Multi-element load prediction method and device for integrated energy system

Abstract

The invention relates to a multi-element load prediction method and device for an integrated energy system, and the method comprises the steps: obtaining historical meteorological data and cooling, heating and power load data of the integrated energy system, and carrying out the data preprocessing; standardizing the data in the database, and dividing the data into a training set, a verification set and a test set according to a preset proportion; calculating a Spearman correlation coefficient between the cold and heat electric loads in each season, establishing a neural network load prediction model of the target season according to the Spearman correlation coefficient, and training the model based on the training set; and inputting the data in the test set into the trained neural network load prediction model, and outputting a cooling, heating and power load prediction result. According to the multi-element load prediction method of the integrated energy system, seasonal performance, nonlinearity and coupling performance of the multi-element load are fully considered, coupling information between loads is extracted and shared from different scales, and a more refined prediction model is established for each season.

Description

Technical Field

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

[0002] As a new energy supply model that integrates multiple energy carriers such as electricity, cold energy, and heat energy, the integrated energy system has become an effective solution to the problem of energy resource shortage. Through various energy conversion and storage devices, different energy flows can be converted and supplemented with each other, greatly improving energy utilization. Accurate load forecasting can help energy suppliers predict user needs in advance, promote the economic and low-carbon operation of the integrated energy system, and improve the response speed of the system. However, compared with the traditional single energy system, the more frequent and complex interactions of multiple energy sources in the integrated energy system bring great difficulties and challenges to accurate load forecasting.

[0003] In recent years, with the development of artificial intelligence technology, deep learning has gradually taken a dominant position in the field of load forecasting. The existing deep learning-based multi-load forecasting technology for integrated energy systems has the following defects: (1) Many forecasting models do not fully consider the coupling characteristics between the multi-loads in the integrated energy system, and cannot mine and share the coupling characteristic information between strongly correlated loads from multiple scales. In addition, the information sharing layer in some forecasting models will reduce the computational efficiency of the system; (2) Many existing forecasting models ignore the seasonal characteristics of the multi-loads in the integrated energy system and do not analyze the seasonal changes in the coupling relationship between the multi-loads, which leads to the establishment of a prediction model that is not sophisticated enough; (3) The existing joint prediction loss function weight optimization method is difficult to simultaneously balance the loss size and training speed of each task during training. Summary of the invention

[0004] In view of this, it is necessary to provide a multi-element load forecasting method for an integrated energy system to solve the above-mentioned defects of the prior art.

[0005] In order to solve the above problems, in a first aspect, an embodiment of the present invention provides a multi-element load forecasting method for an integrated energy system, comprising:

[0006] Obtain historical meteorological data and cooling, heating and electricity load data of the integrated energy system, perform data preprocessing, and store the preprocessed data in the database;

[0007] Standardize the data in the database and divide it into training set, validation set and test set according to the preset ratio;

[0008] Calculate the Spearman correlation coefficient between the cooling, heating and electricity loads in each season, and establish a neural network load forecasting model for the target season based on the Spearman correlation coefficient, and train the model based on the training set;

[0009] The data in the test set is input into the trained neural network load forecasting model, and the cooling, heating and electricity load forecasting results are output.

[0010] Preferably, the historical meteorological data include temperature, humidity, air pressure, wind speed, dew point and precipitation; the cooling, heating and electricity load data include cooling load, heating load and electricity load.

[0011] The data preprocessing step includes:

[0012] The outliers in the data were eliminated by using the quartile method.

[0013] Fill missing data values ​​using linear interpolation.

[0014] Preferably, the step of standardizing the data in the database includes:

[0015] Divide the data in the database into four parts according to season;

[0016] The data corresponding to each season are standardized respectively; the expression of standardized processing is:

[0017]

[0018] Where x and x' are the sample values ​​before and after standardization, μ and σ are the mean and standard deviation of the samples in the training set, respectively.

[0019] Preferably, the calculation of the Spearman correlation coefficient between the cooling, heating and electricity loads in each season, and establishing a neural network load forecasting model for the target season according to the Spearman correlation coefficient, includes:

[0020] The calculation formula of Spearman correlation coefficient is as follows:

[0021]

[0022] Where, d i = rg(x i )-rg(y i ); rg(x i ) and rg(y i ) are x i and i rank; n is the total number of samples; ρ s The closer the absolute value of is to 1, the stronger the correlation between x and y is. i andi Represent the i-th element in sequences x and y respectively.

