Ordered charging load short-term prediction method and system based on vehicle network interaction, and medium

Abstract

The invention belongs to the field of charging station charging load prediction, and provides an ordered charging load short-term prediction method and system based on vehicle network interaction, and a medium. The orderly charging load short-term prediction method based on vehicle network interaction comprises the steps of obtaining charging station charging loads and corresponding vehicle identification marks in a historical set time period, and dividing the charging station charging loads into bus charging loads and non-bus charging loads according to the vehicle identification marks; the historical bus charging loads are clustered according to the date types, and the historical bus charging loads of each date type are subjected to bus charging load prediction through a bus charging load prediction model based on similar days; the historical non-bus charging load is subjected to variational mode decomposition, a plurality of mode components are obtained, and a non-bus charging load prediction value is obtained after mapping of a pre-trained non-bus charging load prediction model. The method can improve the accuracy of the short-term prediction result of the ordered charging load based on vehicle-network interaction, so as to accurately guide the actual distribution decision of the charging load of the charging station.

Description

Technical Field

[0001] This invention belongs to the field of charging load prediction for charging stations, and particularly relates to a method, system and medium for predicting orderly charging load based on vehicle-grid interaction. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Accurate short-term load forecasting for charging stations is crucial for ensuring the safe, efficient, and stable operation of the power grid, and is also an important prerequisite for the multi-energy coordinated operation of the charging station energy management system. Current charging station load forecasting algorithms can be divided into model-driven methods based on physical mechanisms and data-driven methods based on statistical learning. While model-driven methods based on physical mechanisms do not require a large amount of historical data, the accuracy of the modeling directly affects the accuracy of the final prediction. For charging entities with strong randomness and large fluctuations, low modeling accuracy leads to large prediction errors. Data-driven methods based on statistical learning are more adaptable to complex nonlinear scenarios, are more flexible, and their prediction accuracy increases over time. However, they require a large amount of sample data, and their performance depends on the quality and scale of the data.

[0004] In summary, current charging station load forecasting algorithms are not suitable for charging entities with strong randomness and large fluctuations. Furthermore, due to insufficient sample data, the short-term forecasting accuracy of orderly charging load based on vehicle-grid interaction is poor, affecting the actual allocation decision of charging station charging load. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method, system, and medium for predicting short-term orderly charging load based on vehicle-to-grid interaction. This method can improve the accuracy of short-term orderly charging load prediction results based on vehicle-to-grid interaction, thereby accurately guiding the actual allocation decision of charging load at charging stations.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: The first aspect of the present invention provides a method for predicting short-term charging load in an orderly manner based on vehicle-to-grid interaction.

[0007] In one or more embodiments, a method for predicting short-term charging load based on vehicle-to-grid interaction is provided, including: Obtain the charging load of charging stations and their corresponding vehicle identification numbers within a set historical time period. Based on the vehicle identification numbers, divide the charging load of charging stations into bus charging load and non-bus charging load. Historical bus charging loads are clustered according to date type. The historical bus charging load for each date type is predicted using a bus charging load prediction model based on similar days. The date types include weekdays, weekends, and public holidays. Variational mode decomposition is performed on historical non-bus charging load to obtain several mode components. These mode components are combined with time-of-use electricity price, weather type, ambient temperature, and holiday type data to form input vectors. After being mapped by a pre-trained non-bus charging load prediction model, the predicted value of non-bus charging load is obtained.

[0008] A second aspect of the present invention provides an orderly charging load short-term prediction system based on vehicle-to-grid interaction.

