Load prediction method, device and equipment of virtual power plant, medium and product

Abstract

The invention discloses a load prediction method and device for a virtual power plant, equipment, a medium and a product. The method comprises the steps of obtaining first influence factor data of a virtual power plant in a target time period; determining a target category set according to the first influence factor data of the target time period and a predetermined data classification result; determining a target load prediction model according to the target category set; the second influence factor data at least comprises the first influence factor data; and based on the target load prediction model, determining a load prediction result of the target time period according to the second influence factor data of the target time period. According to the technical scheme, the problem of low load prediction efficiency is solved, the learning task of the load prediction model can be simplified by performing category division on the target time period based on the influence factor data in advance, and the load prediction efficiency is improved while the load prediction accuracy is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular, to a load prediction method, device, equipment, medium and product for a virtual power plant. Background Art

[0002] A virtual power plant (VPP) is a "virtualized" power plant that aggregates dispersed multiple energy resources such as wind and light through information and communication technologies and software systems to form a unified, coordinated, and flexibly schedulable power plant. The load in future periods of a virtual power plant usually needs to be predicted based on multiple influencing factors, such as historical load, meteorology, electricity price, equipment status, and user behavior, etc.

[0003] Currently, the prior art usually pre-trains a load prediction model based on a deep learning algorithm. For example, a load prediction model is obtained by training a pre-built neural network. After obtaining the load prediction model, multi-factor data is directly used as the input data of the load prediction model to predict the load in future periods. Since the mapping relationship between multi-factor data and load is complex, a deep learning algorithm with a complex structure or a large number of iterative trainings of the deep learning algorithm is required to achieve accurate load prediction, resulting in low load prediction efficiency. Summary of the Invention

[0004] The present invention provides a load prediction method, device, equipment, medium and product for a virtual power plant to solve the problem of low load prediction efficiency. By pre-classifying the target period based on influencing factor data, the learning task of the load prediction model can be simplified, and while ensuring the accuracy of load prediction, the load prediction efficiency is improved.

[0005] According to one aspect of the present invention, there is provided a load prediction method for a virtual power plant, the method comprising:

[0006] Obtaining first influencing factor data of the virtual power plant in a target period;

[0007] Determining a target category set according to the first influencing factor data of the target period and a pre-determined data classification result; the data classification result is obtained by unsupervised classification based on the first influencing factor data of the virtual power plant in each historical period;

[0008] Determining a target load prediction model according to the target category set; the target load prediction model is pre-trained based on the second influencing factor data of each historical period in the target category set; the second influencing factor data at least includes the first influencing factor data;

[0009] Based on the target load prediction model, determine the load prediction result for the target time period according to the second influencing factor data of the target time period.

[0010] According to another aspect of the present invention, there is provided a load prediction device for a virtual power plant, the device comprising:

[0011] An influencing factor data acquisition module, configured to acquire first influencing factor data of the virtual power plant in the target time period;

[0012] A category set determination module, configured to determine a target category set according to the first influencing factor data of the target time period and a pre-determined data classification result; the data classification result is obtained by unsupervised classification based on the first influencing factor data of the virtual power plant in each historical time period;

[0013] A prediction model determination module, configured to determine a target load prediction model according to the target category set; the target load prediction model is pre-trained based on the second influencing factor data of each historical time period in the target category set; the second influencing factor data at least includes the first influencing factor data;

[0014] A prediction result determination module, configured to determine the load prediction result for the target time period based on the target load prediction model according to the second influencing factor data of the target time period.

[0015] According to another aspect of the present invention, there is provided an electronic device, the electronic device comprising:

[0016] At least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the load prediction method of the virtual power plant according to any embodiment of the present invention.

[0017] According to another aspect of the present invention, there is provided a computer-readable storage medium, the computer-readable storage medium stores computer instructions, and the computer instructions are used to implement the load prediction method of the virtual power plant according to any embodiment of the present invention when executed by a processor.

[0018] According to another aspect of the present invention, there is provided a computer program product, including a computer program, and the computer program implements the load prediction method of the virtual power plant according to any embodiment of the present invention when executed by a processor.