[0023] According to the calculation results of the Spearman correlation coefficient, a neural network load forecasting model is built for the four seasons:

[0024] If the Spearman correlation coefficient of two loads in a certain season is greater than the preset threshold of 0.7, the interactive multi-scale convolution module is used as the core prediction unit for joint prediction;

[0025] If the seasonal Spearman correlation coefficients of one load and the other two loads are all less than the preset threshold of 0.7, the multi-scale feature fusion module is used as the core prediction unit for separate prediction.

[0026] Preferably, the neural network load forecasting model includes an interactive multi-scale convolution module, which is composed of a one-dimensional copy filling layer, a one-dimensional convolution layer, a LeakyRelu activation function, a discard layer, a one-dimensional convolution layer and a Tanh activation function;

[0027] The load prediction step of the interactive multi-scale convolution module includes:

[0028] For any two load sequences F in the input cooling, heating and electricity e With F c , firstly, coupled information sharing is performed through multiple rounds of interactive learning: through two different one-dimensional convolution modules ψ e and ψ c , load sequence F e With F c Mapped to F e1 and F c1 ; After exponential operation, F e1 and F c1 The above process can be expressed by the following formula, where represents the Hadamard product operation;

[0029]

[0030] Then, through a second set of different one-dimensional convolution modules φ e and φ c , load sequence and Mapped to F e2 and F c2 , and complete the second interaction by adding or subtracting each other:

[0031]

[0032] After completing multiple rounds of interactive learning, the third group of different one-dimensional convolution modules η e and η c For load sequence and Then, the feature addition and self-attention mechanism in the multi-scale feature fusion module are introduced to extract the multi-scale features of the two load sequences. Finally, the output features of the self-attention mechanism and the original input sequence are added by introducing the residual network to obtain the final output F' of the interactive multi-scale convolution module. e and F' c .

[0033] Preferably, when an interactive multi-scale convolution module is used as a core prediction unit for joint prediction, a loss function weight optimization method based on Laplace distribution is used to dynamically adjust the weights of the loss functions of each task during training.

[0034] Preferably, the calculation formula of the total loss function L of the neural network load forecasting model is:

[0035]

[0036] In the formula, is the average loss of task k in the tth training round, K is the total number of tasks, W is the parameter in the neural network, T is an adjustable hyperparameter, and λ k is the model output noise for task k.

[0037] Preferably, after inputting the data in the test set into the trained neural network load forecasting model, the method further comprises:

[0038] The output data of the neural network load forecasting model is de-normalized to obtain the cooling, heating and electricity load forecasting results.

[0039] In a second aspect, an embodiment of the present invention provides a multi-element load forecasting device for an integrated energy system, comprising:

[0040] An acquisition module is used to acquire historical meteorological data and cooling, heating and electricity load data of the integrated energy system, perform data preprocessing, and store the preprocessed data in a database;

[0041] The standardization processing module is used to standardize the data in the database and divide it into a training set, a validation set, and a test set according to a preset ratio;

[0042] The model building module is used to calculate the Spearman correlation coefficient between the cooling, heating and electricity loads in each season, and to establish a neural network load forecasting model for the target season based on the Spearman correlation coefficient, and to train the model based on the training set;

[0043] The load forecasting module is used to input the data in the test set into the trained neural network load forecasting model and output the cooling, heating and electricity load forecasting results.

[0044] In a third aspect, the present invention further provides an electronic device, including a memory and a processor, wherein:

[0045] The memory is used to store programs;

[0046] The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the multi-element load forecasting method for an integrated energy system as described in the embodiment of the first aspect of the present invention.

[0047] In a fourth aspect, the present invention further provides a computer-readable storage medium for storing computer-readable programs or instructions, which, when executed by a processor, can implement the steps in the multi-element load forecasting method for an integrated energy system as described in the embodiment of the first aspect of the present invention.

