Method and device for predicting multi-energy load of integrated energy system

Abstract

The invention provides a multi-energy load prediction method and device for an integrated energy system, and belongs to the field of load prediction. The method comprises the steps that target data in a preset time period are acquired, and the target data comprise load data and climate data of various energy sources; performing clustering processing and variational mode decomposition on the load data of the multiple energy sources to obtain first processing data; performing correlation analysis on the first processing data and the climate data to obtain second processing data; inputting the second data into a feature extraction model to obtain high-dimensional time features of the various energy sources, and inputting the high-dimensional time features into a pre-constructed load prediction model to obtain predicted loads of the various energy sources; wherein the feature extraction model comprises an attention mechanism module, and the load prediction model comprises a time attention mechanism. According to the invention, the accuracy of multi-energy load prediction can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of load forecasting, and in particular to a method and device for forecasting multi-energy loads in an integrated energy system. Background Art

[0002] With the increasing global warming, the gradual depletion of fossil energy and the increasingly prominent problems such as the imbalance between supply and demand in the power system, there is an urgent need to build an accurate and efficient integrated energy system (IES).

[0003] The prediction of multi-energy loads is the basis of IES operation and scheduling. Accurate prediction of cooling, heating and electricity loads is of great significance to the development of IES. The complementary effect between various energy forms in the IES system leads to the coupling characteristics between multi-energy loads.

[0004] At present, the commonly used traditional load prediction methods are mainly aimed at the prediction of a single load type, including statistical methods and machine learning methods. Statistical methods mainly include regression analysis, exponential smoothing and other methods, and machine learning methods mainly include random forest, support vector machine and recurrent neural network and its variants. The complex coupling relationship between multiple energy sources makes single load prediction unable to meet the current needs of IES development. How to improve the accuracy of multi-energy load prediction has become a technical problem that needs to be solved urgently. Summary of the invention

[0005] The embodiment of the present invention provides a method and device for predicting multi-energy loads of an integrated energy system, so as to solve the problem of low prediction accuracy of the current traditional load prediction method.

[0006] In a first aspect, an embodiment of the present invention provides a prediction of multi-energy loads of an integrated energy system, including:

[0007] Obtaining target data within a preset time period, wherein the target data includes load data and climate data of multiple energy sources;

[0008] Performing clustering processing and variational mode decomposition on load data of multiple energy sources to obtain first processed data;

[0009] Performing correlation analysis on the first processed data and the climate data to obtain second processed data;

[0010] The second data is input into the feature extraction model to obtain high-dimensional time features of multiple energy sources, and the high-dimensional time features are input into a pre-built load forecasting model to obtain predicted loads of various energy sources; wherein the feature extraction model includes an attention mechanism module, and the load forecasting model includes a time attention mechanism.

[0011] In a possible implementation, clustering processing and variational mode decomposition are performed on load data of multiple energy sources to obtain first processed data, including:

[0012] Clustering and dividing the load data of various energy sources based on the elbow method to obtain the first sub-data;

[0013] Perform variational mode decomposition on the first sub-data to obtain first processed data.

[0014] In a possible implementation, performing a correlation analysis on the first processed data and the climate data to obtain the second processed data includes:

[0015] Performing correlation analysis on the first processed data and the climate data to determine correlation data between the loads of multiple energy sources and the climate;

[0016] Based on the correlation data between the load of various energy sources and the climate, the first processed data and the climate data are eliminated to obtain the second processed data.

[0017] In a possible implementation, the center of the cluster division is 4.

[0018] In a possible implementation, the load data of the multiple energy sources include electric load data, cooling load data, and heating load data;

[0019] Climate data includes air pressure, wind speed, temperature, solar radiation, precipitation, humidity, and dew point temperature.

[0020] In a possible implementation, the correlation analysis includes any one or more of a Pearson correlation coefficient, a Spearman correlation coefficient, or a rank correlation coefficient;

[0021] The correlation data between the loads of various energy sources and the climate include: the electric load data is related to the air pressure, temperature, precipitation and dew point temperature; the heat load data is related to the air pressure, temperature, solar radiation and dew point temperature; the cooling load data is related to the air pressure, temperature, solar radiation, humidity and dew point temperature.

