Smart energy station site selection optimization method and device, electronic equipment and storage medium

Abstract

The invention provides a smart energy station site selection optimization method and device, electronic equipment and a storage medium, and relates to the field of new energy, and the method comprises the steps: constructing a space-time diagram through a space-time mask auto-encoder in a space-time generation type pre-training model; pre-training the space-time diagram by using a space-time generation type pre-training model, learning an internal space-time mode of a truck energy charging behavior through an adaptive mask strategy, obtaining space-time representation, and training a space-time prediction dynamic diffusion model; and by taking the lowest comprehensive cost as an objective function, constructing a smart energy station site selection optimization model so as to perform optimal site selection of the smart energy station. The intelligent energy station site selection optimization model constructed by the invention is a hydrogenation, charging, photovoltaic and energy storage integrated multi-objective optimization model, different energy supply demands of new energy trucks are comprehensively considered, and the accuracy and real-time performance of site selection prediction are improved. In addition, environmental protection and power grid stability are more emphasized in site selection decision, and comprehensive optimization of the site selection scheme is realized.

Description

Technical Field

[0001] The present application relates to the field of new energy technology, and specifically to a method, device, electronic device and storage medium for optimizing the site selection of a smart energy station. Background Art

[0002] With the continuous advancement of new energy truck technology and the increase in policy support, its share is showing a rapid growth trend. Many regions have formulated specific new energy truck development goals, including the expected growth in new energy commercial vehicle production, the comprehensive electrification plan for public vehicles, and specific promotion measures for new energy trucks.

[0003] The rapid development of new energy trucks faces the following challenges: ① Although charging stations for new energy vehicles have been built on a large scale, the existing charging station infrastructure mostly considers the charging needs of new energy family passenger cars with 7 seats or less, and rarely considers the charging needs of electric trucks: in terms of charging power, truck charging piles usually support higher charging power, and electric trucks basically support dual-gun charging of one car, with two fast charging interfaces; in terms of auxiliary power supply, the BMS auxiliary power supply of electric trucks is 24V, while the BMS auxiliary power supply of small cars is 12V. ② With the gradual rise of the hydrogen fuel cell vehicle market, the demand for hydrogen refueling stations is also growing. However, the existing hydrogen refueling station network still has obvious deficiencies in quantity and distribution, and cannot fully meet the growing demand for charging. The current hydrogen refueling stations are mostly concentrated in specific areas, which makes it difficult for many potential users to easily obtain hydrogen energy due to geographical restrictions. In addition, the service capacity of existing hydrogen refueling stations is also limited by their scale and technical level, and they cannot efficiently provide fast charging services for a large number of vehicles during peak hours. Based on this, the existing charging stations and hydrogen refueling stations cannot meet the charging needs of new energy trucks. Summary of the invention

[0004] In response to the shortcomings of the prior art, the present application provides a smart energy station site selection optimization method, device, electronic device and storage medium, which solves the problems of excessive reliance on specific energy types, lack of comprehensive energy management strategies, and defects in dynamic demand response and multi-objective optimization in the site selection of smart energy stations for new energy trucks.

[0005] To achieve the above objectives, this application is implemented through the following technical solutions:

[0006] In a first aspect, an embodiment of the present application provides a method for optimizing the site selection of a smart energy station, the method comprising: obtaining first data information of a new energy truck in a historical period, and preprocessing the first data information to obtain second data information; analyzing the operation mode of the new energy truck based on the second data information, and determining the main operation area and potential growth area of ​​the new energy truck; wherein the operation mode includes the driving route, the type of cargo, and the transportation frequency; constructing a spatiotemporal graph through the spatiotemporal masked autoencoder in the spatiotemporal generative pre-training model to characterize the charging behavior of the truck; wherein the spatiotemporal graph uses the truck and the charging station as nodes, and the time between the truck and the charging station is The space-time relationship is used as an edge; the space-time graph is pre-trained using a space-time generative pre-training model, and the intrinsic space-time pattern of the truck charging behavior is learned through an adaptive masking strategy to obtain a space-time representation; the space-time representation is used as input information to train a space-time prediction dynamic diffusion model to learn the change in charging demand from the current state to the future state in order to predict the charging demand; based on the charging demand prediction of the space-time prediction dynamic diffusion model, with the lowest comprehensive cost as the objective function, a smart energy station site selection optimization model is constructed to optimize the site selection of the smart energy station; among them, the smart energy station site selection optimization model is a multi-objective optimization model that integrates hydrogen refueling, charging, photovoltaics and energy storage.

[0007] In the second aspect, an embodiment of the present application provides a smart energy station site selection optimization device, which includes a preprocessing module, an analysis and determination module, a space-time graph construction module, a pretraining module, a learning prediction module and a model construction module; wherein the preprocessing module is used to obtain first data information of new energy trucks in a historical period, and preprocess the first data information to obtain second data information; the analysis and determination module is used to analyze the operation mode of the new energy truck based on the second data information, and determine the main operation area and potential growth area of ​​the new energy truck; wherein the operation mode includes driving routes, cargo types and transportation frequencies; the space-time graph construction module is used to construct a space-time graph through the space-time mask autoencoder in the space-time generative pre-training model to characterize the truck charging behavior; wherein the space-time graph converts cargo The truck and the charging station are taken as nodes, and the spatiotemporal relationship between the truck and the charging station is taken as the edge; the pre-training module is used to pre-train the spatiotemporal graph using the spatiotemporal generative pre-training model, and learn the intrinsic spatiotemporal pattern of the truck charging behavior through an adaptive masking strategy to obtain the spatiotemporal representation; the learning prediction module is used to take the spatiotemporal representation as input information, train a spatiotemporal prediction dynamic diffusion model to learn the change of charging demand from the current state to the future state in order to predict the charging demand; the model construction module is used to predict the charging demand based on the spatiotemporal prediction dynamic diffusion model, take the lowest comprehensive cost as the objective function, and construct a smart energy station site selection optimization model to optimize the site selection of the smart energy station; among them, the smart energy station site selection optimization model is a multi-objective optimization model integrating hydrogen refueling, charging, photovoltaics and energy storage.

[0008] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor, a memory, and a program stored in the memory and executable on the processor, wherein when the program is executed by the processor, the smart energy station site selection optimization method in the aforementioned first aspect is implemented.

[0009] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores a program or instruction. When the program or instruction is executed by a processor, the smart energy station site selection optimization method in the aforementioned first aspect is implemented.