[0009] In one or more embodiments, an ordered charging load short-term prediction system based on vehicle-to-grid interaction includes: The charging load division module is used to obtain the charging load of the charging station and its corresponding vehicle identification mark within a historical set time period, and divide the charging load of the charging station into bus charging load and non-bus charging load according to the vehicle identification mark. The bus charging load prediction module is used to cluster historical bus charging loads according to date type. The historical bus charging load for each date type is predicted using a bus charging load prediction model based on similar days. The date types include weekdays, weekends, and public holidays. The non-bus charging load prediction module is used to perform variational mode decomposition on historical non-bus charging loads to obtain several mode components. These mode components are combined with time-of-use electricity prices, weather types, ambient temperatures, and holiday types to form input vectors. After being mapped by a pre-trained non-bus charging load prediction model, the predicted non-bus charging load value is obtained.

[0010] A third aspect of the present invention provides an electronic device.

[0011] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the above-described method for orderly short-term charging load prediction based on vehicle-to-grid interaction.

[0012] A fourth aspect of the present invention provides a computer storage medium.

[0013] A computer storage medium having a computer program stored thereon, the program being configured to perform the steps of the orderly short-term charging load forecasting method based on vehicle-to-grid interaction as described above.

[0014] A fifth aspect of the present invention provides a computer storage medium.

[0015] A computer program product includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the above-described method for orderly short-term charging load prediction based on vehicle-to-grid interaction.

[0016] Compared with the prior art, the beneficial effects of the present invention are: This invention targets integrated charging stations, considering the differences in characteristics among various charging entities. It leverages the uncertainty and volatility of non-bus charging behavior by mapping historical non-bus charging loads through variational mode decomposition to pre-trained non-bus charging load prediction models, thus predicting the actual non-bus charging load. Since electric buses have clearly defined working hours and routes, are less affected by external environments, and exhibit distinct driving and charging patterns, this invention further improves the accuracy of short-term charging load predictions by dividing historical charging load data into weekdays, weekends, and holidays, and using data from similar days for charging load prediction. This achieves the goal of accurately guiding actual charging load allocation decisions at charging stations. Attached Figure Description

[0017] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0018] Figure 1 This is a flowchart illustrating the method for predicting short-term charging load based on vehicle-to-grid interaction according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of the orderly charging load short-term prediction system based on vehicle-to-grid interaction according to an embodiment of the present invention; Figure 3 This is a schematic diagram of an electronic device according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the principle of orderly charging load short-term prediction based on vehicle-to-grid interaction in an embodiment of the present invention; Figure 5 This is a schematic diagram of the SSA algorithm optimization according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the optimization of the non-bus charging load prediction model according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the LSTM model according to an embodiment of the present invention; Figure 8 This is a schematic diagram of a bus charging load prediction model based on similar days according to an embodiment of the present invention. Figure 9 This is a waveform diagram of the load forecast value according to an embodiment of the present invention; Figure 10 This is the historical charging load curve of an embodiment of the present invention. Detailed Implementation

[0019] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0020] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0021] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0022] Model-driven methods based on physical mechanisms are mostly based on modeling user travel characteristics, while taking into account multiple influencing factors such as traffic and road networks.

[0023] Data-driven methods based on statistical learning make predictions based on historical statistical data. They build models by mining historical data, first preprocessing the data to ensure its quality and validity, then using algorithms such as LSTM to learn time-series patterns, and finally using the trained model to predict the load and generate short-term charging load forecast data.

[0024] Figure 1 This is a flowchart illustrating a method for predicting short-term charging load based on vehicle-to-grid interaction according to an embodiment of the present invention. Figure 1 and Figure 4 The method for predicting short-term charging load based on vehicle-to-grid interaction in this embodiment may include the following steps S101 to S103.

[0025] The specific implementation process of steps S101 to S103 is as follows: Step S101: Obtain the charging load of the charging station and its corresponding vehicle identification number within a historical set time period. Based on the vehicle identification number, divide the charging load of the charging station into bus charging load and non-bus charging load.