[0048] The method and device for predicting multiple loads in an integrated energy system provided by the present invention fully consider the seasonality, nonlinearity and coupling of multiple loads, can extract and share coupling information between loads from different scales, and establish a more refined prediction model for each season. For the joint prediction situation, the proposed loss function weight optimization method based on Laplace distribution can more robustly balance the loss size and training speed of each task at the same time, and avoid the occurrence of negative loss values. The present invention achieves an improvement in the prediction accuracy of multiple loads in an integrated energy system, makes up for the deficiency of insufficient utilization of the collected load big data, and provides data support for the subsequent economic and low-carbon operation of the integrated energy system. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A flow chart of a multi-element load forecasting method for an integrated energy system provided by the present invention;

[0050] Figure 2 A heat map of the Spearman correlation coefficients between the cooling, heating and electricity loads in each season of the integrated energy system provided by the present invention;

[0051] Figure 3 The structure diagram of the interactive multi-scale convolution module provided by the present invention;

[0052] Figure 4 Four possible neural network load forecasting model structures provided by the present invention;

[0053] Figure 5 A structural block diagram of a multi-element load forecasting device for an integrated energy system provided by the present invention;

[0054] Figure 6This is a structural block diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0055] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0056] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0057] The existing deep learning-based multi-load forecasting technology for integrated energy systems has the following defects: (1) Many forecasting models do not fully consider the coupling characteristics between the multi-loads in the integrated energy system, and cannot mine and share the coupling feature information between strongly correlated loads from multiple scales. In addition, the information sharing layer in some forecasting models will reduce the computational efficiency of the system; (2) Many existing forecasting models ignore the seasonal characteristics of the multi-loads in the integrated energy system and do not analyze the seasonal changes in the coupling relationship between the multi-loads, which leads to the established forecasting model being not sophisticated enough; (3) The existing joint forecasting loss function weight optimization method is difficult to simultaneously balance the loss size and training speed during the training of each task.

[0058] In view of this, the present invention provides a multi-load forecasting method for an integrated energy system, which fully considers the seasonality, nonlinearity and coupling of multi-loads, extracts and shares coupling information between loads from different scales, and establishes a more refined forecasting model for each season. The following will be explained and introduced through multiple embodiments.

[0059] Figure 1 The flowchart of the multi-element load forecasting method for the integrated energy system provided by the present invention is as follows. Figure 1 As shown, the multi-element load forecasting method for the integrated energy system includes steps S1 to S4, wherein:

[0060] Step S1, obtaining historical meteorological data and cooling, heating and electricity load data of the integrated energy system, performing data preprocessing, and storing the preprocessed data in a database.

[0061] Specifically, historical meteorological data include temperature, humidity, air pressure, wind speed, dew point and precipitation. Cold, heat and electricity load data include cold load, heat load and electricity load.

[0062] In this embodiment, the data preprocessing step includes:

[0063] The outliers in the data were eliminated by using the quartile method.

[0064] Fill missing data values ​​using linear interpolation.

[0065] Step S2, standardize the data in the database and divide it into a training set, a validation set and a test set according to a preset ratio.

[0066] Specifically, in this embodiment, the data in the database is divided into four parts according to seasons, specifically, four season libraries, wherein March to May is spring, June to August is summer, September to November is autumn, and December to February is winter.

[0067] The data of each seasonal database are standardized respectively; the expression of the standardized processing is:

[0068]

[0069] Where x and x' are the sample values ​​before and after standardization, μ and σ are the mean and standard deviation of the samples in the training set, respectively.

[0070] After the data is standardized, it is divided into a training set, a validation set, and a test set according to a preset ratio. Preferably, the division ratio can be set to 8:1:1.

[0071] Step S3, calculate the Spearman correlation coefficient between the cooling, heating and electricity loads in each season, and establish a neural network load forecasting model for the target season based on the Spearman correlation coefficient, and train the model based on the training set.

[0072] Among them, the Spearman correlation coefficient is the Spearman rank correlation coefficient named after Charles Edward Spearman. The Spearman correlation coefficient is used to measure the correlation between two variables without assuming that the relationship between them is linear or that the data follows any specific distribution. It is calculated based on the data rank (that is, the position of the data after sorting) rather than the original numerical value.

[0073] In step S3, the Spearman correlation coefficient between the cooling, heating and electricity loads in each season is calculated, wherein the calculation formula of the Spearman correlation coefficient is as follows:

[0074]

[0075] Where, d i = rg(x i )-rg(y i ); rg(x i ) and rg(yi ) are x i and i rank; n is the total number of samples; ρ s The closer the absolute value of is to 1, the stronger the correlation between x and y is.