[0010] This application provides a method, device, electronic device and storage medium for optimizing the site selection of a smart energy station. Compared with the prior art, it has the following beneficial effects:

[0011] The smart energy station site selection optimization model constructed in this application is a multi-objective optimization model that integrates hydrogenation, charging, photovoltaics and energy storage. It comprehensively considers the different energy supply needs of new energy trucks, and improves the accuracy and real-time performance of site selection prediction by introducing deep learning algorithms. In addition, the present invention pays more attention to environmental protection and grid stability in site selection decisions, and achieves comprehensive optimization of site selection plans. The present invention pays more attention to environmental protection in site selection decisions, optimizes the energy structure and improves energy utilization efficiency by introducing photovoltaic power generation and energy storage equipment. This site selection optimization that comprehensively considers multiple energy supply methods such as charging and hydrogenation will help reduce the impact on the environment and promote the logistics industry to develop in a greener and smarter direction. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0013] Figure 1 is an exemplary schematic diagram of the smart energy station site selection optimization method provided in an embodiment of the present application;

[0014] Figure 2 is another exemplary schematic diagram of the smart energy station site selection optimization method provided in an embodiment of the present application;

[0015] Figure 3 This is a framework diagram for predicting the charging demand of new energy trucks for logistics enterprises provided in an embodiment of the present application;

[0016] Figure 4 It is a schematic diagram of the design of the smart energy station system provided in the embodiment of the present application;

[0017] Figure 5 It is a flow chart of the smart energy station site selection optimization algorithm provided in the embodiment of the present application;

[0018] Figure 6 It is a structural schematic diagram of a smart energy station site selection optimization device provided in an embodiment of the present application;

[0019] Figure 7 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0020] 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 are clearly and completely described. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0022] The embodiments of the present application provide a smart energy station site selection optimization method, device, electronic device and storage medium to solve the problems of excessive reliance on specific energy types, lack of comprehensive energy management strategies, and defects in dynamic demand response and multi-objective optimization in the site selection of smart energy stations for new energy trucks.

[0023] The technical solution in the embodiment of the present application is to solve the above technical problems, and the overall idea is as follows:

[0024] With the continuous advancement of new energy truck technology and the increase in policy support, its share is showing a rapid growth trend. Many regions have formulated specific new energy truck development goals, including the expected growth in new energy commercial vehicle production, the comprehensive electrification plan for public vehicles, and specific promotion measures for new energy trucks.

[0025] The rapid development of new energy trucks faces the following challenges: ① Although charging stations for new energy vehicles have been built on a large scale, the existing charging station infrastructure mostly considers the charging needs of new energy family passenger cars with 7 seats or less, and rarely considers the charging needs of electric trucks: in terms of charging power, truck charging piles usually support higher charging power, and electric trucks basically support dual-gun charging of one car, with two fast charging interfaces; in terms of auxiliary power supply, the BMS auxiliary power supply of electric trucks is 24V, while the BMS auxiliary power supply of small cars is 12V. ② With the gradual rise of the hydrogen fuel cell vehicle market, the demand for hydrogen refueling stations is also growing. However, the existing hydrogen refueling station network still has obvious deficiencies in quantity and distribution, and cannot fully meet the growing demand for charging. The current hydrogen refueling stations are mostly concentrated in specific areas, which makes it difficult for many potential users to easily obtain hydrogen energy due to geographical restrictions. In addition, the service capacity of existing hydrogen refueling stations is also limited by their scale and technical level, and they cannot efficiently provide fast charging services for a large number of vehicles during peak hours.

[0026] Based on the above two points, the existing charging stations and hydrogen refueling stations cannot meet the charging needs of new energy trucks. Existing technologies often ignore the comprehensive layout of multi-energy refueling stations in site selection optimization methods, that is, the coordinated planning of charging and hydrogen refueling facilities. In addition, most methods do not make full use of renewable energy sources, such as solar energy, and energy storage technology, which is an important deficiency in terms of energy costs and environmental impacts. Furthermore, existing technologies usually lack the ability to integrate and dynamically adjust real-time data, which is particularly important when responding to rapid changes in logistics demand. Finally, existing methods are not comprehensive enough in considering the comprehensive evaluation of cost, efficiency and environmental impact. They often only focus on a single indicator and ignore the importance of multi-objective optimization.

[0027] In summary, existing technologies have certain limitations in the site selection optimization of smart energy stations for new energy trucks, including over-reliance on specific energy types, lack of comprehensive energy management strategies, and deficiencies in dynamic demand response and multi-objective optimization.

[0028] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0029] The following first introduces a smart energy station site selection optimization method provided in an embodiment of the present application.

[0030] The present invention provides a flow chart of a smart energy station site selection optimization method, please refer to Figure 1 , Figure 2 and Figure 3 , the smart energy station site selection optimization method may include the following steps S110-S160.

[0031] S110: Acquire first data information of new energy trucks in a historical period, and pre-process the first data information to obtain second data information.

[0032] S120. Analyze the operation mode of the new energy truck based on the second data information, and determine the main operation area and potential growth area of ​​the new energy truck; wherein the operation mode includes driving routes, cargo types and transportation frequencies.

[0033] S130. Construct a spatiotemporal graph through a spatiotemporal masked autoencoder in a spatiotemporal generative pre-trained model to characterize the charging behavior of the truck; wherein the spatiotemporal graph uses the truck and the charging station as nodes, and uses the spatiotemporal relationship between the truck and the charging station as an edge.

[0034] S140. Pre-train the spatiotemporal graph using a spatiotemporal generative pre-training model, learn the intrinsic spatiotemporal pattern of the truck charging behavior through an adaptive masking strategy, and obtain a spatiotemporal representation.

[0035] S150, taking the spatiotemporal representation as input information, training a spatiotemporal prediction dynamic diffusion model to learn the change of charging demand from the current state to the future state to perform charging demand prediction.

[0036] S160. Based on the charging demand prediction of the spatiotemporal prediction dynamic diffusion model, with the lowest comprehensive cost as the objective function, a smart energy station site selection optimization model is constructed to optimize the site selection of smart energy stations; among them, the smart energy station site selection optimization model is a multi-objective optimization model that integrates hydrogen refueling, charging, photovoltaics and energy storage.