[0026] The charging station control system collects electric vehicle charging load data in real time through IoT sensors, generating a complete dataset daily containing 96 time-series data points. Each data point represents the average load value within a 15-minute interval, forming a complete daily load curve dataset, providing basic data support for subsequent load analysis and forecasting. The information on buses in the integrated charging station is relatively fixed, and license plate numbers and VINs are recorded in the station control system. Based on the VIN code information during charging, it is easy to distinguish between bus charging load and non-bus charging load. During data collection, the charging load data is smoothed using a moving average method to clean up abrupt data and eliminate noise interference. The moving average method calculates the average of continuous subsequences of time-series data using a fixed window size, which can smooth short-term fluctuations and highlight trend changes. The moving average formula is expressed as:

[0027] In the formula, Let be the moving average at time t. Here are the original data points, and n is the sliding coefficient, with a value of n=3.

[0028] Before performing data prediction modeling, historical datasets must be systematically preprocessed. This crucial step includes core processes such as data cleaning (handling missing and outlier values), feature engineering (variable transformation and feature selection), data standardization, and training / test set partitioning, ensuring that the raw data is transformed into high-quality input suitable for machine learning algorithms.

[0029] Data cleaning, including handling missing and outlier values, is crucial. The prediction model dataset constructed in this invention contains multi-dimensional feature variables, integrating not only charging load time-series data but also time-of-use electricity pricing policy data, meteorological data, and holiday types. By establishing a unified timestamp alignment mechanism and outlier handling process, standardized cleaning of multi-source heterogeneous data is achieved, providing high-quality training samples for the load prediction model.

[0030] The data cleaning objectives of this embodiment are large-scale data gaps and anomalies, and localized data gaps and jumps. The data to be cleaned is daily time-series data at 96 points. Data cleaning steps: 1) If there are more than 3 consecutive missing data points in the daily time series data, the time series data will be deleted directly, and other feature data of the same date will be deleted simultaneously and will no longer be used as sample data.

[0031] 2) If the total number of missing data points in a daily time series reaches 6 or more, the time series data will be deleted directly, and other feature data of the same date will be deleted simultaneously and will no longer be used as sample data.

[0032] 3) Data showing continuous zero load during periods of low electricity prices is considered abnormal and will no longer be used as sample data.

[0033] 4) If the daily time series data shows that the load value remains unchanged for more than 3 consecutive points, it is judged as abnormal data and will no longer be used as sample data.

[0034] 4) When the load fluctuation exceeds 3 times the standard deviation, it is judged as abnormal data and data repair is performed.

[0035] 5) For missing or abrupt data points, the moving average method is used for data repair, as shown in the following formula:

[0036] In the formula Let be the moving average at time t, and k be the window radius, which is set to k=3. The data are the neighboring points on both sides at time t.

[0037] Data normalization processing is used in this embodiment to build a prediction model using historical charging load, time-of-use electricity price, weather type, ambient temperature, and holiday type data. Parameter normalization can unify the numerical range of different parameters, avoid model bias due to differences in units, accelerate algorithm convergence, reduce gradient descent oscillations, and enhance numerical stability. Min-Max normalization, also known as linear function normalization, does not change the original data distribution but only performs a linear transformation on the original data, mapping it to the range of 0 to 1. The calculation formula is as follows:

[0038] In the formula, The data is after normalization. The original input data, These are the maximum and minimum values ​​of the original data.

[0039] Step S102: Cluster the historical bus charging load according to the date type. The historical bus charging load of each date type is predicted using a bus charging load prediction model based on similar days. The date types include three types: weekdays, weekends, and holidays.

[0040] Electric buses have clearly defined operating hours and routes, making them relatively stable in both time and space. Charging times are regular and less affected by weather. Bus schedules exhibit the following characteristics: During weekday morning and evening rush hours, departure frequency increases and intervals shorten; some routes even have earlier first departures to meet commuting needs. On weekends, some routes switch to scheduled services with longer intervals, and while the first and last departure times are the same as on weekdays, the number of trips is reduced. Holiday express lines are added, and some routes operate only on weekends and holidays, suspending service on weekdays. Based on these different date type characteristics, this invention employs LSTM-based electric bus charging load prediction using similar day data. Dates are categorized into three types: weekdays, weekends, and holidays. Electric vehicle charging load data is classified according to date type, and then an LSTM model is used to train the model on the data from different categories.