[0019] The technical solution of the embodiment of the present invention is as follows: Obtain the first influencing factor data of the virtual power plant in the target period; determine the target category set according to the first influencing factor data in the target period and the pre-determined data classification result; the data classification result is obtained by unsupervised classification based on the first influencing factor data of the virtual power plant in each historical period; determine the target load forecasting model according to the target category set; the target load forecasting model is pre-trained based on the second influencing factor data in each historical period in the target category set; the second influencing factor data at least includes the first influencing factor data; based on the target load forecasting model, determine the load forecasting result in the target period according to the second influencing factor data in the target period. This technical solution solves the problem of low load forecasting efficiency. By pre-classifying the target period based on the influencing factor data, it can simplify the learning task of the load forecasting model and improve the load forecasting efficiency while ensuring the accuracy of the load forecasting.

[0020] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0022] Figure 1 is a flowchart of a method for forecasting the load of a virtual power plant according to Embodiment 1 of the present invention;

[0023] Figure 2 is a flowchart of a method for forecasting the load of a virtual power plant according to Embodiment 2 of the present invention;

[0024] Figure 3 is a schematic structural diagram of a device for forecasting the load of a virtual power plant according to Embodiment 3 of the present invention;

[0025] Figure 4 is a schematic structural diagram of an electronic device for implementing the method for forecasting the load of a virtual power plant in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] To enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0027] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that such data used in appropriate cases can be interchanged so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices. The acquisition, storage, use, processing, etc. of data in the technical solution of this application all comply with the relevant regulations of national laws and regulations.

[0028] Embodiment 1

[0029] Figure 1 A flowchart of a load forecasting method for a virtual power plant is provided for Embodiment 1 of the present invention. This embodiment is applicable to the power planning scenario, especially the load estimation of the virtual power plant. This method can be executed by a load forecasting device of the virtual power plant. The device can be implemented in the form of hardware and / or software, and the device can be configured in an electronic device. As Figure 1 shown, the method includes:

[0030] S110. Obtain the first influencing factor data of the virtual power plant in the target period.

[0031] This solution can be executed by the energy management system of the virtual power plant. The energy management system can obtain various influencing factor data of the virtual power plant in the target period. It can be understood that the virtual power plant can be associated with physical power plants of various power generation energy sources such as thermal power, wind power, and solar energy. The influencing factor data can include environmental data such as light, temperature, wind speed, and rainfall in the areas where the physical power plants associated with the virtual power plant are located, or power consumption data such as current and power of all users in the power supply area covered by the virtual power plant during the target period. It can also include power market disclosure data associated with the virtual power plant, such as data on the unified regulated power consumption load, clearing price, new energy output, and tie-line plan associated with the virtual power plant.

[0032] Specifically, the energy management system can collect the environmental data of the area where the physical power plant is located through the sensing devices deployed in the area where the physical power plant is located, or communicate with the meteorological monitoring system in the area where the physical power plant is located to obtain the environmental data of the area where the physical power plant is located. The energy management system can obtain the electricity consumption data of all users in the power supply area covered by the virtual power plant during the target period through power consumption measurement devices such as smart meters deployed on the user side. The energy management system can obtain information such as documents and news released by the power management department and extract the power market disclosure data from them. Among them, the first influencing factor data can be all or part of the influencing factor data.

[0033] S120. Determine the target category set according to the first influencing factor data in the target period and the pre-determined data classification result; the data classification result is obtained by unsupervised classification based on the first influencing factor data of the virtual power plant in each historical period.

[0034] The energy management system can perform unsupervised classification processing on the first influencing factor data of the virtual power plant in multiple historical periods in advance, divide each historical period into multiple category sets, and obtain the data classification result. The energy management system can classify the first influencing factor data of each historical period based on unsupervised classification models such as K-means clustering, hierarchical clustering, and DBSCAN (Density-Based Spatial Clustering of Applications with Noise) to obtain multiple category sets. It should be noted that there is no intersection between every two category sets.

[0035] Among them, the data classification result can include information such as various category sets and the reference features of various category sets. The reference feature of a category set can be used to represent the features of the first influencing factor data of each historical period in this category set. For example, the average value of the first influencing factor data of each historical period in the category set can be used as the reference feature of this category set.

[0036] The energy management system can compare the first influencing factor data in the target period with the data in each category set in the data classification result to determine the category set to which the target period belongs. In a feasible solution, the data classification result can include the reference features of each category set, and the reference features of each category set can have the same dimension as the first influencing data. The energy management system can calculate the distance between the first influencing factor data in the target period and the reference features of each category set, and use the category set with the reference feature having the smallest distance from the first influencing factor data in the target period as the target category set matched by the target period.

[0037] S130. Determine a target load forecasting model according to the target category set; the target load forecasting model is pre-trained based on the second influencing factor data of each historical period in the target category set; the second influencing factor data includes at least the first influencing factor data.