[0037] The above is a specific implementation method of the smart energy station site selection optimization method provided in the embodiment of the present application. The charging demand prediction process of new energy trucks includes spatiotemporal generative pre-training and diffusion model prediction of charging demand; the pre-training stage uses the masked autoencoder MAE task as the training target, and reconstructs the masked information of the spatiotemporal data based on the unmasked spatiotemporal data by learning the spatiotemporal representation function. The smart energy station site selection optimization model takes into account the geographical location, traffic convenience, land cost, environmental impact, and grid access conditions, and takes into account the charging aggregator cost-user cost-grid stability cost; the smart energy station site selection optimization model can evaluate the energy needs of different types of new energy trucks, including electricity and hydrogen, and consider the energy consumption characteristics and recharging time of different models.

[0038] It should be noted that the present invention predicts the incremental charging / hydrogenation demand of new energy trucks of logistics enterprises in the future by collecting and analyzing the historical and current data of new energy trucks, including the number, type, distribution and frequency of use of vehicles, as well as the infrastructure data of existing charging stations and hydrogenation stations. Using the spatiotemporal generative pre-training model and the spatiotemporal prediction dynamic diffusion model, the present invention can predict the future trend of charging demand for new energy trucks, and design a smart energy station model that integrates hydrogenation, charging, photovoltaic power generation and energy storage, taking into account key factors such as geographical location, traffic convenience, land cost, environmental impact and grid access conditions.

[0039] It should also be noted that the smart energy station site selection optimization model constructed by the present invention is a multi-objective optimization model, which achieves the optimal balance of charging aggregator costs, user costs and grid stability costs while ensuring economy, environmental friendliness and operational efficiency. By using the dung beetle optimization algorithm that integrates the black widow idea, the present invention completes the optimal site selection of smart energy stations. This method combines local search with global search, balances exploration and development, achieves efficient search, and outputs the optimal or suboptimal charging station site selection plan after meeting the convergence criteria.

[0040] In one example, in the process of constructing a spatiotemporal graph, in order to capture the spatiotemporal information of charging of new energy trucks, a three-dimensional tensor is defined: in, represents the region, i.e., the location of the truck charging station; Ψ represents the set of time points; F represents various characteristics of each region at each time point, including land type, traffic flow, charging occupancy rate, charging pricing, and weather conditions; spatiotemporal hypergraph is used for spatiotemporal modeling; hypergraph It consists of three parts:

[0041] ①Vertex Each vertex represents a region r at time point t;

[0042] ②Hyperedge ε={e1...e H}, each of which connects multiple vertices to reflect a multi-part region-like relationship (e.g., all charging stations can be connected by one hyperedge);

[0043] ③Vertex-hyperedge connection Where N represents the number of vertices and H is a trainable parameter.

[0044] In one example, in the stage of pre-training the spatiotemporal graph, a masked autoencoder MAE task is used as the training target. The goal of the MAE task is to reconstruct the masked information of the spatiotemporal data according to the unmasked spatiotemporal data by learning the spatiotemporal representation function f; for [t K-L+1, t K] The objective function for the spatiotemporal data trained during the training period is as follows:

[0045]

[0046] In the formula, represents a measure of prediction deviation, M∈{0,1} represents a mask tensor, ⊙ represents an element-wise multiplication operation, and W represents a prediction linear layer.

[0047] In one example, the preprocessing of the first data information to obtain the second data information may specifically include the following steps:

[0048] S210: Clean the first data information, fill in missing values, and identify and remove abnormal data.

[0049] S220. Merge different data as needed to form a unified data set and data format.

[0050] S230: Identify the time period, day of the week, and whether it is a holiday of the timestamp, and identify the spatial location of the logistics park, etc.

[0051] S240, performing sliding window processing on the time series data, visualizing the geographic location information, and dividing the data set into a training set, a validation set, and a test set.

[0052] S250: Check whether the processed data set is balanced and whether there is a data leakage problem, and save the processed data in a format acceptable to model training to obtain second data information.

[0053] In some embodiments, the spatiotemporal generative pre-trained model is also used for customized temporal pattern coding, hierarchical spatial pattern coding, and downstream tasks.

[0054] In the embodiment of the present application, it can be understood that the stage corresponding to the downstream task occurs after the pre-training stage, and the spatiotemporal representation generated by the pre-trained model is used to facilitate the downstream prediction task; the downstream task stage is represented as:

[0055]

[0056] Where ζ refers to the spatiotemporal features generated by f based on the first L timestamps before the Kth timestamp; the output is the prediction for the next P time intervals, and the DYffusion model is used as the prediction function g.

[0057] In some embodiments, the spatiotemporal prediction dynamic diffusion model performs charging demand prediction including problem definition, diffusion process and generation process; it can be understood that the present application fuses the spatiotemporal features generated by the spatiotemporal generative pre-training model with the real real-time traffic flow, real-time charging occupancy rate, real-time charging pricing and real-time weather conditions of new energy trucks; the fused features are used as the input of the spatiotemporal prediction dynamic diffusion model, and the spatiotemporal prediction dynamic diffusion model will use these features to predict charging demand;

[0058] Define the fusion dataset x t =Concat(ζ,η)

[0059] In the formula, ζ represents the spatiotemporal features generated by the pre-training model, η represents the real-time spatiotemporal features, Represents the space where the data is located, including spatial dimensions (latitude and longitude of charging stations) and channel dimensions (land type, traffic flow, charging occupancy rate, charging pricing, and weather conditions);

[0060] (1) Problem definition

[0061] The prediction of the charging demand of new energy trucks of logistics enterprises is defined as a spatiotemporal probability prediction problem, which predicts the charging demand of each charging station in a certain period of time in the future. The task of probability prediction is to learn the conditional distribution P(x t+1：t+h |x t-l+1：t ), that is, using information from the past l data points to predict the charging requirements of h data points;

[0062] (2) Diffusion process

[0063] The core idea of ​​the diffusion model is to "destroy" the data by gradually adding noise, and then restore the data through the denoising process; in the standard diffusion model, the forward diffusion process gradually degrades the data by gradually adding Gaussian noise, and the reverse diffusion process gradually removes the noise to reconstruct the original data through a denoising network. The network is trained to minimize the reconstruction error, thereby learning the generation process of the data; in the case of dynamic prediction, the diffusion model can be trained to minimize the prediction error:

[0064]

[0065] In the formula, is the data distribution, ‖·‖ is the norm, and s (0) =x t+1:t+h is the prediction target; Describes a uniform distribution on the set {a, a+1, …, b}; in practice, R θcan be trained to predict the Gaussian noise that has been added to data points using the score matching objective;