[0041] like Figure 7 As shown, the internal structure of an LSTM neuron consists of four core components working together: cell state, input gate, forget gate, and output gate. Selective memorization and forgetting are achieved through a gating mechanism. The cell state level runs through the core channels of the neuron, responsible for the transmission and storage of long-term information, essentially acting as the network's memory backbone. Gating units regulate the addition and deletion of information, avoiding the gradient vanishing problem of traditional RNNs.

[0042] There are various prediction models for time series data. Different models can be selected depending on the object of prediction. For example, the Transformer model can also be used for bus load prediction. It captures long-distance dependencies through a self-attention mechanism and has more advantages when analyzing long-term historical data.

[0043] This patent, based on the regularity of bus routes and time patterns and combined with date type, eliminates minor factors such as weather and environment, and uses an LSTM model to predict the next day's bus charging load, resulting in higher accuracy in the predicted load data. LSTM is an improved version of RNN, solving the problem that RNN cannot handle long sequences due to gradient vanishing / exploding. LSTM selectively retains or forgets information through a gating mechanism, effectively capturing long-term dependencies in the sequence. Then, separate modeling is performed according to date type, which can effectively improve the accuracy of daily charging load prediction for electric buses.

[0044] The LSTM modeling process is as follows: 1) Input data, normalize the feature data, and format it as follows: , as input data for the model.

[0045] 2) The cell state at the previous moment was The hidden state at the previous moment was The input data at the current moment is The current cell state is The hidden state output at the current moment is .

[0046] 3) The forget gate is implemented using the sigmoid function, which receives the current input. Hidden state from the previous moment After activation by the sigmoid function, a decay coefficient is generated, outputting a value between 0 and 1. This determines the cell state. Which information should be retained or discarded? The closer the value is to 0, the more information is forgotten. Forget Gate The formula is as follows:

[0047] In the formula, For the Gate of Oblivion It is the sigmoid activation function. and This is the weight matrix. This is the bias vector.

[0048] 4) The input gate updates the cell state. The input gate formula is as follows:

[0049]

[0050] In the formula, Here, tanh is the input gate, and tanh is the hyperbolic tangent activation function. , , , This is the weight matrix. This is the bias vector.

[0051] 5) The output gate determines the value of the next hidden state. The output gate formula is as follows:

[0052]

[0053]

[0054] In the formula, For output gate, is the sigmoid activation function, and tanh is the hyperbolic tangent activation function. , This is the weight matrix. This is the bias vector.

[0055] The training of an LSTM neural network employs the backpropagation algorithm, propagating the error term along both the previous layer and the time direction. The gradient of the loss function with respect to the weights is calculated using the chain rule. The parameters are first input into the neural network, and the output of each memory block is then calculated. , , , Then calculate the error term for each node. The gradients of the weights along the neuron and time directions are calculated; finally, the weights of the network are updated using gradient descent.

[0056] like Figure 5 As shown, the SSA algorithm performs well when solving problems where the optimal solution is near the origin, but its performance degrades when the optimal solution is far from the origin. Furthermore, because individuals in the SSA algorithm converge to the current optimal solution by jumping directly to its vicinity, rather than moving towards the optimal solution as in other classic algorithms, the SSA algorithm is prone to getting trapped in local optima and has weak global search capabilities. Based on these shortcomings, the position update formulas for the discoverer, followers, and watchdogs were optimized.

[0057] 1) Discoverer location updated; the discoverer explores new areas.