[0038] After unsupervised classification of each historical period, the energy management system can train the load forecasting model corresponding to each category based on the second influencing factor data of the historical periods in each category set. The second influencing factor data can include the first influencing factor data. If the first influencing factor data is not all the influencing factor data, the second influencing factor data can also include other influencing factor data other than the first influencing factor data.

[0039] It can be understood that the first influencing factor data of the historical periods in the same category set has similarity. Training the load forecasting model for the influencing factor data of the same category reduces the complexity of the load forecasting task on the one hand, and on the other hand, the influencing factor data of each historical moment in the same category set implicitly carries the reference characteristics of the first influencing factor data, which is conducive to improving the fineness of the load forecasting model training, thus ensuring the accuracy of the load forecasting.

[0040] After determining the target category set to which the target period belongs, the energy management system can, according to the target category set, use the load forecasting model corresponding to the target category set as the target load forecasting model.

[0041] S140. Based on the target load forecasting model, determine the load forecasting result of the target period according to the second influencing factor data of the target period.

[0042] The energy management system can obtain the second influencing factor data of the target period, input the second influencing factor data into the target load forecasting model, and use the output of the target load forecasting model as the load forecasting result of the target period.

[0043] The technical solution of the embodiment of the present invention is to obtain the first influencing factor data of the virtual power plant in the target period; determine the target category set according to the first influencing factor data in the target period and the pre-determined data classification result; the data classification result is obtained by unsupervised classification based on the first influencing factor data of the virtual power plant in each historical period; determine the target load prediction model according to the target category set; the target load prediction model is pre-trained based on the second influencing factor data of each historical period in the target category set; the second influencing factor data at least includes the first influencing factor data; based on the target load prediction model, determine the load prediction result of the target period according to the second influencing factor data of the target period. This technical solution solves the problem of low load prediction efficiency. By pre-classifying the target period based on the influencing factor data, the learning task of the load prediction model can be simplified, and while ensuring the accuracy of the load prediction, the load prediction efficiency is improved.

[0044] Embodiment 2

[0045] Figure 2 It is a flowchart of a load prediction method for a virtual power plant provided by Embodiment 2 of the present invention. This embodiment is based on the above embodiment and refines the determination process of the data classification result. As Figure 2 shown, the method includes:

[0046] S201. Obtain the first influencing factor data of the virtual power plant in each historical period, and determine a set of clustering centers matching each classification quantity element from the first influencing factor data of each historical period according to the pre-set set of classification quantities.

[0047] This solution can classify the first influencing factor data of each historical period based on the K-means clustering model. To ensure the reliability of the classification quantity, the energy management system can pre-set a set of classification quantities, list different classification situations. For example, the set of classification quantities can be {2, 3, 4, 5, 6, 7, 8}. The energy management system can perform clustering in sequence according to each classification quantity element in the set of classification quantities to obtain clustering results matching each classification quantity element.

[0048] The energy management system can obtain the first influencing factor data of the virtual power plant in each historical period, take the first influencing factor data of each historical period as a sample point, and randomly select a sample point with the classification quantity from each sample point as the set of clustering centers matching this classification quantity element. For example, for the classification quantity element 8, the energy management system can select 8 different sample points from each sample point clock to form the set of clustering centers corresponding to this classification quantity element, and each sample point can be used as the clustering center of a category.

[0049] Since the data of different influencing factors are usually from different sources, the energy management system can normalize the data of influencing factors before using it to avoid the influence of heterogeneous data and abnormal data. Specifically, the normalization calculation formula of the data of influencing factors can be expressed as:

[0050]

[0051] Among them, q represents the influencing factor index, and x q represents the value of the influencing factor q, and x qmin represents the minimum value of the first influencing factor q, and x qmax represents the maximum value of the first influencing factor q.

[0052] S202. Determine the clustering results matching each classification quantity element according to the first influencing factor data of each historical period and the set of clustering centers matching each classification quantity element.

[0053] According to the set of clustering centers matching each classification quantity element, the energy management system can perform at least one clustering iteration on the first influencing factor data of each historical period according to the preset classification quantity to obtain the clustering results corresponding to each classification quantity element.