[0076] Figure 2 The Spearman correlation coefficient heat map between the cooling, heating and electricity loads in each season of the comprehensive energy system provided by the present invention is Figure 2 It can be seen that in spring, the Spearman correlation coefficients between the electric load and the cooling load and between the cooling load and the heating load exceeded the preset threshold of 0.7; in summer, only the Spearman correlation coefficient between the electric load and the cooling load exceeded the preset threshold of 0.7; in autumn, the Spearman correlation coefficients between the three loads all exceeded the preset threshold of 0.7; in winter, only the Spearman correlation coefficient between the cooling load and the heating load exceeded the preset threshold of 0.7.

[0077] The neural network load forecasting model established in the embodiment of the present invention includes an interactive multi-scale convolution module. Figure 3 The interactive multi-scale convolution module structure diagram provided by the present invention is shown in FIG. Figure 3 ,The interactive multi-scale convolution module consists of a one-dimensional replication padding layer, a one-dimensional convolution layer, a LeakyRelu activation function, a dropout layer, a one-dimensional convolution layer, and a Tanh activation function. The load prediction steps of the interactive multi-scale convolution module include:

[0078] For any two load sequences F in the input cooling, heating and electricity e With F c , firstly, coupled information sharing is performed through multiple rounds of interactive learning: through two different one-dimensional convolution modules ψ e and ψ c , load sequence F e With F c Mapped to F e1 and F c1 ; After exponential operation, F e1 and F c1 The above process can be expressed by the following formula, where represents the Hadamard product operation;

[0079]

[0080] Then, through a second set of different one-dimensional convolution modules φ e and φ c , load sequence and Mapped to F e2 and F c2, and complete the second interaction by adding or subtracting each other:

[0081]

[0082] After completing multiple rounds of interactive learning, the third group of different one-dimensional convolution modules η e and η c For load sequence and Then, the feature addition and self-attention mechanism in the multi-scale feature fusion module are introduced to extract the multi-scale features of the two load sequences. Finally, the output features of the self-attention mechanism and the original input sequence are added by introducing the residual network to obtain the final output F' of the interactive multi-scale convolution module. e and F' c .

[0083] In this embodiment, according to the calculation results of the Spearman correlation coefficient, a neural network load forecasting model is built for each of the four seasons: if the Spearman correlation coefficient of two loads in a certain season is greater than the preset threshold value of 0.7, the interactive multi-scale convolution module is used as the core prediction unit for joint prediction. If the Spearman correlation coefficient of one load and the other two loads in the season are both less than the preset threshold value of 0.7, the multi-scale feature fusion module is used as the core prediction unit for separate prediction. Figure 4 The four possible neural network load forecasting model structures are shown in Figure 1. Figure 4 Four possible neural network load forecasting model structures provided by the present invention.

[0084] Depend on Figure 2 The Spearman correlation coefficient heat map between the cooling, heating and electricity loads in each season can be shown in Table 1.

[0085]

[0086] In this embodiment, when an interactive multi-scale convolution module is used as a core prediction unit for joint prediction, the present invention adopts a loss function weight optimization method based on Laplace distribution to dynamically adjust the weights of the loss functions of each task during training.

[0087] The derivation process of the loss function weight optimization method based on Laplace distribution will be described below.

[0088] Let f W (x) is the output of the prediction model with x as input and W as the model parameters, then the likelihood between f(x) and the observed value y is defined by the following Laplace distribution:

[0089]

[0090] Among them, λ represents noise, corresponding to the homoscedastic uncertainty of the current task. p(y|f W (x)) represents the conditional probability distribution of observing data y given the model f W (x), and Lp(y|f W (x), λ) is the log-likelihood function.

[0091] According to the maximum likelihood estimation, the parameters in the Laplace distribution are obtained by maximizing the log-likelihood of the model with respect to W and λ:

[0092]

[0093] Taking the case where joint predictions are made for all three types of loads as an example, assume that the three outputs y1, y2, and y3 of the model all follow the Laplace distribution:

[0094] p(y1,y2,y3|f W (x))

[0095] =p(y1|f W (x))·p(y2|f W (x))·p(y3|f W (x))

[0096] =Lp(y1|f W (x),λ1)·Lp(y2|f W (x),λ2)·Lp(y3|f W (x),λ3)

[0097] Then the loss function of the neural network load prediction model is equivalent to minimizing the following negative log-likelihood:

[0098]

[0099] Among them, L1(W)=|y1 - f W (x)| is the loss of the first task. λ k is the model output noise for task k, used to learn and determine the weight of L k (W).