[0066] ① Forward process

[0067] Temporal interpolation is used as the forward process. To impose temporal bias, a time-conditioned network is trained to interpolate between data; specifically, given a baseline h, train such that For i ∈ {1, …, h - 1}, the objective is:

[0068]

[0069] ② Reverse process

[0070] In the second stage, the predictor network F θ is used to predict x t+h such that For where S represents the schedule that maps diffusion steps to interpolation time steps; the network is frozen, inference randomness is enabled, represented by the random variable, and ξ represents the randomly dropped weights of the neural network;

[0071]

[0072] (3) Generation process

[0073] To incorporate the settings where F θ learns to predict the initial conditions, define i0 := 0 and := x t ; In the simplest case, the predictor network is supervised by all possible time steps given by the temporal resolution of the training data, N = h and Typically, the interpolation time steps should satisfy 0 = i0 < i n < i m < h, where 0 < n < m ≤ N - 1; The equivalent role of the predictor and denoising networks in the diffusion model is called the diffusion backbone, and the temporal condition of the diffusion backbone is i n ; The generation process of DYffusion is as follows:

[0074]

[0075] where s (0) = x t and correspond to the predictions of the initial condition and intermediate steps, respectively; n = 0 is the start of the reverse process, and n = N is the output of the reverse process;

[0076] Let s be the time variable, and DYffusion models the dynamics as:

[0077]

[0078] The initial condition is x(t)=x t Given x, in the prediction process, the objective function is:

[0079]

[0080] The prediction process is equivalent to evaluating at discrete points in the prediction window [t, t+h].

[0081] In some embodiments, the aforementioned customized time mode encoding includes:

[0082] S310, normalize the spatiotemporal data, construct an embedding matrix, and use an embedding layer to convert discrete features into continuous embedding vectors.

[0083] S320: Initialize the embedding vector by random initialization or using a pre-trained model.

[0084] S330, periodically encode the time feature, positionally encode the space feature, and fuse the obtained periodic encoding result and position encoding result to obtain spatiotemporal embedding information.

[0085] S340. Based on the spatiotemporal embedding information, the parameters of the embedding layer are optimized during the model training process and fine-tuned according to the task requirements to obtain an effective representation that can capture the intrinsic characteristics and structure of the spatiotemporal data.

[0086] In an example, the specific steps of customizing the time pattern encoding are as follows:

[0087] ①Data preprocessing

[0088] The original data is normalized using the Z-Score function and then masked using the mask operation; the calculation formula of the Z-Score function is as follows:

[0089]

[0090] Where X is the original data point, μ is the mean of the data set, and σ is the standard deviation of the data set;

[0091] ②Construct the embedding matrix

[0092] Initialization of the embedding matrix E: Where V is the number of features and D is the dimension of embedding;

[0093] ③ Initialize the embedding vector

[0094] Randomly initialize the embedding vector: e i ~Uniform(-1,1), where ei is the embedding vector of the i-th feature;

[0095] ④Periodic encoding of time characteristics

[0096] Hypergraph neural network is used for temporal pattern encoding; for each vertex (time point), cyclic encoding is applied to capture its position in the time period; cyclic encoding usually uses sine and cosine functions to represent time points, and the formula is as follows:

[0097]

[0098] In the formula, e t Represents the code at time point t, where T is the cycle period.

[0099] In some embodiments, the aforementioned hierarchical spatial pattern encoding is performed based on a hypergraph capsule clustering network and includes the following steps:

[0100] S410, constructing a hypergraph structure by initializing region embedding and calculating transfer information from the region to the cluster center.

[0101] S420. Utilize the dynamic routing mechanism of the capsule network to iteratively update the hyperedge representation and the regional hyperedge connection to enhance the clustering capability.

[0102] S430. Modeling category embeddings through advanced hypergraph neural networks for cross-category relationship learning.

[0103] S440. Based on the category-aware mask strategy, the prediction difficulty is gradually increased to guide the model to learn robust spatiotemporal representation in the pre-training stage and realize spatial hierarchical pattern encoding from local to global.

[0104] S450, the temporal coding vector and the spatial coding vector as well as the embedding vector obtained by the hypergraph capsule clustering network are merged to form a comprehensive spatiotemporal representation.

[0105] In one example, the specific steps of hierarchical spatial pattern encoding are as follows:

[0106] ① Initialization area embedding

[0107] For each region r at time point t, the embedding is initialized using the following formula:

[0108] v r,t =squash(VΓ r,t +c)

[0109] In the formula, Γ r,t is the original feature embedding of region r at time t, V and c are learnable weight matrices and biases, and squash is a nonlinear activation function (such as softmax) used to ensure the scale of the embedding;

[0110] ②Calculation and transmission of information

[0111] Delivering information The calculation formula from each region t to the cluster center (hyperedge) i is:

[0112]

[0113] In the formula, H′ i,r,t represents the connection weight from region r to hyperedge i, represents the vector outer product;

[0114] ③Iterative Hypergraph Learning

[0115] Iteratively update the hyperedge representation s using the dynamic routing mechanism of the capsule network i,t and region-hyperedge connection c i,r,t :

[0116]

[0117] In the formula, is the embedding of hyperedge i at time t in the jth iteration, is the weight of the jth iteration, is the log probability;

[0118] ④Cross-category relationship learning

[0119] Category Embedding Using Advanced Hypergraph Neural Networks To model:

[0120]

[0121] Where H″ represents the high-level hypergraph structure, σ is the activation function;

[0122] ⑤Category-aware masking strategy

[0123] Design an adaptive masking strategy, combining the previously learned clustering information To gradually increase the difficulty of prediction:

[0124]

[0125] Where M r,t It is a mask matrix. Mask is used to selectively mask the information of the region in time according to the clustering information and the mask ratio mask_rate;

[0126] ⑥Integration of spatiotemporal features

[0127] The time encoding vector e t and the spatial encoding vector e sAnd the embedding vector v obtained by the hypergraph capsule clustering network r,t Combined together to form a comprehensive space-time representation:

[0128] e st =Concat(e t ,e s ,v r,t )

[0129] In the formula, Concat represents the concatenation of vectors.

[0130] In some embodiments, the solution algorithm idea of ​​the smart energy station site selection optimization model includes: combining the spatiotemporal distribution prediction results of charging demand, using the dung beetle optimization algorithm that integrates the black widow idea to optimize the site selection of the smart energy station; when the algorithm is initialized, a candidate solution population representing possible charging station locations is generated; in each generation of iteration, by imitating the behavior of the black widow, some poor candidate solutions are strategically abandoned, and the current solution is updated through the dung beetle rolling strategy to explore the solution space; the algorithm combines local search with global search, balancing exploration and development to achieve efficient search; through iterative updates and fitness evaluation, the algorithm gradually optimizes the population, and finally outputs the optimal or suboptimal charging station site selection plan after meeting the convergence criteria, completing the smart energy station site selection planning.