[0054] In this solution, optionally, the determining the clustering results matching each classification quantity element according to the first influencing factor data of each historical period and the set of clustering centers matching each classification quantity element includes:

[0055] Take a classification quantity element in the classification quantity set as the target element, and take the set of clustering centers matching the target element as the target clustering center set;

[0056] Calculate the distances between the first influencing factor data of each historical period and each clustering center in the target clustering center set;

[0057] Perform at least one clustering iteration according to the distances between the first influencing factor data of each historical period and each clustering center in the target clustering center set, and determine the clustering results matching the target element according to the clustering results output by each clustering iteration;

[0058] Take the classification quantity elements in the classification quantity set whose clustering results have not been determined as new target elements, and return to execute the calculation of the distances between the first influencing factor data of each historical period and each clustering center in the target clustering center set until the processing of all classification quantity elements in the classification quantity set is completed.

[0059] For each classification quantity element, the energy management system may use the classification quantity element as the target element, use the set of cluster centers of the target element as the target cluster center set, and extract each cluster center in the target cluster center set. The energy management system may use the first influencing factor data of each historical period as sample points, calculate the distances between each sample point and each cluster center line in the target cluster center set in sequence, output the clustering result of this clustering iteration according to the distances between each sample point and each cluster center in the target cluster center set, update the cluster centers of the next clustering iteration according to the clustering result of this clustering iteration, return to perform the next clustering iteration until the clustering iteration termination condition is met, and output the clustering result of the last clustering iteration as the clustering result matching the target element.

[0060] Among them, during the clustering iteration process, the update formula of the cluster center can be shown as follows:

[0061]

[0062] Among them, q represents the first influencing factor index, i represents the clustering iteration number index, x q represents the first influencing factor q, represents the set of categories to which the first influencing factor q belongs in the i-th clustering iteration, represents the cluster center of the set of categories to which the first influencing factor q belongs in the i-th clustering iteration, n represents the number of elements in the set of categories and is the cluster center of the set of categories to which the first influencing factor q belongs in the (i + 1)-th clustering iteration.

[0063] The clustering iteration termination condition may be that the number of clustering iterations reaches a preset number, such as 100 generations, or the clustering result of this clustering iteration meets the preset evaluation condition, such as the data dispersion degree of the clustering result of this clustering iteration is below the preset dispersion degree threshold, or for another example, the clustering results of adjacent clustering iterations do not change. Specifically, the data dispersion degree can be calculated by the dispersion degree calculation formula, and the dispersion degree calculation formula can be expressed as:

[0064]

[0065] Among them, x q represents the first influencing factor q, D q represents the set of categories to which the first influencing factor q belongs, p represents the category set index, m represents the number of category sets, and X q represents the cluster center of the set of categories to which the first influencing factor q belongs.

[0066] The energy management system may perform the above clustering operation for each classification quantity element in the classification quantity set to obtain the clustering results matching each classification quantity element.

[0067] S203. Determine the data classification result according to the clustering results matched by each classification quantity element.

[0068] The energy management system can evaluate the clustering results matched by each classification quantity element, and select one from the clustering results matched by each classification quantity as the data classification result according to the clustering results matched by each classification quantity element.

[0069] In a feasible solution, the energy management system can use the distance from the sample point in the clustering result to the clustering center of the sample point as the clustering loss, and calculate the loss of each classification quantity element. Specifically, the energy management system can use the sum of the distances from each sample point in the clustering result of the classification quantity element to the corresponding clustering center as the loss matched by the classification quantity element. According to the loss matched by the classification quantity element, determine the association relationship between the classification quantity and the loss, such as drawing a line chart of the classification quantity and the loss. Through the line chart of the classification quantity and the loss, based on the elbow method, determine the optimal classification quantity, and use the clustering result matched by the optimal classification quantity element as the data classification result. Among them, the calculation principle of the elbow method is the cost function, and the cost function is the sum of the class distortion degrees. The distortion degree of each class is equal to the sum of the squares of the position distances from each variable point to its clustering center. The more compact the members within the class set are, the smaller the distortion degree of the class is, and the more dispersed it is, the larger it is. As the number of clusters increases, the number of samples included in each class will decrease, so the samples will be closer to the clustering center, and the average distortion degree will decrease. As the number of clusters continues to increase, the improvement effect of the average distortion degree will continue to decrease. During the process of increasing the number of clusters, the clustering number corresponding to the position where the decrease amplitude of the improvement effect of the distortion degree is the largest is the elbow.

