Load prediction method and device of integrated energy system, terminal equipment and storage medium

Abstract

The invention discloses a load prediction method and device for an integrated energy system, terminal equipment and a storage medium, and the method comprises the steps: carrying out the decoupling of the historical integrated load data of the integrated energy system according to the load types, generating the historical load data of each load type, selecting one target historical load data from the historical load data, and carrying out the load prediction, the method comprises the steps of generating target prediction load data according to the correlation between a target load type corresponding to the target load data and each load type, converting the target prediction load data according to the correlation between the target load type corresponding to the target load data and each load type, and generating prediction load data of each load type. Therefore, only the load data of one load type is predicted, so that the interference of other loads during prediction is avoided, otherwise, the target prediction load data is converted by using the correlation between the load types, and the prediction load data of other load types are generated; and the predicted comprehensive load data of the comprehensive energy system can be accurately predicted.

Description

Technical Field

[0001] The present invention relates to the technical field of load forecasting, and particularly to a load forecasting method, device, terminal device and storage medium for an integrated energy system. Background Art

[0002] In recent years, with the explosive growth of new energy power generation and the rapid increase in power consumption demand in China, the contradiction of power grid peak shaving has become increasingly prominent, and the peak shaving demand has continued to rise. Relying solely on the peak shaving capacity on the power supply side has been difficult to meet the peak shaving demand. As another major adjustable resource that can effectively participate in the peak shaving market, by predicting the regulation potential of multi-energy flow loads, it can help power grid companies accurately control the peak shaving market, effectively formulate peak shaving strategies, which is of great significance for alleviating the power grid peak shaving pressure, promoting new energy consumption, and ensuring the safe operation of the power grid.

[0003] An integrated energy system (IES) integrates various energy forms such as cooling, heating, and electricity, involves the conversion, distribution, and organic coordination of energy, and is an important physical carrier for realizing the multi-energy complementarity characteristics of the energy Internet. Through scientific scheduling among various energies within the system, goals such as efficient consumption of renewable energy, efficient utilization of integrated energy, and safe and economic energy use by users can be achieved. Since the cooling, heating, and electricity loads are not only related to their own historical data, there is also a correlation among the loads, showing strong coupling. Traditional load forecasting models, such as linear regression and autoregression, cannot analyze the correlation among various types of loads, resulting in interference from the correlation among various loads when they are applied to integrated energy load forecasting, and ultimately leading to large errors in the forecasting results. Summary of the Invention

[0004] An embodiment of the present invention provides a load forecasting method for an integrated energy system, which can accurately forecast the predicted integrated load data of the integrated energy system.

[0005] An embodiment of the present invention provides a load forecasting method for an integrated energy system, including:

[0006] Obtain the historical integrated load data and weather forecast data of the integrated energy system; wherein, the historical integrated load data is composed of historical load data of several load types;

[0007] Perform decoupling analysis on the historical integrated load data to generate the historical load data of each load type and a joint distribution function for characterizing the correlation of each load type, and determine a target historical load data from the historical load data;

[0008] Perform load forecasting according to the weather forecast data and the target historical load data to generate target forecast load data;

[0009] Generate predicted load data for several other load types except the target historical load data according to the target predicted load data and the correlation between the target load type corresponding to the target historical load data and each load type in the joint distribution function.

[0010] Generate the predicted comprehensive load data of the integrated energy system according to the predicted load data and the target predicted load data.

[0011] Further, the decoupling analysis of the historical comprehensive load data to generate the historical load data of each load type and the joint distribution function for characterizing the correlation between each load type includes:

[0012] Decouple the historical comprehensive load data to generate the historical load data of each load type;

[0013] Generate the marginal probability density function and Copula function corresponding to each historical load data according to the historical load data;

[0014] Use the maximum likelihood estimation method to generate the static correlation coefficient between the load types according to each Copula function;

[0015] Construct an evolution equation using the marginal probability density function and calculate the time-varying correlation coefficient between the load types according to the evolution equation;

[0016] Generate the joint distribution function for characterizing the correlation between each load type according to the time-varying correlation coefficient and the static correlation coefficient.

[0017] Further, the load forecasting to generate the target predicted load data according to the weather forecast data and the target historical load data includes:

[0018] Perform modal decomposition on the target historical load data to generate the modal components of the target historical load data;

[0019] Input the weather forecast data and the modal components into a preset load forecasting model, so that the load forecasting model generates the target predicted load data according to the modal components; wherein, the load forecasting model is jointly constructed by several support vector machines for load forecasting.

[0020] Further, the performing modal decomposition on the target historical load data to generate the modal components of the target historical load data includes:

[0021] Perform constrained variation on the target historical load data to generate a constrained variation function of the target historical load data;

[0022] Using Lagrange multipliers and quadratic penalty functions, convert the constrained variation function into an unconstrained variation function;

[0023] Iteratively optimize the unconstrained variation function to generate the modal components and center frequencies of the target historical load data.

[0024] Furthermore, the construction of the load prediction model includes:

[0025] Obtain a number of training load data of the load type and corresponding weather data, and construct a training data set;

[0026] Construct a number of support vector machines as base predictors, and assign a weight value to each of the base predictors;

[0027] According to the weight values, combine a number of the base predictors to construct an initial load prediction model;

[0028] Use the training data set to repeatedly perform iterative training operations on the initial load prediction model until the load prediction model is generated.

