initial model for battery swap load prediction based on multivariate time series is used to capture the time dependency of load data, and the input of the initial model for battery swap load prediction based on multivariate time series includes multiple key feature data and multiple composite feature data, and the output includes load forecast values ​​for each future time period; Based on the key features and the composite features, the initial model for battery swap load prediction based on multivariate time series is trained, and the L2 regularization technology is introduced to optimize the initial model for battery swap load prediction based on multivariate time series to prevent the initial model for battery swap load prediction based on multivariate time series from overfitting during the training process; Using the Bayesian optimization method to automatically adjust the hyperparameters of the initial model for battery swap load prediction based on multivariate time series to optimize the prediction accuracy of the initial model for battery swap load prediction based on multivariate time series, and obtain a basic model for battery swap load prediction based on multivariate time series; Based on the key features and the composite features, an influencing factor matrix is ​​constructed, wherein the influencing factor matrix contains a plurality of external factor data that affect the battery swap load; The influencing factor matrix is ​​fitted into the basic model for battery swap load prediction based on multivariate time series, and the basic model for battery swap load prediction based on multivariate time series is made to fit and correct the prediction results based on the external factor data to obtain the final battery swap load prediction model based on multivariate time series.

5. The dispatching method according to claim 1, It is characterized in that Based on the first power situation data of the power swap station and the second power situation data of the electric vehicle, a power swap demand prediction model is constructed, including: Performing feature engineering processing on the first power situation data based on the battery swap station and the second power situation data of the electric vehicle to extract key features; and constructing composite features based on the key features; Construct an initial model for battery swap demand prediction based on a long short-term memory network, wherein the initial model for battery swap demand prediction based on a long short-term memory network is used to capture the time dependency and long-term time relationship of battery swap demand data, and the input layer of the initial model for battery swap demand prediction based on a long short-term memory network is used to obtain multiple key feature data and multiple composite feature data, the LSTM layer is used to perform multi-layer stacking processing on the data input by the input layer, and the fully connected layer is used to output the predicted value of the battery swap demand of electric vehicles in the future time period; Based on the key features and the composite features, the initial model for predicting battery swapping demand based on the long short-term memory network is iteratively trained, and the gradient value of the loss function of the initial model for predicting battery swapping demand based on the long short-term memory network relative to the parameters of the initial model for predicting battery swapping demand based on the long short-term memory network is calculated by a back propagation algorithm, and the parameters of the initial model for predicting battery swapping demand based on the long short-term memory network are updated according to the gradient value by using the Adam optimization algorithm, so that the initial model for predicting battery swapping demand based on the long short-term memory network is accelerated to converge; Furthermore, in the process of iteratively training the initial model for battery swap demand prediction based on the long short-term memory network based on the key features and the composite features, a Dropout layer is added after each LSTM layer of the initial model for battery swap demand prediction based on the long short-term memory network, and a preset proportion of neurons are randomly discarded to prevent the initial model for battery swap demand prediction based on the long short-term memory network from overfitting the data training of the key features and the composite features, so as to obtain the final battery swap demand prediction model based on the long short-term memory network.

6. The dispatching method according to claim 1, It is characterized in that In response to the battery swap request order sent by the client, the battery swap order optimization allocation model is used to optimize the allocation processing for the battery swap request order, determine the battery swap station corresponding to each battery swap request order, and generate a battery swap delivery order based on the battery swap station corresponding to the battery swap request order, including: In response to a battery swap request order sent by a client, a set of initial solutions is generated for the battery swap request order using the battery swap order optimization allocation model, wherein each solution in the set of initial solutions represents a battery swap station corresponding to the battery swap request order and a delivery route based on the corresponding battery swap station; The fitness of each initial solution is calculated according to the objective function in the battery swap order optimization allocation model, and a plurality of initial solutions whose fitness is higher than a first preset threshold are selected by a roulette wheel method, and individual reproduction is performed on the selected plurality of initial solutions to obtain a new initial solution; Re-execute the step of "calculating the fitness of each initial solution according to the objective function in the battery swap order optimization allocation model" for the new initial solution until the maximum number of iterations is reached or the fitness improvement is less than the second preset threshold; The retained initial solutions are organized into a group of initial particles, and the speed and position of each initial particle are initialized, and the speed and position of each initial particle are calculated and processed according to the objective function in the battery swap order optimization allocation model to obtain the fitness of each initial solution, wherein each initial solution corresponds to an initial particle; The first formula and the second formula are used to calculate the optimal position of each particle and the optimal position of the group to update the speed and position of the particle, wherein the first formula is: ， is the particle velocity, is the particle position, is the preset optimal position of the particle itself, is the preset optimal position of the group, is the inertia weight, and is the acceleration constant, and is a random number; The second formula is: ， Re-execute the step of "calculating and processing the speed and position of each initial particle according to the objective function in the battery swap order optimization allocation model to obtain the fitness of each initial solution" for each particle until the maximum number of iterations is reached or the fitness improvement is less than the second preset threshold; The particle with the highest fitness is adopted as the optimal node of the battery swap request order, and the battery swap station corresponding to the battery swap request order is determined according to the particle with the highest fitness, and a battery swap delivery order is generated based on the battery swap station corresponding to the battery swap request order and the delivery route of the battery swap station.

7. A dispatching device for a battery swap platform, It is characterized in that include: A collecting unit, used to collect first power situation data describing at least the historical power usage and real-time power usage of the battery swap station and second power situation data describing at least the historical power usage and real-time power usage of the electric vehicle; A first prediction unit is used to construct a power-swap load prediction model based on the first power condition data of the power-swap station and the second power condition data of the electric vehicle, wherein the power-swap load prediction model is used to predict the load change of each power-swap station in a future time period; A second prediction unit is used to construct a battery swap demand prediction model based on the first power situation data of the battery swap station and the second power situation data of the electric vehicle, wherein the battery swap demand prediction model is used to predict the usage demand of the battery swap station in different time periods and different areas; An allocation unit, configured to adopt a multi-objective optimization method to construct a battery swap order optimization allocation model based on the battery swap demand prediction model and the battery swap load prediction model, wherein the battery swap order optimization allocation model is used to allocate battery swap stations to battery swap orders according to the load conditions and usage requirements of each battery swap station; A response unit, used to respond to the battery swap request order sent by the client, use the battery swap order optimization allocation model to optimize the allocation processing for the battery swap request order, determine the battery swap station corresponding to each battery swap request order, and generate a battery swap delivery order based on the battery swap station corresponding to the battery swap request order; Wherein, a multi-objective optimization method is adopted to construct a battery swap order optimization allocation model based on the battery swap demand prediction model and the battery swap load prediction model, including: Generate a load balancing objective function and an inventory balancing objective function, and combine the battery swap demand prediction model and the battery swap load prediction model to construct an initial model for optimizing the allocation of battery swap orders based on a multi-objective optimization algorithm. The load balancing objective function is: , is the load of the i-th battery swap station, is the capacity of the i-th battery swap station; The inventory balance objective function is: , For the The battery inventory of each battery swap station, For the The battery replacement demand of each battery replacement station; The genetic algorithm and the particle swarm optimization algorithm are integrated into the solution module of the initial model of the battery swap order optimization allocation based on the multi-objective optimization algorithm to obtain the final battery swap order optimization allocation model based on the multi-objective optimization algorithm, wherein the final battery swap order optimization allocation model based on the multi-objective optimization algorithm adopts the genetic algorithm and the particle swarm optimization algorithm to solve the load balancing objective function and the inventory balance objective function.

8. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory. It is characterized in that When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images