[0022] In a possible implementation, before clustering and variational mode decomposition of load data of multiple energy sources, the following steps are also included:

[0023] Use the standard score method to detect abnormal data and remove them;

[0024] Use exponential smoothing to fill in the removed abnormal data and missing data to obtain the processed data;

[0025] The processed data are normalized.

[0026] In one possible implementation, the feature extraction model includes a deep residual network and an attention mechanism module.

[0027] In one possible implementation, the load forecasting model includes LSTM and a temporal attention mechanism.

[0028] In a first aspect, an embodiment of the present invention provides a device for predicting multi-energy loads of an integrated energy system, comprising:

[0029] A data acquisition module, used to acquire target data within a preset time period, wherein the target data includes load data and climate data of various energy sources;

[0030] A first processing module is used to perform clustering processing and variational mode decomposition on load data of multiple energy sources to obtain first processed data;

[0031] A second processing module is used to perform correlation analysis on the first processed data and the climate data to obtain second processed data;

[0032] The prediction module is used to input the second data into the feature extraction model to obtain high-dimensional time features of multiple energy sources, and input the high-dimensional time features into a pre-built load prediction model to obtain predicted loads of various energy sources; wherein the feature extraction model includes an attention mechanism module, and the load prediction model includes a time attention mechanism.

[0033] In a possible implementation, the first processing module is used to cluster and divide the load data of multiple energy sources based on the elbow method to obtain first sub-data;

[0034] Perform variational mode decomposition on the first sub-data to obtain first processed data.

[0035] In a possible implementation, the second processing module is used to perform a correlation analysis on the first processed data and the climate data to determine the correlation data between the loads of multiple energy sources and the climate;

[0036] Based on the correlation data between the load of various energy sources and the climate, the first processed data and the climate data are eliminated to obtain the second processed data.

[0037] In a possible implementation, the center of the cluster division is 4.

[0038] In a possible implementation, the load data of the multiple energy sources include electric load data, cooling load data, and heating load data;

[0039] Climate data includes air pressure, wind speed, temperature, solar radiation, precipitation, humidity, and dew point temperature.

[0040] In a possible implementation, the correlation analysis includes any one or more of a Pearson correlation coefficient, a Spearman correlation coefficient, or a rank correlation coefficient;

[0041] The correlation data between the loads of various energy sources and the climate include: the electric load data is related to the air pressure, temperature, precipitation and dew point temperature; the heat load data is related to the air pressure, temperature, solar radiation and dew point temperature; the cooling load data is related to the air pressure, temperature, solar radiation, humidity and dew point temperature.

[0042] In a possible implementation, the data acquisition module is used to detect abnormal data using a standard score method and remove the abnormal data;

[0043] Use exponential smoothing to fill in the removed abnormal data and missing data to obtain the processed data;

[0044] The processed data are normalized.

[0045] In one possible implementation, the feature extraction model includes a deep residual network and an attention mechanism module.

[0046] In one possible implementation, the load forecasting model includes LSTM and a temporal attention mechanism.

[0047] The embodiment of the present invention provides a method and device for predicting multi-energy loads in an integrated energy system, which predicts multi-energy loads in a preset time period by acquiring target data in a preset time period. Among them, the target data covers load data and climate data of multiple energy sources, which can reflect the usage of multiple energy sources in a preset time period. Then, since the load data of multiple energy sources are random and volatile, it is necessary to cluster the load data of multiple energy sources, so that the load data can be classified to ensure that accurate and consistent data sets can be obtained in subsequent analysis and prediction. In addition, irregular load data can also be decomposed into relatively stable data components of different frequencies through variational mode decomposition, so that accurate load prediction can be achieved. Then, since there are many climate data, and each energy is related to different climates, it is necessary to find out which climate factors each energy is related to, so as to determine the prediction model and the input second processing data based on the correlation. Finally, the second data is input into the feature extraction model to obtain high-dimensional time features of multiple energy sources, and the high-dimensional time features are input into the pre-built load prediction model to obtain the predicted loads of various energy sources. The present invention first performs cluster analysis and variational mode processing on the load data, thereby improving the stability of the extracted data and the accuracy of the features. In addition, by adding an attention mechanism module to the feature extraction model and adding a time attention mechanism to the load forecasting model, the accuracy of the model can be further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 It is a flow chart of the implementation of the method for predicting multi-energy loads of an integrated energy system provided by an embodiment of the present invention;