[0070] In another feasible solution, the step of determining the data classification result according to the clustering results matched by each classification quantity element includes:

[0071] Determine the clustering evaluation index matched by each classification quantity element according to the clustering results matched by each classification quantity element; the clustering evaluation index is used to evaluate the distance between the first influencing factor data in each historical period and the data in each category set;

[0072] Determine the data classification result according to the clustering evaluation index matched by each classification quantity element.

[0073] It can be understood that the energy management system can evaluate the clustering results matched by each classification quantity element and determine the clustering evaluation index corresponding to each classification quantity element. Specifically, the energy management system can evaluate the distance between the first influencing factor data in each historical period and the data in each category set, and the clustering evaluation index can be expressed as:

[0074]

[0075] Among them, N represents the number of historical time periods, a represents the historical time period index, P(a) represents the average distance from the first influencing factor data of historical time period a to the first influencing factor data of each historical time period in the non - belonging category set, and q(a) represents the average distance from the first influencing factor data of any historical time period a to the first influencing factor data of other historical time periods in the belonging category set.

[0076] P can be used to characterize the overall clustering result of all historical time periods. The larger P is, the better the overall clustering effect. The energy management system can calculate the clustering evaluation indicators matching each classification quantity element in turn, sort the clustering evaluation indicators matching each classification quantity element, select the classification quantity element with the largest P among each classification quantity element, and use the clustering result of this classification quantity element as the data classification result.

[0077] This solution can evaluate the clustering results of different classification quantities and select the best data classification result, which is beneficial to ensuring the reliability of the data classification result.

[0078] S204. Take each category set in the data classification result as the to - be - processed category set in turn, take the second influencing factor data of each historical time period in the to - be - processed category set as the to - be - processed samples, and obtain the labels matching each to - be - processed sample to form a to - be - trained data set; the label matching the to - be - processed sample is the actual load matching the second influencing factor data of the historical time period.

[0079] After obtaining the data classification result, the energy management system can use the influencing factor data of the historical time periods in each category set in the data classification result to train the load prediction model corresponding to each category set in turn. Specifically, the energy management system can take each category set in the data classification result as the to - be - processed category set in turn, take the second influencing factor data of each historical time period in the to - be - processed category set as the to - be - processed samples, and obtain the labels matching each to - be - processed sample to form a to - be - trained data set. It can be understood that the label matching the to - be - processed sample is the actual load matching the second influencing factor data of the historical time period.

[0080] S205. Input each to - be - processed sample in the to - be - trained data set into the pre - constructed neural network in turn to obtain the predicted load matching each to - be - processed sample.

[0081] The energy management system can pre - build a neural network as the to - be - trained load prediction model. The neural network can include an input layer, at least one hidden layer, and an output layer. The number of neurons in the hidden layer of the neural network can be set based on the number of neurons in the input layer and the number of neurons in the output layer. For example, the number of neurons in the hidden layer can satisfy the following formula: Among them, a represents the number of neurons in the input layer, c represents the number of neurons in the output layer, k is an integer, and 1 < k < 10. The activation functions of the output layer can be functions such as Sigmoid, Tanh, ReLU, and Softmax. Each sample to be processed in the dataset to be trained is sequentially input into the pre-constructed neural network, and the neural network can output the predicted load matching each sample to be processed.

[0082] S206. According to the predicted load matching each sample to be processed and the label matching each sample to be processed, perform at least one training iteration on the neural network to determine the load prediction model matching the set of samples to be processed.

[0083] The energy management system can compare the predicted load matching each sample to be processed with the actual load matching each sample to be processed to determine the prediction error of this iteration, and perform at least one training iteration on the weight coefficients of each neuron in the neural network according to the prediction error to obtain a load prediction model that meets the prediction accuracy requirements. After completing the training of the load prediction model for the set of samples to be processed, the energy management system can use other unprocessed sets of samples as the new set of samples to be processed, and train the load prediction model for this set of samples to be processed according to steps S204 - S206 until the training of the load prediction models for all sets of samples is completed, and the energy management system can obtain the load prediction models matching each set of samples.

[0084] In a preferred solution, the performing at least one training iteration on the neural network according to the predicted load matching each target sample and the label matching each target sample to determine the target load prediction model includes:

[0085] Determine the prediction error matching this iteration according to the predicted load matching each target sample and the label matching each target sample;

[0086] Determine the weight adjustment parameter matching this iteration according to the prediction error matching this iteration and the neural network weight coefficients obtained in this iteration;

[0087] Perform at least one training iteration on the neural network according to the weight adjustment parameter matching this iteration to determine the target load prediction model.