[0131] In an example, see Figure 5 , the solution algorithm steps of the smart energy station site selection optimization model can specifically include the following steps:

[0132] 1. Initialize the population

[0133] ICMIC is used as the final chaotic mapping initial population generation method. The specific expression of ICMIC mapping is:

[0134]

[0135] In the formula, α is the mapping coefficient. In order to further explore the influence of mapping parameters, the Lyapunov index (λ) is used to measure the mapping effect. The calculation formula is:

[0136]

[0137] 2. Update the position of dung beetles

[0138] (1) Random number < δ

[0139] The black widow algorithm is used to replace the rolling ball behavior algorithm. The black widow algorithm converts the motion model of the black widow spider in the spider web into two forms: linear and spiral. The position update formula is as follows:

[0140]

[0141] In the formula, x * is the optimal position, m is a random floating point number between 0.4 and 0.9, and β is a random floating point number between -1 and 1. for other individuals in the population;

[0142] Introducing the pheromone defined by the Black Widow algorithm:

[0143]

[0144] Where, fitness max is the global optimal individual fitness value, fitness min is the global worst individual fitness value, and fitness(i) is the current individual fitness value;

[0145] ①Pheromone≤0.1

[0146] When the pheromone is too small, the individual will be replaced by a new individual, and the position update formula is:

[0147]

[0148] In the formula, and are two different individuals in the population;

[0149] ②Pheromone>0.1

[0150] Go directly to the next step;

[0151] (2) Random number ≥ δ

[0152] The mathematical model of the dancing behavior of the rolling dung beetle is expressed as follows:

[0153] x i (t+1)=x i (t)+tan(θ)|x i (t)-x i (t-1)

[0154] Where, θ∈[0,π], when θ=0,π / 2,π, the position of the dung beetle will not be updated;

[0155] (3) Update the locations of egg-laying dung beetles, baby dung beetles, and stealing dung beetles

[0156] ① Egg-laying dung beetle

[0157] To mimic the best location for egg-laying dung beetles, a boundary selection strategy was adopted to simulate the area where female dung beetles lay eggs. The strategy is described as:

[0158] Lb *=max(X * *(1-R),Lb),

[0159] Ub * =min(X * *(1-R),Ub)

[0160] Where, X * As the best position in this round, Lb * , Ub * As the upper and lower bounds of the spawning area, Lb and Ub are the upper and lower bounds of the optimization algorithm; t is the current round, T max is the maximum number of iterations.

[0161] Once the egg-laying area is determined, the egg-laying dung beetle will lay an egg in the selected area. At the same time, it should be pointed out that according to the above formula, the boundary range of the egg-laying area is dynamically changing and will gradually become smaller; the mathematical model of the specific location of the egg-laying dung beetle is:

[0162] B i (t+1)=X * +b1*(B i (t)-Lb * )+b2*(B i (t)-Ub * )

[0163] In the formula, B i (t) is the position information of the i-th dung beetle at the t-th iteration, b1 and b2 are independent random vectors that conform to the dimension, and the egg-laying position will be strictly limited to the egg-laying area;

[0164] ② Small dung beetle

[0165] The t-distribution variation disturbance factor with the number of iterations as the freedom parameter was used to perturb the foraging behavior of the dung beetle. The probability density function of the t-distribution is:

[0166]

[0167] In the formula, It is the second type of Euler integral, the decision parameter is the degree of freedom parameter m, when m = 1, the t distribution is the Cauchy distribution; when m → ∞, the t distribution is the Gaussian distribution; the specific position update method is as follows:

[0168]

[0169] In the formula, t represents a random number under t distribution;

[0170] ③ Stealing cockroaches

[0171] Imitating the behavior pattern of stealing cockroaches, the mathematical model is expressed as:

[0172] x i (t+1)=X b +S*g*(|x i (t)-X * ∣+∣x i (t)-X b ∣)

[0173] In the formula, g is a D-dimensional random vector that obeys the normal distribution, and S is a constant;

[0174] f(x i ) → Fitness value

[0175] The fitness of each individual is evaluated according to the objective function f;

[0176] 3. Black Widow Pheromone Update

[0177] (1) Mobile

[0178] x i (t+1)=x i (t)+α·(x * (t)-x i (t))+β·Δx i (t)

[0179] Among them, x * (t) is the current best individual position, α and β are random coefficients, Δx i (t) indicates position change;

[0180] (2)Pheromone Update

[0181]

[0182] Among them, τ(x i ) is the pheromone concentration of the ith individual, δ is the pheromone evaporation rate;

[0183] 4. Determination of the position and boundary of dung beetles

[0184] The main purpose of performing boundary judgment on each dung beetle is to ensure that the position of the dung beetle in the search space does not exceed the predefined upper and lower bounds; the purpose of boundary judgment is to maintain the feasibility of the algorithm and ensure that the solution after each iteration is valid; for each dung beetle i at the position x of the tth iteration i (t), the boundary determination formula can be expressed as:

[0185]

[0186] In the formula, x i (t+1) is the position of dung beetle i after the (t+1)th iteration; Lbi is the upper bound of the position of dung beetle i in the i-th dimension; Ub i is the lower bound of the position of dung beetle i in the i-th dimension;

[0187] 5. Calculate the fitness and optimal solution of all dung beetles

[0188]

[0189] Where N is the number of dung beetles, and the position of each dung beetle is represented by x i , the global optimal position is represented by x * ,f(x i ) is the fitness of dung beetle i, f(x * ) is the global optimal fitness;

[0190] 6. Repeat

[0191] Determine whether t is less than the maximum number of iterations. If so, return to the random number judgment, otherwise the iteration ends.

[0192] In some embodiments, the smart energy station site selection optimization model includes an objective function and objective constraints corresponding to the objective function; the objective constraints include energy demand constraints, technical parameter constraints, environmental impact constraints, grid access and stability constraints, land and geographical location constraints, return on investment constraints, and user satisfaction constraints.