[0050] Figure 2 It is a block diagram of a method for predicting multi-energy loads of an integrated energy system provided by an embodiment of the present invention;

[0051] Figure 3 is a schematic diagram of cluster analysis provided by an embodiment of the present invention;

[0052] Figure 4 is a schematic diagram of correlation analysis using the Pearson correlation coefficient provided by an embodiment of the present invention;

[0053] Figure 5 is a schematic diagram of correlation analysis using the Spearman correlation coefficient provided by an embodiment of the present invention;

[0054] Figure 6 is a schematic diagram of correlation analysis using a rank correlation coefficient provided by an embodiment of the present invention;

[0055] Figure 7 is a schematic diagram of the prediction result of the electric load provided by an embodiment of the present invention;

[0056] Figure 8 is a schematic diagram of the prediction result of the cooling load provided by an embodiment of the present invention;

[0057] Fig. 9 is a schematic diagram of the prediction result of the heat load provided by an embodiment of the present invention;

[0058] Fig.10 It is a structural schematic diagram of a device for predicting multi-energy loads of an integrated energy system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0060] In order to make the purpose, technical solutions and advantages of the present invention more clear, specific embodiments will be described below in conjunction with the accompanying drawings.

[0061] Although there are many load forecasting methods, most of them are based on a single energy source. For multiple energy sources, since the load data of each energy source will be affected by a variety of climate factors, the load data will be random and volatile. If the load data of multiple energy sources are directly used for forecasting, the accuracy of the forecast will be affected.

[0062] The execution subjects of each embodiment of the present application can be servers, processors, microprocessors and other devices with data processing functions. In the actual implementation process, the specific implementation method of the execution subject can be selected according to actual needs. This embodiment does not impose any special restrictions on this, as long as it is a device with data processing functions.

[0063] Please also see Figure 1 and Figure 2 , Figure 1 The implementation flow chart of the method for predicting multi-energy loads of an integrated energy system provided by the embodiment of the present invention is described in detail as follows:

[0064] S110: Acquire target data within a preset time period.

[0065] The target data include load data of various energy sources and climate data.

[0066] In some embodiments, the load data of the energy in the Tempe campus can be obtained through the official website of Arizona State University. The climate data can be obtained from the National Renewable Energy Laboratory of the United States.

[0067] The load data here include electric load data in (KW), cooling load data in (Tons) and heating load data in (mmBTU).

[0068] Climate data include: air pressure (hpa), wind speed (m / s), temperature (℃), solar radiation (w / m^2), precipitation (mm), humidity (%) and dew point temperature (℃).

[0069] In this embodiment, the sampling frequency of the load data and the climate data is 1 hour.

[0070] For example, the data may be two years of data from 0:00 on January 1, 2021 to 23:00 on December 31, 2022.

[0071] S120, performing clustering processing and variational mode decomposition on load data of multiple energy sources to obtain first processed data.

[0072] In some embodiments, due to power outages, network disconnections, sensor failures, etc., the collected data may contain obvious abnormal data and missing data. Therefore, before analyzing the data, the data must first be processed.

[0073] In this embodiment, first, the standard score method Z-score may be used to detect abnormal data and then remove the abnormal data.

[0074] Then, exponential smoothing is used to fill in the removed abnormal data and missing data to obtain processed data, thereby improving the quality of the data.

[0075] Finally, in order to normalize the original data to the same range, the processed data also needs to be normalized.