[0100] In the above formula, the loss functions of each task are derived from the Laplace likelihood, which has a linear relationship with the absolute error, so that the model shows stronger robustness when facing outliers in the dataset. More generally, for K joint prediction tasks, there is:

[0101]

[0102] log2λk The role of is to act as a regular term to avoid noise λ K grows too much, but when λ k When <0.5, it may lead to the appearance of negative loss value. k A slight modification gives the following formula:

[0103]

[0104] In addition, the present invention also introduces another weight In order to balance the training speed of each task, the total loss function L of the neural network load forecasting model is calculated by the following formula:

[0105]

[0106] In the formula, is the average loss of task k in the tth training round, K is the total number of tasks, W is the parameter in the neural network, T is an adjustable hyperparameter, and λ k is the model output noise for task k, is the ratio of the average loss of task k in the t-1th and t-2th training rounds, is the ratio of the average loss of task i in the t-1th and t-2th training rounds, i∈[1,K]; is the weight.

[0107] By learning and calculating the task-related homoscedastic uncertainty and average loss variation, the weight optimization method of the loss function based on Laplace distribution proposed in the present invention can balance the loss size and training speed of each task at the same time. In addition, the 1-norm loss function that is linearly related to the absolute error can reduce the impact of outliers on the estimation results.

[0108] Step S4, input the data in the test set into the trained neural network load forecasting model, and output the cooling, heating and electricity load forecasting results.

[0109] Specifically, after the data in the test set is input into the trained neural network load forecasting model, the output data of the neural network load forecasting model is de-standardized to obtain the cooling, heating and electricity load forecasting results.

[0110] In this embodiment, in order to verify the superiority and effectiveness of the multivariate load forecasting method for the integrated energy system, the present invention is compared with two classic machine learning models: gradient boosting regression GBR, artificial neural network ANN and two commonly used deep learning models based on recurrent neural network architecture: long short-term memory recurrent neural network LSTM and bidirectional long short-term memory recurrent neural network BiLSTM on the Pytorch platform, and the MAPE performance index of each model on the test set is calculated as shown in Table 2.

[0111] Table 2

[0112]

[0113]

[0114] The calculation formula of MAPE performance index is as follows:

[0115]

[0116] In the formula, n is the total number of samples, y′ i and i are the predicted value and the actual value of the load respectively.

[0117] Referring to Table 2, the present invention performs best in 11 of the 12 load forecasting tasks, and can achieve the greatest improvement in forecasting accuracy in autumn when load fluctuations are strongest. Further calculations show that the present invention improves the average accuracy by 7.82% over BiLSTM, 6.28% over LSTM, 15.79% over ANN, and 22.49% over GBR.

[0118] The multi-element load forecasting method for an integrated energy system provided by the present invention fully considers the seasonality, nonlinearity and coupling of the multi-element loads, can extract and share the coupling information between loads from different scales, and establish a more refined forecasting model for each season. For the joint forecasting situation, the proposed loss function weight optimization method based on Laplace distribution can more robustly balance the loss size and training speed of each task at the same time, and avoid the occurrence of negative loss values. The present invention achieves an improvement in the accuracy of multi-element load forecasting for an integrated energy system, makes up for the deficiency of insufficient utilization of the collected load big data, and provides data support for the subsequent economic and low-carbon operation of the integrated energy system.

[0119] Figure 5 The structural block diagram of the multi-element load forecasting device for the integrated energy system provided by the present invention is shown in FIG. Figure 5 , the multi-element load forecasting device of the integrated energy system includes:

[0120] The acquisition module 501 is used to acquire the historical meteorological data and the cooling, heating and electricity load data of the integrated energy system, perform data preprocessing, and store the preprocessed data in the database;

[0121] The standardization processing module 502 is used to standardize the data in the database and divide it into a training set, a validation set and a test set according to a preset ratio;

[0122] The model building module 503 is used to calculate the Spearman correlation coefficient between the cooling, heating and electricity loads in each season, and to establish a neural network load forecasting model for the target season according to the Spearman correlation coefficient, and to train the model based on the training set;

[0123] The load prediction module 504 is used to input the data in the test set into the trained neural network load prediction model and output the cooling, heating and electricity load prediction results.