[0058]

[0059] In the formula, This indicates the position of the i-th sparrow in the j-th generation of the population. Let Q be a uniformly distributed random number in the range [0,1], and ST be a normally distributed random number. ST is the warning value. That is, the strategy for the discoverer to update the location is changed to multiplying by a normally distributed random number (1+Q) with a mean of 1 and a variance of 1, or adding a standard normally distributed random number Q.

[0060] 2) Follower position update: If a follower's fitness is better than the discoverer's, it will take the discoverer's position; otherwise, the follower will forage randomly.

[0061]

[0062] In the formula, This represents the position of the i-th sparrow in the j-th generation of the population, where Q is a normally distributed random number. This represents the worst position of a sparrow in the population at the (t-1)th iteration. This represents the optimal position of the sparrows in the population during the (t-1)th iteration.

[0063] 3) Guardian Location Update: While sparrows are foraging, some individuals will act as guards. When danger approaches, they will abandon their current food, regardless of whether the sparrow is the discoverer or a follower, and move to a new location. Each iteration randomly selects 20% of individuals to perform this warning behavior.

[0064]

[0065] In the formula, The random disturbance factor is a standard normally distributed random number. This represents the worst position during the (t-1)th iteration. This represents the optimal position at the (t-1)th iteration. That is, if an individual is in the optimal position, it will move to a random position between the optimal and worst positions; otherwise, it will move to a random position between itself and the optimal position.

[0066] Through multiple iterations, SSA can gradually approach the global optimal solution, providing an efficient solution for parameter optimization of complex systems.

[0067] like Figure 8 As shown, the steps for predicting the charging load of electric buses using LSTM based on similar daily data are as follows: 1) Data filtering: Collect data from the most recent year and handle outliers and missing values; 2) Data normalization: Normalize the data. 3) Classify the data according to date type, with the Spring Festival as a special category for separate prediction. Multiply the prediction result by the coefficient (1+k) to obtain the final prediction result for the Spring Festival. ; 4) Use the SSA algorithm to find the optimal hyperparameters of the LSTM model, including the number of hidden layer neurons, the number of iterations, and the learning rate; 5) Create an LSTM time series model using the optimal hyperparameters and train the model using data from different categories. For weekdays, weekends, and holidays, the charging load data of the corresponding date type in the most recent six months is used as the model training set. For the data prediction model during the Spring Festival, the charging load data of the previous year's Spring Festival is used as the training set. 6) Record the training parameters, predict the data on the test set, and calculate the evaluation metrics.

[0068] Step S103: Perform variational mode decomposition on the historical non-bus charging load to obtain several mode components. Combine these mode components with time-of-use electricity price, weather type, ambient temperature, and holiday type data to form input vectors. After mapping by the pre-trained non-bus charging load prediction model, the predicted value of non-bus charging load is obtained.

[0069] Variational Mode Decomposition (VMD) is an adaptive and completely non-recursive method for variational mode and signal processing. It has the advantage of being able to determine the number of mode decompositions. Its adaptability is reflected in determining the number of mode decompositions for a given sequence based on the actual situation. In the subsequent search and solution process, it can adaptively match the optimal center frequency and finite bandwidth of each mode. It can also achieve effective separation of intrinsic mode components (IMF), frequency domain partitioning of the signal, and thus obtain the effective decomposition components of a given signal, ultimately obtaining the optimal solution to the variational problem.

[0070] Due to the large load volume, strong volatility, and large fluctuations in electric vehicle charging load, prediction models directly based on historical charging load data have low accuracy. The modal components obtained after VMD processing exhibit strong periodicity and small amplitude variations. Electric vehicle load prediction models built using these modal components as inputs can improve the accuracy of load prediction. The specific implementation method is as follows: 1) Constructing a constrained variational problem The signal f(t) is decomposed into K modal components. To ensure that the decomposed modes can accurately reconstruct the original signal, the sum of all modal components must equal the original signal, and the sum of the bandwidths of each mode must be minimized. To reduce or avoid aliasing, each modal component should ideally contain only a finite number of frequency components. To ensure that each modal function is mainly concentrated around its center frequency, the objective function needs to be minimized to minimize the sum of the estimated bandwidths of all modes. Based on these constraints, a constrained variational model is constructed as shown in the following equation.