[0076] As described above, the loads of various energy sources are affected by climate data, which makes the load data random and volatile. Therefore, in order to accurately predict, it is necessary to divide similar sample data based on the characteristics of load data. The purpose is to identify and classify load data in the same operating state, so as to ensure that accurate and consistent data sets can be obtained in subsequent analysis and prediction.

[0077] Accurate load data division can not only capture the dynamic and potential characteristics of load data more accurately, but also provide a corresponding basis for subsequent predictions.

[0078] In this embodiment, first, the load data of multiple energy sources may be clustered and divided based on the elbow method to obtain the first sub-data.

[0079] For example, the cluster center is 4, such as Figure 3 As shown in Figure 1, by calculating the minimum Euclidean distance from the load data to the cluster center, the sample mean of each cluster is updated to the cluster center.

[0080] In addition, due to the volatility and randomness of load data, it is difficult for the load forecasting model to capture the characteristic information between load data when processing these data. The highly nonlinear relationship between the original load data makes it difficult for the load forecasting model to construct the corresponding subtasks. Therefore, it is also necessary to use variational mode decomposition VMD to decompose the clustered data. VMD decomposes the different similar loads after clustering into multiple sequences with simple structures. VMD realizes the frequency domain division of different load data, decomposing irregular load data into relatively stable data components of different frequencies, so that accurate load forecasting can be achieved.

[0081] S130, performing correlation analysis on the first processed data and the climate data to obtain second processed data.

[0082] In some embodiments, the correlation analysis includes a Pearson correlation coefficient, a Spearman correlation coefficient, or a rank correlation coefficient.

[0083] The Pearson correlation coefficient (PCC) quantifies the linear correlation between two variables X and Y, that is, the correlation between load data and load data and climate data. The calculation process is as follows:

[0084]

[0085] Among them, P represents PCC, n represents the total number of samples of a single variable, and x i and i represents the i-th sample of x and y, and Represents the average of two variables.

[0086] The results of correlation analysis using Pearson correlation coefficient are as follows: Figure 4 shown.

[0087] The Spearman correlation coefficient quantifies the monotonic correlation between variables through a linear correlation analysis of the ranks of two variables. The calculation process is as follows:

[0088]

[0089] Among them, n represents the total number of samples of a single variable, d i Represents x i and i The rank difference between .

[0090] The results of correlation analysis using Spearman correlation coefficient are as follows Figure 5 shown.

[0091] The rank correlation coefficient is a method used to assess the general correlation between variables and is calculated as follows:

[0092]

[0093] Where n represents the total number of samples of a single variable, and C and D represent the number of pairs of elements with consistency and inconsistency in the sample data x and y.

[0094] The results of correlation analysis using rank correlation coefficient are as follows: Figure 6 shown.

[0095] In this embodiment, one of the Pearson correlation coefficient, the Spearman correlation coefficient, or the rank correlation coefficient may be used, or the three correlation calculation methods may be used simultaneously.

[0096] In some embodiments, first, correlation analysis may be performed on the first processed data and the climate data to determine correlation data between the loads of multiple energy sources and the climate.

[0097] The correlation data between the loads of various energy sources and the climate include: the electric load data is related to the air pressure, temperature, precipitation and dew point temperature; the heat load data is related to the air pressure, temperature, solar radiation and dew point temperature; the cooling load data is related to the air pressure, temperature, solar radiation, humidity and dew point temperature.

[0098] Then, based on the correlation data between the loads of various energy sources and the climate, the first processed data and the climate data are eliminated to obtain the second processed data.

[0099] After removing irrelevant data, the second processed data can be obtained.

[0100] In some embodiments, a load prediction model for each energy source may also be established based on the correlation.

[0101] The relevant data of electric load, air pressure, temperature, precipitation and dew point temperature are collected, and the pre-established neural network model is trained to obtain the electric load prediction model.

[0102] The relevant data of heat load, air pressure, temperature, solar radiation and dew point temperature are collected, and the pre-established neural network model is trained to obtain the heat load prediction model.

[0103] The relevant data of cooling load, air pressure, temperature, solar radiation, humidity and dew point temperature are collected, and the pre-established neural network model is trained to obtain the cooling load prediction model.