[0124] The integrated energy system multi-load prediction device provided by the present invention executes the integrated energy system multi-load prediction method provided by the above-mentioned embodiments through the above-mentioned modules. The integrated energy system multi-load prediction method has been described in detail in the above-mentioned embodiments, and this embodiment will not be repeated here.

[0125] The multi-load forecasting device for the integrated energy system provided by the present invention fully considers the seasonality, nonlinearity and coupling of multi-loads, can extract and share the coupling information between loads from different scales, and establish a more refined forecasting model for each season. For the joint forecasting situation, the proposed loss function weight optimization method based on Laplace distribution can more robustly balance the loss size and training speed of each task at the same time, and avoid the occurrence of negative loss values.

[0126] Figure 6 The structural block diagram of the electronic device provided by the present invention is as follows: Figure 6 As shown, the present invention also provides an electronic device, and the electronic device 600 can be a computing device such as a mobile terminal, a desktop computer, a notebook, a palm computer, and a server. The electronic device 600 includes a processor 601 and a memory 602, wherein the memory 602 stores a multi-element load forecasting program 606 for an integrated energy system.

[0127] In some embodiments, the memory 602 may be an internal storage unit of a computer device, such as a hard disk or memory of the computer device. In other embodiments, the memory 602 may also be an external storage device of a computer device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the computer device. Furthermore, the memory 602 may also include both an internal storage unit of a computer device and an external storage device. The memory 602 is used to store application software and various types of data installed on the computer device, such as program codes for installing the computer device. The memory 602 may also be used to temporarily store data that has been output or is to be output. In one embodiment, when the integrated energy system multi-load forecasting program 606 is executed by the processor 601, the following steps are implemented:

[0128] Obtain historical meteorological data and cooling, heating and electricity load data of the integrated energy system, perform data preprocessing, and store the preprocessed data in the database;

[0129] Standardize the data in the database and divide it into training set, validation set and test set according to the preset ratio;

[0130] Calculate the Spearman correlation coefficient between the cooling, heating and electricity loads in each season, and establish a neural network load forecasting model for the target season based on the Spearman correlation coefficient, and train the model based on the training set;

[0131] The data in the test set is input into the trained neural network load forecasting model, and the cooling, heating and electricity load forecasting results are output.

[0132] In some embodiments, the processor 601 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run program codes or process data stored in the memory 602, such as executing a multi-element load forecasting program for an integrated energy system.

[0133] This embodiment further provides a computer-readable storage medium on which a multi-component load prediction program for an integrated energy system is stored. When the multi-component load prediction program for an integrated energy system is executed by a processor, the following steps are implemented:

[0134] Obtain historical meteorological data and cooling, heating and electricity load data of the integrated energy system, perform data preprocessing, and store the preprocessed data in the database;

[0135] Standardize the data in the database and divide it into training set, validation set and test set according to the preset ratio;

[0136] Calculate the Spearman correlation coefficient between the cooling, heating and electricity loads in each season, and establish a neural network load forecasting model for the target season based on the Spearman correlation coefficient, and train the model based on the training set;

[0137] The data in the test set is input into the trained neural network load forecasting model, and the cooling, heating and electricity load forecasting results are output.

[0138] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

[0139] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-element load forecasting method for an integrated energy system, characterized in that: include: Obtain historical meteorological data and cooling, heating and electricity load data of the integrated energy system, perform data preprocessing, and store the preprocessed data in the database; Standardize the data in the database and divide it into training set, validation set and test set according to the preset ratio; Calculate the Spearman correlation coefficient between the cooling, heating and electricity loads in each season, and establish a neural network load forecasting model for the target season based on the Spearman correlation coefficient, and train the model based on the training set; The data in the test set is input into the trained neural network load forecasting model, and the cooling, heating and electricity load forecasting results are output.

2. The method for multi-element load forecasting of an integrated energy system according to claim 1, characterized in that: The historical meteorological data include temperature, humidity, air pressure, wind speed, dew point and precipitation; the cooling, heating and electricity load data include cooling load, heating load and electricity load. The data preprocessing step includes: The outliers in the data were eliminated by using the quartile method. Fill missing data values ​​using linear interpolation.

3. The multi-element load forecasting method for an integrated energy system according to claim 1, characterized in that: The step of standardizing the data in the database includes: Divide the data in the database into four parts according to season; The data corresponding to each season are standardized respectively; the expression of standardized processing is: Where x and x' are the sample values ​​before and after standardization, μ and σ are the mean and standard deviation of the samples in the training set, respectively.