[0088] The energy management system can sum the difference between the predicted load matched with each target sample and the actual load matched with each target sample to obtain the prediction error matched in this iteration. According to the prediction error matched in this iteration and the weight coefficients of each neuron in the neural network obtained in this iteration, the weight adjustment parameter matched in this iteration is determined for updating the weight coefficients of each neuron in the neural network in the next iteration. The energy management system realizes the training of the neural network by updating the weight coefficients of each neuron in the neural network during each training iteration, so as to obtain the target load prediction model.

[0089] In a feasible solution, the weight adjustment parameter matched in this iteration can be expressed as:

[0090] Δw = (J T J + μI) -1 ·J T e;

[0091] Where, J represents the Jacobian matrix generated in this iteration, e represents the prediction error in this iteration, I represents the identity matrix, and μ represents a preset coefficient, which is a constant.

[0092] It can be understood that the prediction error can be expressed as a multivariate vector function of the weight coefficients of each neuron in the neural network. The energy management system can take the partial derivatives of each weight coefficient according to the multivariate vector function representing the prediction error and calculate the Jacobian matrix.

[0093] This solution adopts the LM algorithm to improve problems such as slow convergence speed and easy to fall into local minimum of the neural network, which is beneficial to improving the training accuracy and convergence speed of the neural network to save computational cost and time cost.

[0094] S207. Obtain the first influencing factor data of the virtual power plant in the target period.

[0095] S208. Determine the target category set according to the first influencing factor data in the target period and the pre-determined data classification result.

[0096] S209. Determine the target load prediction model according to the target category set; the target load prediction model is pre-trained based on the second influencing factor data of each historical period in the target category set; the second influencing factor data at least includes the first influencing factor data.

[0097] S210. Based on the target load prediction model, determine the load prediction result of the target period according to the second influencing factor data of the target period.

[0098] This solution realizes the classification of the first influencing factor data in the historical period based on K-means clustering. By evaluating the clustering results of multiple classification quantity elements in the classification quantity set, the optimal data classification result is obtained, which is beneficial to ensuring the reliability of data classification. By pre-classifying the target period based on the influencing factor data, the learning task of the load forecasting model can be simplified, while ensuring the accuracy of load forecasting, the load forecasting efficiency is improved.

[0099] Embodiment III

[0100] Figure 3 It is a schematic structural diagram of a load forecasting device for a virtual power plant provided by Embodiment III of the present invention. As Figure 3 shown, the device includes:

[0101] The factor data acquisition module 310 is used to acquire the first influencing factor data of the virtual power plant in the target period;

[0102] The category set determination module 320 is used to determine the target category set according to the first influencing factor data in the target period and the pre-determined data classification result; the data classification result is obtained by unsupervised classification based on the first influencing factor data of the virtual power plant in each historical period;

[0103] The prediction model determination module 330 is used to determine the target load forecasting model according to the target category set; the target load forecasting model is pre-trained based on the second influencing factor data of each historical period in the target category set; the second influencing factor data at least includes the first influencing factor data;

[0104] The prediction result determination module 340 is used to determine the load forecasting result of the target period based on the target load forecasting model according to the second influencing factor data of the target period.

[0105] In this solution, the determination process of the data classification result includes:

[0106] Acquire the first influencing factor data of the virtual power plant in each historical period, and determine the set of clustering centers matching each classification quantity element from the first influencing factor data of each historical period according to the pre-set classification quantity set;

[0107] Determine the clustering results matching each classification quantity element according to the first influencing factor data of each historical period and the set of clustering centers matching each classification quantity element;

[0108] Determine the data classification result according to the clustering results matching each classification quantity element.

[0109] Based on the above solution, determining the clustering results corresponding to each classification quantity element according to the first influencing factor data of each historical period and the set of clustering centers matched with each classification quantity element includes:

[0110] Taking a classification quantity element in the classification quantity set as the target element, and taking the set of clustering centers matched with the target element as the target clustering center set;

[0111] Calculating the distances between the first influencing factor data of each historical period and each clustering center in the target clustering center set;

[0112] Performing at least one clustering iteration according to the distances between the first influencing factor data of each historical period and each clustering center in the target clustering center set, and determining the clustering result matched with the target element according to the clustering results output by each clustering iteration;

[0113] Taking the classification quantity elements in the classification quantity set for which the clustering results have not been determined as new target elements, and returning to execute the calculation of the distances between the first influencing factor data of each historical period and each clustering center in the target clustering center set until the processing of all classification quantity elements in the classification quantity set is completed.