[0104] S140, inputting the second data into a feature extraction model to obtain high-dimensional time features of multiple energy sources, and inputting the high-dimensional time features into a pre-built load forecasting model to obtain predicted loads of various energy sources.

[0105] Among them, the feature extraction model includes an attention mechanism module, and the load forecasting model includes a time attention mechanism.

[0106] In this embodiment, the feature extraction model includes a deep residual network and an attention mechanism module.

[0107] Deep Residual Network (ResNet) is a feedforward neural network with deep structure and convolution calculation. It mainly includes input layer, convolution layer, activation function layer, pooling layer, fully connected layer and output layer. The input layer receives the raw data to be processed, mainly including load and climate data. The input data is convolved by the convolution kernel to extract the high-order features of the raw data. After convolution, the activation function is added to increase the nonlinear characteristics of the network. The pooling layer compresses the input features, which reduces the size of the feature map and simplifies the complexity of network calculation on the one hand, and compresses the features and extracts the main features on the other hand. The fully connected layer integrates the output of the convolution and pooling layers into the final classification or regression results. Therefore, ResNet can be used to extract high-dimensional spatiotemporal features of load data.

[0108] However, a simple ResNet network cannot effectively extract the features of the original data. In order to effectively obtain the features of the original data, the attention mechanism module CBAM can be used to enhance the feature extraction ability of ResNet. CBAM is an attention mechanism based on feature channels, which is used to strengthen the network's attention to different feature channels, thereby improving the performance of the network. The core idea is to redistribute the information between channels by learning an attention weight vector that represents the relationship between channels, thereby enhancing the network's attention to important feature channels.

[0109] In this embodiment, the load prediction model includes an electric load prediction model, a thermal load prediction model and a cooling load prediction model. The training process of these three models has been introduced above and will not be repeated here.

[0110] In some embodiments, the load forecasting model includes LSTM and temporal attention mechanism.

[0111] LSTM realizes multi-energy load forecasting. LSTM is a special recurrent neural network. Compared with traditional RNN, LSTM is more suitable for processing and predicting important events with longer intervals in time series. There is a strong time dependency between multi-energy load data and climate data, which means that different time samples under the same variable are highly correlated. The coupling relationship between multiple energies is analyzed through the above three methods, and the corresponding load forecasting model is designed according to the corresponding relationship to increase the prediction accuracy of the model.

[0112] The LSTM calculation process is as follows:

[0113] f t =σ(W fy y t-1 +W fx x t +b f )

[0114] i t =σ(W iy y t-1 +W ix x t +b i )

[0115] h t =tanh(W hy y t-1 +W hx x t +b h )

[0116] c t =i t *h t +f t *h t-1

[0117] o t =σ(W oy y t-1 +W ox x t +b o )

[0118] y t =o t +tanh(c t )

[0119] where f t 、i t 、h t 、c t , o t ,y tW represents the forgetting factor, input gate, unit candidate state, neuron state, output gate and prediction result output at time t. fy , W fx , W iy , W ix , W hy , W hx , W oy , W ox Corresponding to the weight matrix of the network at time t, y t-1 and x t They represent the output at time t-1 and the input at time t, respectively. f , b i , b h , b o represents the bias matrix. σ and tanh represent the Sigmond and tanh activation functions.

[0120] Although LSTM can realize multi-energy load forecasting, as the time series grows, it is difficult for the LSTM model to learn long time series, which makes important information easily ignored, reducing the accuracy of LSTM model prediction. Therefore, the present invention combines the time attention mechanism with the electricity, cooling and heating load forecasting models to realize the learning of important time series information.

[0121] By using the load prediction model provided by the present invention to predict the electric load, cooling load and heating load respectively, the prediction results are as follows: Figure 7-9 shown.

[0122] from Figure 7-9 It can be seen that the difference between the result predicted by the load forecasting model provided by the present invention and the actual result is small, and the prediction accuracy is high.