4. The method for multi-element load forecasting of an integrated energy system according to claim 2, characterized in that: The Spearman correlation coefficient between the cooling, heating and electricity loads in each season is calculated, and a neural network load forecasting model for the target season is established according to the Spearman correlation coefficient, including: The calculation formula of Spearman correlation coefficient is as follows: Where, d i = rg(x i )-rg(y i ); rg(x i ) and rg(y i ) are x i and i rank; n is the total number of samples; ρ s The closer the absolute value of is to 1, the stronger the correlation between x and y is. According to the calculation results of the Spearman correlation coefficient, a neural network load forecasting model is built for the four seasons: If the Spearman correlation coefficient of two loads in a certain season is greater than the preset threshold, the interactive multi-scale convolution module is used as the core prediction unit for joint prediction; If the seasonal Spearman correlation coefficients of one load and the other two loads are all less than the preset threshold, the multi-scale feature fusion module is used as the core prediction unit for separate prediction.

5. The method for multi-element load forecasting of an integrated energy system according to claim 4, characterized in that: The neural network load forecasting model includes an interactive multi-scale convolution module, which is composed of a one-dimensional copy filling layer, a one-dimensional convolution layer, a LeakyRelu activation function, a discard layer, a one-dimensional convolution layer and a Tanh activation function; The load prediction step of the interactive multi-scale convolution module includes: For any two load sequences F in the input cooling, heating and electricity e With F c , firstly, coupled information sharing is performed through multiple rounds of interactive learning: through two different one-dimensional convolution modules ψ e and ψ c , load sequence F e With F c Mapped to F e1 and F c1 ; After exponential operation, F e1 and F c1 The above process can be expressed by the following formula, where represents the Hadamard product operation; Then, through a second set of different one-dimensional convolution modules φ e and φ c , load sequence and Mapped to F e2 and F c2 , and complete the second interaction by adding or subtracting each other: After completing multiple rounds of interactive learning, the third group of different one-dimensional convolution modules η e and η c For load sequence and Then, the feature addition and self-attention mechanism in the multi-scale feature fusion module are introduced to extract the multi-scale features of the two load sequences. Finally, the output features of the self-attention mechanism and the original input sequence are added by introducing the residual network to obtain the final output F' of the interactive multi-scale convolution module. e and F' c .

6. The method for multi-element load forecasting of an integrated energy system according to claim 4, characterized in that: When the interactive multi-scale convolution module is used as the core prediction unit for joint prediction, the loss function weight optimization method based on Laplace distribution is adopted to dynamically adjust the weights of the loss function of each task during training. The calculation formula of the total loss function L of the neural network load forecasting model is: In the formula, is the average loss of task k in the tth training round, K is the total number of tasks, W is the parameter in the neural network, T is an adjustable hyperparameter, and λ k is the model output noise for task k, is the weight.

7. The method for multi-element load forecasting of an integrated energy system according to claim 1, characterized in that: After inputting the data in the test set into the trained neural network load forecasting model, the method further includes: The output data of the neural network load forecasting model is de-normalized to obtain the cooling, heating and electricity load forecasting results.

8. A multi-element load forecasting device for an integrated energy system, characterized in that: include: An acquisition module is used to acquire historical meteorological data and cooling, heating and electricity load data of the integrated energy system, perform data preprocessing, and store the preprocessed data in a database; The standardization processing module is used to standardize the data in the database and divide it into a training set, a validation set, and a test set according to a preset ratio; The model building module is used to calculate the Spearman correlation coefficient between the cooling, heating and electricity loads in each season, and to establish a neural network load forecasting model for the target season based on the Spearman correlation coefficient, and to train the model based on the training set; The load forecasting module is used to input the data in the test set into the trained neural network load forecasting model and output the cooling, heating and electricity load forecasting results.

9. An electronic device, It is characterized in that comprising a memory and a processor, wherein: The memory is used to store programs; The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the multi-element load forecasting method for an integrated energy system as described in any one of claims 1 to 7 above.

10. A computer-readable storage medium, characterized in that: Used to store computer-readable programs or instructions, which, when executed by a processor, can implement the steps in the multi-element load forecasting method for an integrated energy system as described in any one of claims 1 to 7 above.

Images