[0071]

[0072]

[0073]

[0074] In the formula, For the k-th modal component, For its center frequency, It is the Dirac function. This represents the time partial derivative.

[0075] 2) Transform into an unconstrained problem Construct the Lagrangian function L and introduce a penalty factor. and Lagrange multipliers The above constraints are integrated into the objective function. This is achieved by finding a suitable... , , Minimize the function L, thereby transforming the constrained problem into an unconstrained problem.

[0076]

[0077] In the formula, the penalty factor Control modal bandwidth, value =2000, H is a Lagrange multiplier, and H is the function in step (1).

[0078] 3) Iterative solution and update Solving by alternating direction multipliers and iterative updates , , Finally, a set of mode functions was found. and the corresponding center frequency This ensures that each mode function has a finite bandwidth and that the sum of all mode components equals the original signal as much as possible.

[0079] Setting a penalty factor when decomposing data using VMD The value is 2000, the number of decomposed modes k is 6, and the charging load data is decomposed using VMD. Through VMD decomposition, the original complex signal is effectively separated into three IMF components (IMF1 to IMF3) and residual trends in three different frequency bands.

[0080] In one or more embodiments, the non-bus charging load prediction model employs an LSTM model.

[0081] like Figure 6 As shown, in this embodiment, SSA is used to optimize LSTM parameters. The number of hidden layer neurons directly affects the model's complexity and learning ability. Too few neurons prevent the model from capturing complex nonlinear relationships in the data, while too many increase model complexity and are prone to overfitting. The number of iterations is another key parameter determining the degree of model training. If set too few, the model cannot fully learn the training data; too many also lead to overfitting. The learning rate is an important parameter for adjusting model weight updates, affecting the model's convergence speed. SSA achieves efficient search of the solution space by simulating the cooperation and competition among three roles: discoverer, follower, and watchdog. This invention uses SSA to optimize three parameters of the VMD-LSTM network: the number of hidden layer neurons, the number of iterations, and the learning rate.

[0082] For non-bus charging loads, this embodiment establishes a model based on VMD-SSA-LSTM. The three IMF components of VMD decomposition are combined with time-of-use electricity price, weather type, ambient temperature, and holiday type data as inputs to the LSTM network, respectively. Simultaneously, SSA is used to optimize hyperparameters to construct the VMD-LSTM model. The model's input layer receives raw time-series data, the hidden layer is a two-layer LSTM network with 32 neurons per layer, specifically designed to capture long-term dependencies in the sequence data, dynamically updating cell states through a gating mechanism. Following this is a Dropout layer, used to randomly discard a portion of neurons during training to reduce the model's dependence on training samples and improve its generalization ability. The output layer is a fully connected layer containing one neuron. The model uses the Adam optimizer with a learning rate of 0.008, and the mean squared error (MSE) loss function.

[0083] To verify the accuracy of the optimized LSTM model based on VMD and similar days for charging load prediction, this model was compared with the LSTM and VMD-SSA-LSTM models under the same experimental environment and conditions. The prediction results of the three models are shown in the table below. The optimized LSTM model based on VMD and similar days is able to better capture the features in the data, resulting in an improvement in the mean absolute percentage error (MAPE) compared to the other two models. Figure 9 This is a waveform diagram of the load forecast value according to an embodiment of the present invention; Figure 10 This is the historical charging load curve of an embodiment of the present invention.

[0084]

[0085] This embodiment targets comprehensive charging stations, considering the differences in characteristics among various charging entities. It leverages the uncertainty and volatility of non-bus charging behavior by mapping historical non-bus charging loads through variational mode decomposition to pre-trained non-bus charging load prediction models, thus predicting the actual non-bus charging load. Since electric buses have clearly defined working hours and routes, are less affected by external environments, and exhibit distinct driving and charging patterns, historical charging load data is divided into weekdays, weekends, and holidays. Charging load prediction is then performed based on similar day data, improving the accuracy of short-term charging load predictions for charging stations and achieving the goal of precisely guiding actual charging load allocation decisions.