[0123] The embodiment of the present invention provides a prediction method, which predicts the multi-energy load within the preset time period by acquiring the target data within the preset time period. Among them, the target data covers the load data and climate data of multiple energy sources, which can reflect the usage of multiple energy sources within the preset time period. Then, since the load data of multiple energy sources are random and volatile, it is necessary to cluster the load data of multiple energy sources, so that the load data can be classified to ensure that accurate and consistent data sets can be obtained in subsequent analysis and prediction. In addition, the irregular load data can also be decomposed into relatively stable data components of different frequencies through variational mode decomposition, so that accurate load prediction can be achieved. Then, since there are many climate data, and each energy is related to different climates, it is necessary to find out which climate factors each energy is related to, so as to determine the prediction model and the input second processing data based on the correlation. Finally, the second data is input into the feature extraction model to obtain the high-dimensional time features of multiple energy sources, and the high-dimensional time features are input into the pre-built load prediction model to obtain the predicted load of various energy sources. The present invention first performs cluster analysis and variational mode processing on the load data, thereby improving the stability of the extracted data and the accuracy of the features. In addition, by adding an attention mechanism module to the feature extraction model and adding a time attention mechanism to the load forecasting model, the accuracy of the model can be further improved.

[0124] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0125] The following is an embodiment of the device of the present invention. For details not described in detail therein, reference may be made to the corresponding method embodiment described above.

[0126] Fig.10 The following is a schematic diagram showing the structure of a device for predicting multi-energy loads in an integrated energy system provided by an embodiment of the present invention. For ease of explanation, only the parts related to the embodiment of the present invention are shown, which are described in detail as follows:

[0127] like Fig.10 As shown, the prediction device 1000 for multi-energy load of the integrated energy system includes:

[0128] The data acquisition module 1100 is used to acquire target data within a preset time period, wherein the target data includes load data and climate data of various energy sources;

[0129] The first processing module 1200 is used to perform clustering processing and variational mode decomposition on load data of multiple energy sources to obtain first processed data;

[0130] The second processing module 1300 is used to perform correlation analysis on the first processed data and the climate data to obtain second processed data;

[0131] The prediction module 1400 is used to input the second data into the feature extraction model to obtain high-dimensional time features of multiple energy sources, and input the high-dimensional time features into a pre-built load prediction model to obtain predicted loads of various energy sources; wherein the feature extraction model includes an attention mechanism module, and the load prediction model includes a time attention mechanism.

[0132] In a possible implementation, the first processing module 1200 is used to cluster and divide the load data of multiple energy sources based on the elbow method to obtain first sub-data;

[0133] Perform variational mode decomposition on the first sub-data to obtain first processed data.

[0134] In a possible implementation, the second processing module 1200 is used to perform correlation analysis on the first processed data and the climate data to determine correlation data between the loads of multiple energy sources and the climate;

[0135] Based on the correlation data between the load of various energy sources and the climate, the first processed data and the climate data are eliminated to obtain the second processed data.

[0136] In a possible implementation, the center of the cluster division is 4.

[0137] In a possible implementation, the load data of the multiple energy sources include electric load data, cooling load data, and heating load data;

[0138] Climate data includes air pressure, wind speed, temperature, solar radiation, precipitation, humidity, and dew point temperature.

[0139] In a possible implementation, the correlation analysis includes any one or more of a Pearson correlation coefficient, a Spearman correlation coefficient, or a rank correlation coefficient;

[0140] The correlation data between the loads of various energy sources and the climate include: the electric load data is related to the air pressure, temperature, precipitation and dew point temperature; the heat load data is related to the air pressure, temperature, solar radiation and dew point temperature; the cooling load data is related to the air pressure, temperature, solar radiation, humidity and dew point temperature.

[0141] In a possible implementation, the data acquisition module 1100 is used to detect abnormal data using a standard score method and remove the abnormal data;

[0142] Use exponential smoothing to fill in the removed abnormal data and missing data to obtain the processed data;

[0143] The processed data are normalized.

[0144] In one possible implementation, the feature extraction model includes a deep residual network and an attention mechanism module.

[0145] In one possible implementation, the load forecasting model includes LSTM and a temporal attention mechanism.