[0114] In a feasible solution, determining the data classification result according to the clustering results matched with each classification quantity element includes:

[0115] Determining the clustering evaluation indexes matched with each classification quantity element according to the clustering results matched with each classification quantity element; the clustering evaluation indexes are used to evaluate the distances between the first influencing factor data of each historical period and the data in each category set;

[0116] Determining the data classification result according to the clustering evaluation indexes matched with each classification quantity element.

[0117] In this embodiment, optionally, the training process of the target load prediction model includes:

[0118] Taking the second influencing factor data of each historical period in the target category set as target samples, and obtaining the labels matched with each target sample to form a target training data set; the label matched with the target sample is the actual load matched with the second influencing factor data of the historical period;

[0119] Sequentially inputting each target sample in the target training data set into a pre-constructed neural network to obtain the predicted load matched with each target sample;

[0120] Performing at least one training iteration on the neural network according to the predicted load matched with each target sample and the label matched with each target sample to determine the target load prediction model.

[0121] Based on the above solution, determining the target load prediction model by performing at least one training iteration on the neural network according to the predicted load matched with each target sample and the label matched with each target sample includes:

[0122] Determine the prediction error matched in this iteration according to the predicted load matched with each target sample and the label matched with each target sample;

[0123] Determine the weight adjustment parameter matched in this iteration according to the prediction error matched in this iteration and the neural network weight coefficient obtained in this iteration;

[0124] Perform at least one training iteration on the neural network according to the weight adjustment parameter matched in this iteration to determine the target load prediction model.

[0125] The load prediction device of the virtual power plant provided by the embodiments of the present invention can execute the load prediction method of the virtual power plant provided by any embodiment of the present invention, and has corresponding functional modules and beneficial effects for executing the method.

[0126] Embodiment 4

[0127] Figure 4 FIG. shows a schematic structural diagram of an electronic device 410 that can be used to implement the embodiments of the present invention. The electronic device is intended to represent various forms of digital computers, such as, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as, personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0128] As Figure 4 shown, the electronic device 410 includes at least one processor 411, and a memory communicatively connected to at least one processor 411, such as a read-only memory (ROM) 412, a random access memory (RAM) 413, etc. Among them, the memory stores a computer program executable by at least one processor. The processor 411 can execute various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 412 or the computer program loaded from the storage unit 418 into the random access memory (RAM) 413. In the RAM 413, various programs and data required for the operation of the electronic device 410 can also be stored. The processor 411, the ROM 412, and the RAM 413 are connected to each other through a bus 414. The input / output (I / O) interface 415 is also connected to the bus 414.

[0129] Multiple components in the electronic device 410 are connected to the I / O interface 415, including: an input unit 416, such as a keyboard, a mouse, etc.; an output unit 417, such as various types of displays, speakers, etc.; a storage unit 418, such as a disk, an optical disc, etc.; and a communication unit 419, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 419 allows the electronic device 410 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0130] The processor 411 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 411 include but are not limited to a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor 411 executes the various methods and processes described above, such as the load prediction method of the virtual power plant.

[0131] In some embodiments, the load prediction method of the virtual power plant can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as the storage unit 418. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 410 via the ROM 412 and / or the communication unit 419. When the computer program is loaded into the RAM 413 and executed by the processor 411, one or more steps of the load prediction method of the virtual power plant described above can be executed. Alternatively, in other embodiments, the processor 411 can be configured to execute the load prediction method of the virtual power plant in any other suitable manner (e.g., by means of firmware).

[0132] The various embodiments of the systems and techniques described above in this article can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGA), application specific integrated circuits (ASIC), application specific standard products (ASSP), systems on a chip (SOC), complex programmable logic devices (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special or general programmable processor, and can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0133] A computer program for implementing the method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable load prediction devices of a virtual power plant, such that when the computer programs are executed by the processor, the functions / operations specified in the flowchart and / or block diagram are implemented. The computer programs can be executed entirely on the machine, partially on the machine, executed partially on the machine and partially on a remote machine as an independent software package, or executed entirely on a remote machine or server.

[0134] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include electrical connections based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0135] In order to provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball), by which the user can provide input to the electronic device. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and the input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0136] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected with each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), blockchain network, and the Internet.

[0137] A computing system can include a client and a server. The client and the server are generally remote from each other and typically interact through a communication network. The client-server relationship is created by computer programs running on respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, solving the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.