[0086] like Figure 2 As shown, the orderly charging load short-term prediction system based on vehicle-to-grid interaction provided in this embodiment of the invention can be implemented in software. The orderly charging load short-term prediction system based on vehicle-to-grid interaction includes the following software modules: The charging load division module 201 is used to obtain the charging load of the charging station and its corresponding vehicle identification mark within a historical set time period, and divide the charging load of the charging station into bus charging load and non-bus charging load according to the vehicle identification mark. The bus charging load prediction module 202 is used to cluster historical bus charging loads according to date type. The historical bus charging load of each date type is predicted using a bus charging load prediction model based on similar days. The date types include three types: weekdays, weekends, and public holidays. The non-bus charging load prediction module 203 is used to perform variational mode decomposition on historical non-bus charging loads to obtain several mode components. These mode components are combined with time-of-use electricity prices, weather types, ambient temperatures, and holiday types to form input vectors. After being mapped by a pre-trained non-bus charging load prediction model, the predicted value of non-bus charging load is obtained.

[0087] It should be noted that each module in the orderly charging load short-term prediction system based on vehicle-to-grid interaction in this embodiment of the invention corresponds one-to-one with each step in the orderly charging load short-term prediction method based on vehicle-to-grid interaction in the above embodiment, and their specific implementation processes are the same, so they will not be repeated here.

[0088] The structure of the electronic device according to an embodiment of the present invention will be described in detail below. Figure 3 This is a schematic diagram of the composition structure of an electronic device provided in an embodiment of the present invention. It can be understood that... Figure 3 The diagram shows only an exemplary structure of the electronic device, not the entire structure. Some or all of the structures shown may be implemented as needed.

[0089] The electronic device provided in this embodiment of the invention includes: at least one processor 301, a memory 302, a user interface 303, and at least one network interface 304. The various components in the orderly charging load short-term prediction system based on vehicle-to-grid interaction are coupled together via a bus system 305. It can be understood that the bus system 305 is used to realize the connection and communication between these components. In addition to a data bus, the bus system 305 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in... Figure 3 The general designated all buses as Bus System 305.

[0090] The user interface 303 may include a monitor, keyboard, mouse, trackball, click wheel, buttons, touchpad, or touch screen.

[0091] It is understood that memory 302 can be volatile memory or non-volatile memory, or both. In this embodiment of the invention, memory 302 is capable of storing data to support the operation of the terminal. Examples of this data include any computer programs used to operate on the terminal, such as operating systems and applications. The operating system includes various system programs, such as framework layers, core library layers, driver layers, etc., used to implement various basic services and handle hardware-based tasks. Applications can include various applications.

[0092] In some embodiments, the short-term orderly charging load prediction system based on vehicle-to-grid interaction provided in this invention can be implemented using a combination of hardware and software. As an example, the short-term orderly charging load prediction system based on vehicle-to-grid interaction provided in this invention can be a processor in the form of a hardware decoding processor, programmed to execute the short-term orderly charging load prediction method based on vehicle-to-grid interaction provided in this invention. For example, the processor in the form of a hardware decoding processor can employ one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0093] As an example, processor 301 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., wherein the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0094] As an example of the hardware implementation of the orderly charging load short-term prediction system based on vehicle-to-grid interaction provided in this embodiment of the invention, the device provided in this embodiment of the invention can be directly executed by a processor 301 in the form of a hardware decoding processor. For example, it can be executed by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components to implement the orderly charging load short-term prediction method based on vehicle-to-grid interaction provided in this embodiment of the invention.