[0146] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0147] Those of ordinary skill in the art will appreciate that the templates, units, and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0148] If the module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, it can implement the steps of the embodiment of the evaluation method for the adequacy of the distribution network in which each electric vehicle and distributed photovoltaic participate. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory, random access memory, electric carrier signal, telecommunication signal and software distribution medium.

[0149] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for predicting multi-energy loads of an integrated energy system, characterized in that: include: Acquiring target data within a preset time period, wherein the target data includes load data and climate data of multiple energy sources; Performing clustering processing and variational mode decomposition on load data of multiple energy sources to obtain first processed data; performing a correlation analysis on the first processed data and the climate data to obtain second processed data; The second data is input into a feature extraction model to obtain high-dimensional time features of multiple energy sources, and the high-dimensional time features are input into a pre-built load forecasting model to obtain predicted loads of various energy sources; wherein the feature extraction model includes an attention mechanism module, and the load forecasting model includes a time attention mechanism.

2. The method for predicting multi-energy loads of an integrated energy system according to claim 1, characterized in that: The clustering and variational mode decomposition of the load data of the multiple energy sources to obtain the first processed data includes: Clustering the load data of the multiple energy sources based on the elbow method to obtain first sub-data; Perform variational mode decomposition on the first sub-data to obtain the first processed data.

3. The method for predicting multi-energy loads of an integrated energy system according to claim 1 or 2, characterized in that: The performing correlation analysis on the first processed data and the climate data to obtain second processed data includes: Performing a correlation analysis on the first processed data and the climate data to determine correlation data between the loads of the multiple energy sources and the climate; Based on the correlation data between the loads of the various energy sources and the climate, the first processed data and the climate data are eliminated to obtain the second processed data.

4. The method for predicting multi-energy loads of an integrated energy system according to claim 2, characterized in that: The center of the clustering is 4.

5. The method for predicting multi-energy loads of an integrated energy system according to claim 1, characterized in that: The load data of the multiple energy sources include electric load data, cooling load data and heating load data; The climate data includes air pressure, wind speed, temperature, solar radiation, precipitation, humidity and dew point temperature.

6. The method for predicting multi-energy loads of an integrated energy system according to claim 5, characterized in that: The correlation analysis includes any one or more of the Pearson correlation coefficient, the Spearman correlation coefficient or the rank correlation coefficient; The data on correlation between the loads of the various energy sources and the climate include: electric load data is related to air pressure, temperature, precipitation and dew point temperature; thermal load data is related to air pressure, temperature, solar radiation and dew point temperature; and cooling load data is related to air pressure, temperature, solar radiation, humidity and dew point temperature.

7. The method for predicting multi-energy loads of an integrated energy system according to claim 1, characterized in that: Before clustering and performing variational mode decomposition on the load data of multiple energy sources, the method further includes: Use the standard score method to detect abnormal data and remove them; Use exponential smoothing to fill in the removed abnormal data and missing data to obtain the processed data; The processed data are normalized.

8. The method for predicting multi-energy loads of an integrated energy system according to any one of claims 1 to 7, characterized in that: The feature extraction model includes a deep residual network and an attention mechanism module.

9. The method for predicting multi-energy loads of an integrated energy system according to any one of claims 1 to 7, characterized in that: The load forecasting model includes LSTM and temporal attention mechanism.

10. A device for predicting multi-energy loads of an integrated energy system, characterized in that: include: A data acquisition module, used to acquire target data within a preset time period, wherein the target data includes load data and climate data of multiple energy sources; A first processing module is used to perform clustering processing and variational mode decomposition on load data of multiple energy sources to obtain first processed data; A second processing module, configured to perform a correlation analysis on the first processed data and the climate data to obtain second processed data; A prediction module is used to input the second data into a feature extraction model to obtain high-dimensional time features of multiple energy sources, and input the high-dimensional time features into a pre-built load prediction model to obtain predicted loads of various energy sources; wherein the feature extraction model includes an attention mechanism module, and the load prediction model includes a time attention mechanism.

Images