[0138] It should be understood that various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in the present invention can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is made herein.

[0139] The above specific embodiments do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A load forecasting method for a virtual power plant, characterized in that, The method comprises: Obtain the primary influencing factor data of the virtual power plant during the target period; Determining a target category set based on the first influencing factor data of the target period and a predetermined data classification result; the data classification result is obtained by performing unsupervised classification on the first influencing factor data of the virtual power plant in each historical period; Determining a target load forecasting model based on the target category set; the target load forecasting model is pre-trained based on second influencing factor data of each historical period in the target category set; the second influencing factor data at least includes the first influencing factor data; Based on the target load forecasting model and according to the second influencing factor data of the target period, a load forecast result of the target period is determined.

2. The method according to claim 1, wherein The process of determining the data classification result includes: Obtain the first influencing factor data of the virtual power plant in each historical period, and determine the cluster center set that matches each classification quantity element from the first influencing factor data of each historical period according to a preset classification quantity set; Determine the clustering results of the matching of each classification quantity element based on the first influencing factor data of each historical period and the cluster center set of the matching of each classification quantity element; The data classification results are determined based on the clustering results of the matching elements of each classification quantity.

3. The method according to claim 2, wherein The determining of the clustering results of the matching of each classification quantity element based on the first influencing factor data of each historical period and the cluster center set of the matching of each classification quantity element includes: A classification quantity element in the classification quantity set is used as the target element, and the cluster center set that matches the target element is used as the target cluster center set; Calculate the distance between the first influencing factor data of each historical period and each cluster center in the target cluster center set; Perform at least one clustering iteration based on the distance between the first influencing factor data of each historical period and each cluster center in the target cluster center set, and determine the clustering result matching the target element based on the clustering results output by each clustering iteration; The classification quantity elements in the classification quantity set whose clustering results have not been determined are taken as new target elements, and the distance between the first influencing factor data of each historical period and each cluster center in the target cluster center set is calculated again until the processing of all classification quantity elements in the classification quantity set is completed.

4. The method according to claim 2, wherein The data classification result is determined based on the clustering results of the matching elements of each classification quantity, including: Determine the clustering evaluation index of each classification quantity element match based on the clustering results of each classification quantity element match; the clustering evaluation index is used to evaluate the distance between the first influencing factor data of each historical period and the data in each category set; The data classification results are determined based on the clustering evaluation indicators that match the quantitative elements of each classification.

5. The method according to claim 1, wherein The training process of the target load forecasting model includes: The second influencing factor data of each historical period in the target category set is used as the target sample, and the label matched by each target sample is obtained to form a target training data set; the label matched by the target sample is the actual load matched by the second influencing factor data of the historical period; Input each target sample in the target training data set into the pre-built neural network in turn to obtain the predicted load matched by each target sample; Based on the predicted load matched with each target sample and the label matched with each target sample, perform at least one training iteration on the neural network to determine the target load prediction model.

6. The method according to claim 5, characterized in that, The step of performing at least one training iteration on the neural network based on the predicted load matched with each target sample and the label matched with each target sample to determine the target load prediction model includes: Determine the prediction error matched in this iteration based on the predicted load matched with each target sample and the label matched with each target sample. Determine the weight adjustment parameter matched in this iteration based on the prediction error matched in this iteration and the neural network weight coefficient obtained in this iteration. Based on the weight adjustment parameter matched in this iteration, perform at least one training iteration on the neural network to determine the target load prediction model.

7. A load forecasting device for a virtual power plant, characterized in that, The device includes: A factor data acquisition module, configured to acquire first influencing factor data of the virtual power plant in a target time period. A category set determination module, configured to determine a target category set according to the first influencing factor data in the target time period and a pre-determined data classification result; the data classification result is obtained by unsupervised classification based on the first influencing factor data of the virtual power plant in each historical time period. A prediction model determination module, configured to determine a target load prediction model according to the target category set; the target load prediction model is pre-trained based on second influencing factor data of each historical time period in the target category set; the second influencing factor data at least includes the first influencing factor data. A prediction result determination module, configured to determine the load prediction result of the target time period based on the target load prediction model and the second influencing factor data of the target time period.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the load prediction method of the virtual power plant according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a processor to execute the load prediction method of the virtual power plant according to any one of claims 1-6 when executed.

10. A computer program product, including a computer program, where the computer program, when executed by a processor, implements the load prediction method of the virtual power plant according to any one of claims 1-6.

Images