[0193] In one example, the objective function satisfies the expression:

[0194] minZ=C opex +C grid +C users +C env

[0195] Where Z is the total cost to be minimized, C opex is the charging aggregator cost, C grid is the grid stabilization cost, C users is the user cost, C env It is the environmental cost;

[0196] The energy demand constraint satisfies the expression:

[0197]

[0198] In the formula, E i is the supply of the i-th energy module and includes photovoltaic power generation and energy storage release; D i is the demand of the i-th energy module, E total is the total forecast energy demand;

[0199] Technical parameter constraints satisfy the expression:

[0200] P min ≤P i ≤P max

[0201] Where P i is the output power of the ith energy module, P min and P max are the minimum and maximum output powers of the module respectively;

[0202] Environmental impact constraint satisfaction expression:

[0203] E enb,i ≤T env

[0204] In the formula, E env,i is the environmental impact of the ith energy module, T env It is the maximum permissible value of environmental impact;

[0205] The grid access and stability constraints satisfy the expression:

[0206] S grid,i ≤C grid

[0207] In the formula, S grid,i is the contribution of the ith energy module to grid stability, C grid is the maximum permissible value for grid stability;

[0208] The land and geographic location constraints satisfy the expression:

[0209] A i ≥A req,i

[0210] In the formula, A i is the land area available for the i-th energy station, A req,i is the minimum land area required for the energy station;

[0211] The return on investment constraint satisfies the expression:

[0212] ROI ≥ T ROI

[0213] Where ROI is the return on investment, T ROI is the target return on investment;

[0214] User satisfaction constraint satisfies the expression:

[0215] S user,i ≥T user

[0216] In the formula, S user,iis the user satisfaction of the i-th energy station, T user is the minimum acceptable value of user satisfaction.

[0217] In some embodiments, please refer to Figure 4 The smart energy station site selection optimization method also includes building a smart energy station system, which includes an energy supply module, a smart energy station optimization scheduling module and a new energy truck support module; the smart energy station optimization scheduling module includes the following constraints: supply and demand balance constraints, energy flow stability constraints, capacity limitation constraints, equipment status constraints, equipment maintenance constraints and cost optimization constraints; among them, the supply and demand balance constraints are used to ensure that at any given time, the energy supply is equal to the demand, while taking into account the charging and discharging status of the energy storage system.

[0218] In an example, the supply and demand balance constraint satisfies the expression:

[0219]

[0220] Where P i (t) is the output power of supply source i at time t, P stocha (t) and P stodis (t) are the charging and discharging power of the energy storage system at time t, P j (t) is the power demand of load j at time t;

[0221] The energy flow stability constraint is used to ensure that the energy flow is within a predetermined safety range and avoid violent fluctuations; the energy flow stability constraint satisfies the expression:

[0222]

[0223] Where P min and P max They are the minimum and maximum output power of the supply source or load, sto sys represents the energy storage system number library, and load represents the load system number library;

[0224] The capacity limit constraint is used to ensure that the operation of all devices and energy storage systems does not exceed their capacity limits; the capacity limit constraint satisfies the expression:

[0225]

[0226] In the formula, E j (t) is the energy storage level of energy storage system j at time t, E min,j and E max,j are the minimum and maximum energy storage capacities of energy storage system j, respectively;

[0227] The device status constraint is used to consider the device operating status and ensure that all devices are in operation or standby state; the device status constraint satisfies the expression:

[0228]

[0229] In the formula, S i (t) is the state of equipment i at time t, RUN indicates the running state, SBY indicates the standby state, MNT indicates the maintenance state, FLT indicates the fault state, and EQU indicates the equipment number library;

[0230] Equipment maintenance constraints are used to consider regular equipment maintenance to avoid failures and extend service life; equipment maintenance constraints satisfy the expression:

[0231]

[0232] Where, T mnt,i is the recommended time interval between two maintenance of equipment i, T nmnt,i is the time of next maintenance of equipment i, T max,i is the maximum allowed time interval between two maintenances of equipment i;

[0233] The cost optimization constraint considers optimizing the scheduling scheme to minimize the overall operating cost, which includes energy purchase cost, equipment operation and maintenance cost; the cost optimization constraint satisfies the expression:

[0234]

[0235] In the formula, C RUN, i(P i (t)) is the operating cost of equipment i at time t, C MNT, i(S i (t)) is the maintenance cost of equipment i, C BUY (P BUY i(t)) is the cost of purchasing energy from the outside for device i.

[0236] In some embodiments, the present application provides a smart energy station site selection optimization device 500, such as Figure 6 As shown, the device 500 may include the following modules:

[0237] A preprocessing module 510 is used to obtain first data information of new energy trucks in a historical period, and preprocess the first data information to obtain second data information;

[0238] An analysis and determination module 520 is used to analyze the operation mode of the new energy truck based on the second data information, and determine the main operation area and potential growth area of ​​the new energy truck; wherein the operation mode includes the driving route, the type of goods and the transportation frequency;

[0239] A spatiotemporal graph construction module 530 is used to construct a spatiotemporal graph to characterize the charging behavior of the truck by using the spatiotemporal masked autoencoder in the spatiotemporal generative pre-trained model; wherein the spatiotemporal graph uses the truck and the charging station as nodes, and uses the spatiotemporal relationship between the truck and the charging station as an edge;

[0240] A pre-training module 540 is used to pre-train the spatiotemporal graph using a spatiotemporal generative pre-training model, learn the intrinsic spatiotemporal pattern of the truck charging behavior through an adaptive masking strategy, and obtain a spatiotemporal representation;

[0241] A learning prediction module 550 is used to take the spatiotemporal representation as input information and train a spatiotemporal prediction dynamic diffusion model to learn the change of charging demand from the current state to the future state to perform charging demand prediction;

[0242] The model building module 560 is used to predict the charging demand based on the spatiotemporal prediction dynamic diffusion model, and to build a smart energy station site selection optimization model with the lowest comprehensive cost as the objective function to optimize the site selection of the smart energy station; wherein the smart energy station site selection optimization model is a multi-objective optimization model integrating hydrogen refueling, charging, photovoltaics and energy storage.

[0243] According to an embodiment of the present application, any multiple modules among the preprocessing module 510, the analysis and determination module 520, the spatiotemporal graph construction module 530, the pretraining module 540, the learning prediction module 550, and the model construction module 560 can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module.

[0244] Figure 6 Each module in the device shown has the function of implementing each step in the aforementioned smart energy station site selection optimization method and can achieve its corresponding technical effect. For the sake of concise description, it will not be repeated here.

[0245] In some embodiments, the present application provides an electronic device, the structure diagram of the electronic device is as follows Figure 7 shown.