[0095] The memory 302 in this embodiment of the invention is used to store various types of data to support the operation of the orderly charging load short-term forecasting system based on vehicle-to-grid interaction, or to store data for execution. Figure 1 The program code for the method shown. Examples of this data include: any executable instructions for operation on a vehicle-to-grid (V2G) based ordered charging load short-term forecasting system, such as executable instructions that can be included in the executable instructions to implement the V2G based ordered charging load short-term forecasting method of the present invention.

[0096] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including functions for executing... Figure 1 The program code for the method shown. In such an embodiment, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium. When the computer program is executed by the central processing unit, it performs the various functions defined in the apparatus of this application.

[0097] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0098] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predicting short-term charging load based on vehicle-grid interaction, characterized in that, include: Obtain the charging load of charging stations and their corresponding vehicle identification numbers within a set historical time period. Based on the vehicle identification numbers, divide the charging load of charging stations into bus charging load and non-bus charging load. Historical bus charging loads are clustered according to date type. The historical bus charging load for each date type is predicted using a bus charging load prediction model based on similar days. The date types include weekdays, weekends, and public holidays. Variational mode decomposition is performed on historical non-bus charging load to obtain several mode components. These mode components are combined with time-of-use electricity price, weather type, ambient temperature, and holiday type data to form input vectors. After being mapped by a pre-trained non-bus charging load prediction model, the predicted value of non-bus charging load is obtained.

2. The method for short-term orderly charging load prediction based on vehicle-grid interaction as described in claim 1, characterized in that, The bus charging load prediction model based on similar days is an LSTM model, which is trained using data sample sets of different date types to obtain the optimal hyperparameters of the LSTM model.

3. The method for short-term orderly charging load prediction based on vehicle-grid interaction as described in claim 2, characterized in that, The SSA algorithm is used to find the optimal hyperparameters of the LSTM model, including the number of hidden layer neurons, the number of iterations, and the learning rate.

4. The method for short-term orderly charging load prediction based on vehicle-grid interaction as described in claim 1, characterized in that, Variational mode decomposition was performed on historical non-bus charging loads to obtain modal components and residual trends in different frequency bands.

5. The method for short-term orderly charging load prediction based on vehicle-grid interaction as described in claim 1, characterized in that, The non-bus charging load prediction model is an LSTM model. It takes modal components, time-of-use electricity prices, weather types, ambient temperatures, and holiday types as inputs to the LSTM model and outputs predicted values ​​for non-bus charging load.

6. The method for short-term orderly charging load prediction based on vehicle-grid interaction as described in claim 5, characterized in that, The SSA algorithm is used to find the optimal hyperparameters for the non-bus charging load prediction model.

7. A short-term predictive system for orderly charging load based on vehicle-grid interaction, characterized in that, include: The charging load division module is used to obtain the charging load of the charging station and its corresponding vehicle identification mark within a historical set time period, and divide the charging load of the charging station into bus charging load and non-bus charging load according to the vehicle identification mark. The bus charging load prediction module is used to cluster historical bus charging loads according to date type. The historical bus charging load for each date type is predicted using a bus charging load prediction model based on similar days. The date types include weekdays, weekends, and public holidays. The non-bus charging load prediction module is used to perform variational mode decomposition on historical non-bus charging loads to obtain several mode components. These mode components are combined with time-of-use electricity prices, weather types, ambient temperatures, and holiday types to form input vectors. After being mapped by a pre-trained non-bus charging load prediction model, the predicted non-bus charging load value is obtained.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the orderly charging load short-term prediction method based on vehicle-to-grid interaction as described in any one of claims 1-6.

9. A computer storage medium having a computer program stored thereon, characterized in that, The program is configured as a step in the method for predicting short-term charging load based on vehicle-to-grid interaction as described in any one of claims 1-6.

10. A computer program product, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the orderly charging load short-term prediction method based on vehicle-to-grid interaction as described in any one of claims 1-6.

Images