[0246] The electronic device may include a processor 610 and a memory 620 storing computer program instructions.

[0247] Specifically, the processor 610 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0248] The memory 620 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 620 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 620 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 620 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 620 is a non-volatile solid-state memory.

[0249] The memory 620 may include a read-only memory (ROM), a random access memory (RAM), a disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Therefore, generally, the memory 620 includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it can perform the operations described in any of the smart energy station site optimization methods in the above embodiments.

[0250] The processor 610 implements any one of the smart energy station site selection optimization methods in the above embodiments by reading and executing computer program instructions stored in the memory 620.

[0251] In one example, the electronic device may further include a communication interface 630 and a bus 600. Figure 7 As shown, the processor 610, the memory 620, and the communication interface 630 are connected via a bus 600 and communicate with each other.

[0252] The communication interface 630 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0253] Bus 600 includes hardware, software or both, and the parts of online data flow billing equipment are coupled to each other. For example, but not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front-end bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In appropriate cases, bus 600 may include one or more buses. Although the present application embodiment describes and shows a specific bus, the present application considers any suitable bus or interconnection.

[0254] In addition, in combination with the smart energy station site selection optimization method in the above embodiment, the embodiment of the present application can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by the processor, any one of the smart energy station site selection optimization methods in the above embodiment is implemented.

[0255] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0256] The functional blocks shown in the above block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0257] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.

[0258] Aspects of the present disclosure are described above with reference to the flowchart and / or block diagram of the method, device (system) and computer program product according to the embodiment of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0259] In summary, compared with the prior art, this application has the following beneficial effects:

[0260] 1. Through precise data analysis and optimization algorithms, the present invention can determine the optimal energy station location to meet the charging needs of new energy trucks of logistics companies. This method not only takes into account the efficiency and cost of energy replenishment, but also takes into account environmental protection and grid stability. Compared with the prior art, the present invention can more accurately predict charging needs, thereby reducing the over-construction or under-construction of energy stations, avoiding waste of resources, improving the overall efficiency of energy replenishment, and reducing construction and operating costs.

[0261] 2. The present invention pays more attention to environmental protection in site selection decisions. By introducing photovoltaic power generation and energy storage equipment, it optimizes the energy structure and improves energy utilization efficiency. This site selection optimization that comprehensively considers multiple energy replenishment methods such as charging and hydrogenation helps to reduce the impact on the environment and promote the logistics industry to develop in a greener and smarter direction. Compared with the background technology, the present invention achieves comprehensive optimization of the site selection plan through a multi-objective optimization model, which not only improves the layout efficiency of new energy truck charging facilities, but also promotes the green and sustainable development of the logistics industry.

[0262] 3. When constructing the smart energy station site selection optimization model, the present invention takes into account the grid access conditions and stability. By optimizing the layout of energy supply stations, it helps to reduce the impact on the power grid and improve the stability of the power grid. In addition, by predicting the charging needs of new energy trucks and combining real-time data, the present invention can achieve dynamic management and optimal allocation of energy and improve energy utilization efficiency. Compared with the prior art, the present invention completes the optimal site selection of smart energy stations by integrating the dung beetle optimization algorithm with the black widow idea, while improving the contribution to the stability of the power grid.

[0263] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing the site selection of a smart energy station, characterized in that: include: Acquire first data information of new energy trucks in a historical period, and pre-process the first data information to obtain second data information; Analyze the operation mode of new energy trucks based on the second data information, and determine the main operation areas and potential growth areas of new energy trucks; wherein the operation mode includes driving routes, cargo types and transportation frequencies; A spatiotemporal graph is constructed by a spatiotemporal masked autoencoder in a spatiotemporal generative pre-trained model to characterize the charging behavior of the truck; wherein the spatiotemporal graph uses the truck and the charging station as nodes, and uses the spatiotemporal relationship between the truck and the charging station as an edge; Pre-training the spatiotemporal graph using the spatiotemporal generative pre-training model, learning the intrinsic spatiotemporal pattern of the truck charging behavior through an adaptive masking strategy, and obtaining a spatiotemporal representation; Taking the spatiotemporal representation as input information, training a spatiotemporal prediction dynamic diffusion model to learn the change of charging demand from the current state to the future state to perform charging demand prediction; Based on the charging demand prediction of the spatiotemporal prediction dynamic diffusion model, a smart energy station site selection optimization model is constructed with the lowest comprehensive cost as the objective function to optimize the site selection of the smart energy station; wherein the smart energy station site selection optimization model is a multi-objective optimization model integrating hydrogen refueling, charging, photovoltaics and energy storage.

2. The smart energy station site selection optimization method according to claim 1, characterized in that: The preprocessing of the first data information to obtain the second data information includes: Cleaning the first data information, filling missing values, and identifying and removing abnormal data; Merge different data into a unified data set and data format as needed; Identify the time period, day of the week, and whether it is a holiday, and identify the spatial location of the logistics park, etc. Perform sliding window processing on time series data, visualize geographic location information, and divide the data set into training set, validation set, and test set; Check whether the processed data set is balanced and whether there is any data leakage problem, and save the processed data in a format acceptable to model training to obtain second data information.

3. The smart energy station site selection optimization method according to claim 1, characterized in that: The spatiotemporal prediction dynamic diffusion model for predicting charging demand includes defining problems, diffusion processes and generation processes; The spatiotemporal generative pre-trained model is also used for customized temporal pattern coding, hierarchical spatial pattern coding and downstream tasks; wherein the customized temporal pattern coding includes: Normalize the spatiotemporal data and construct an embedding matrix, using an embedding layer to convert discrete features into continuous embedding vectors; Initializing the embedding vector by random initialization or using a pre-trained model; Perform periodic encoding on the time feature and position encoding on the space feature, and fuse the periodic encoding result and position encoding result to obtain the spatiotemporal embedding information; Based on the spatiotemporal embedding information, the parameters of the embedding layer are optimized during the model training process and fine-tuned according to the task requirements to obtain an effective representation that can capture the intrinsic characteristics and structure of the spatiotemporal data.

4. The method for optimizing the site selection of a smart energy station according to claim 3, characterized in that: The hierarchical spatial pattern coding is performed based on a hypergraph capsule clustering network and includes the following steps: The hypergraph structure is constructed by initializing region embedding and calculating the transfer information from the region to the cluster center; The dynamic routing mechanism of capsule network is used to iteratively update hyperedge representation and regional hyperedge connection to enhance clustering ability; Modeling category embeddings through advanced hypergraph neural networks for cross-category relationship learning; Gradually increase the prediction difficulty based on the category-aware mask strategy to guide the model to learn robust spatiotemporal representation in the pre-training stage and achieve spatial hierarchical pattern encoding from local to global; The temporal encoding vector and the spatial encoding vector as well as the embedding vector obtained by the hypergraph capsule clustering network are fused to form a comprehensive spatiotemporal representation.

5. The smart energy station site selection optimization method according to claim 1, characterized in that: The smart energy station site selection optimization model includes an objective function and objective constraints corresponding to the objective function; the objective constraints include energy demand constraints, technical parameter constraints, environmental impact constraints, grid access and stability constraints, land and geographical location constraints, return on investment constraints and user satisfaction constraints.

6. The method for optimizing the site selection of a smart energy station according to claim 5, characterized in that: The objective function satisfies the expression: minZ=C opex +C grid +C users +C env Where Z is the total cost to be minimized, C opex is the charging aggregator cost, C grid is the grid stabilization cost, C users is the user cost, C env It is the environmental cost; The energy demand constraint satisfies the expression: In the formula, E i is the supply of the i-th energy module and includes photovoltaic power generation and energy storage release; D i is the demand of the i-th energy module, E total is the total forecast energy demand; The technical parameter constraints satisfy the expression: P min ≤P i ≤P max Where P i is the output power of the ith energy module, P min and P max are the minimum and maximum output powers of the module respectively; The environmental impact constraint satisfies the expression: AND env,i ≤T env In the formula, E env,i is the environmental impact of the ith energy module, T env It is the maximum permissible value of environmental impact; The grid access and stability constraints satisfy the expression: S grid,i ≤C grid In the formula, S grid,i is the contribution of the ith energy module to grid stability, C grid is the maximum permissible value for grid stability; The land and geographic location constraints satisfy the expression: A i ≥A req,i In the formula, A i is the land area available for the i-th energy station, A req,i is the minimum land area required for the energy station; The ROI constraint satisfies the expression: ROI≥T ROI Where ROI is the return on investment, T ROi is the target return on investment; The user satisfaction constraint satisfies the expression: S user,i ≥T user In the formula, S user,i is the user satisfaction of the i-th energy station, T user is the minimum acceptable value of user satisfaction.

7. The method for optimizing the site selection of a smart energy station according to claim 1, characterized in that: It also includes building a smart energy station system, which includes an energy supply module, a smart energy station optimization scheduling module and a new energy truck support module; The smart energy station optimization scheduling module includes the following constraints: supply and demand balance constraint, energy flow stability constraint, capacity limit constraint, equipment status constraint, equipment maintenance constraint and cost optimization constraint; wherein, the supply and demand balance constraint is used to ensure that at any given time, the energy supply is equal to the demand, while considering the charging and discharging state of the energy storage system; the supply and demand balance constraint satisfies the expression: Where P i (t) is the output power of supply source i at time t, P stocha (t) and P stodis (t) are the charging and discharging power of the energy storage system at time t, P j (t) is the power demand of load j at time t; The energy flow stability constraint is used to ensure that the energy flow is within a predetermined safety range and avoid violent fluctuations; the energy flow stability constraint satisfies the expression: Where P min and P max They are the minimum and maximum output power of the supply source or load, sto sys represents the energy storage system number library, and load represents the load system number library; The capacity limit constraint is used to ensure that the operation of all devices and energy storage systems does not exceed their capacity limits; the capacity limit constraint satisfies the expression: In the formula, E j (t) is the energy storage level of energy storage system j at time t, E min,j and E max,j are the minimum and maximum energy storage capacities of energy storage system j, respectively; The device status constraint is used to consider the device operating status and ensure that all devices are in operation or standby state; the device status constraint satisfies the expression: In the formula, S i (t) is the state of equipment i at time t, RUN indicates the running state, SBY indicates the standby state, MNT indicates the maintenance state, FLT indicates the fault state, and EQU indicates the equipment number library; The equipment maintenance constraint is used to consider regular equipment maintenance to avoid failures and extend service life; the equipment maintenance constraint satisfies the expression: Where, T mnt,i is the recommended time interval between two maintenance of equipment i, T nmnt, i is the time of next maintenance of equipment i, T max, i is the maximum allowed time interval between two maintenances of equipment i; The cost optimization constraint considers optimizing the scheduling scheme to minimize the overall operating cost, which includes energy purchase cost, equipment operation and maintenance cost; the cost optimization constraint satisfies the expression: In the formula, C RUN, i(P i (t)) is the operating cost of equipment i at time t, C MNT, i(S i (t)) is the maintenance cost of equipment i, C BUY (P BUY i(t)) is the cost of energy purchased from outside by device i.

8. A smart energy station site selection optimization device, characterized in that: include: A preprocessing module, used to obtain first data information of new energy trucks in a historical period, and preprocess the first data information to obtain second data information; An analysis and determination module, configured to analyze the operation mode of new energy trucks based on the second data information, and determine the main operation areas and potential growth areas of new energy trucks; wherein the operation mode includes driving routes, cargo types, and transportation frequencies; A spatiotemporal graph construction module, used to construct a spatiotemporal graph to characterize the charging behavior of the truck through the spatiotemporal masked autoencoder in the spatiotemporal generative pre-trained model; wherein the spatiotemporal graph uses the truck and the charging station as nodes, and uses the spatiotemporal relationship between the truck and the charging station as an edge; A pre-training module, configured to pre-train the spatiotemporal graph using the spatiotemporal generative pre-training model, learn the intrinsic spatiotemporal pattern of the truck charging behavior through an adaptive masking strategy, and obtain a spatiotemporal representation; A learning prediction module, used for taking the spatiotemporal representation as input information, training a spatiotemporal prediction dynamic diffusion model to learn the change of charging demand from the current state to the future state to perform charging demand prediction; The model building module is used to predict the charging demand based on the spatiotemporal prediction dynamic diffusion model, take the lowest comprehensive cost as the objective function, and build a smart energy station site selection optimization model to optimize the site selection of the smart energy station; wherein the smart energy station site selection optimization model is a multi-objective optimization model integrating hydrogen refueling, charging, photovoltaics and energy storage.

9. An electronic device, characterized in that: include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein when the program is executed by the processor, the method for optimizing the site selection of a smart energy station as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores programs or instructions, and when the programs or instructions are executed by the processor, the smart energy station site selection optimization method as described in any one of claims 1 to 7 is implemented.

Images