[0029] Furthermore, the iterative training operation includes:

[0030] Obtain the model to be optimized; initially, the model to be optimized is the initial load prediction model;

[0031] Input the training data set into the model to be optimized, so that the model to be optimized outputs the predicted load data to be evaluated, and the predicted data output by each of the base predictors;

[0032] According to the training data set and the load data to be evaluated, generate the mean absolute error, mean square error, and mean absolute percentage error of the load data to be evaluated;

[0033] Judge whether the mean absolute error, the mean square error, and the mean absolute percentage error all respectively meet the corresponding preset thresholds;

[0034] If so, use the model to be optimized as the load prediction model;

[0035] If not, adjust the weights of each of the base predictors according to the load data to be evaluated and the predicted data, and optimize the model parameters of each of the base predictors according to the predicted data to generate the model to be optimized required for the next round of iterative training operation.

[0036] Another embodiment of the present invention provides a load forecasting device for an integrated energy system, comprising:

[0037] A data acquisition module for acquiring historical integrated load data and weather forecast data of the integrated energy system; wherein, the historical integrated load data is composed of historical load data of several load types;

[0038] A load decoupling module for performing decoupling analysis on the historical integrated load data to generate historical load data of each load type, and a joint distribution function for characterizing the correlation between each load type, and determining a target historical load data from the historical load data;

[0039] A load forecasting module for performing load forecasting according to the weather forecast data and the target historical load data to generate target forecast load data;

[0040] A load conversion module for generating forecast load data of several other load types except the target historical load data according to the target forecast load data and the correlation between the target load type corresponding to the target historical load data and each load type in the joint distribution function;

[0041] A load generation module for generating forecast integrated load data of the integrated energy system according to the forecast load data and the target forecast load data.

[0042] Further, the load decoupling module performs decoupling analysis on the historical integrated load data to generate historical load data of each load type, and a joint distribution function for characterizing the correlation between each load type, including:

[0043] Performing decoupling on the historical integrated load data to generate historical load data of each load type;

[0044] Generating marginal probability density functions and Copula functions corresponding to the historical load data;

[0045] Using the maximum likelihood estimation method to generate static correlation coefficients between the load types according to the Copula functions;

[0046] Constructing an evolution equation using the marginal probability density function and calculating the time-varying correlation coefficient between the load types according to the evolution equation;

[0047] Generating a joint distribution function for characterizing the correlation between each load type according to the time-varying correlation coefficient and the static correlation coefficient.

[0048] Another embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, a load forecasting method for an integrated energy system as described in any one of the embodiments is implemented.

[0049] Another embodiment of the present invention provides a storage medium, which includes a stored computer program. When the computer program runs, the device where the storage medium is located is controlled to execute a load forecasting method for an integrated energy system as described in any one of the above embodiments.

[0050] By implementing the present invention, the following beneficial effects are achieved:

[0051] The present invention discloses a load forecasting method, device, terminal device, and storage medium for an integrated energy system. The method decouples the historical integrated load data of the integrated energy system according to the load type, generates the historical load data of each load type, selects a target historical load data from them for load forecasting to generate a target predicted load data, and then converts the target predicted load data according to the correlation between the target load type corresponding to the target load data and each load type to generate the predicted load data of each load type. Therefore, the present invention only forecasts the load data of one of the load types, avoiding interference from other loads during forecasting. Instead, it utilizes the correlation between the loads to convert the target predicted load data to generate the predicted load data of other load types, thereby accurately forecasting the predicted integrated load data of the integrated energy system. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is a schematic flowchart of a load forecasting method for an integrated energy system provided by an embodiment of the present invention.

[0053] Figure 2 is a schematic structural diagram of a load forecasting device for an integrated energy system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0054] To make the objectives, technical solutions, and advantages of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments in the present application belong to the scope of protection of the present application.

[0055] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this application belongs; the terms used herein are for the purpose of describing specific embodiments only and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above drawings are intended to cover non-exclusive inclusion.

[0056] In the description of the embodiments of this application, technical terms such as "first" and "second" are only used to distinguish different objects and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity, specific order, or primary-secondary relationship of the indicated technical features. In the description of the embodiments of this application, the meaning of "a plurality" is more than two, unless otherwise specifically defined.

[0057] Referring to "embodiment" herein means that the specific features, structures, or characteristics described in connection with the embodiment may be included in at least one embodiment of this application. The phrase appears in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art explicitly and implicitly understand that the embodiments described herein may be combined with other embodiments.

[0058] In the description of the embodiments of this application, the term "and / or" is merely a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " herein generally represents an "or" relationship between the associated objects before and after.

[0059] In the description of the embodiments of this application, the term "a plurality" refers to more than two (including two). Similarly, "multiple groups" refers to more than two groups (including two groups), and "multiple pieces" refers to more than two pieces (including two pieces).

[0060] In the description of the embodiments of this application, unless otherwise clearly specified and limited, technical terms such as "installation", "connection", "connection", "fixation", etc. should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or integrated; it can also be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two components or the interaction relationship between two components. For those of ordinary skill in the art, the specific meanings of the above terms in the embodiments of this application can be understood according to specific circumstances.

[0061] See Figure 1 , which is a schematic flow chart of a load forecasting method for an integrated energy system provided by an embodiment of the present invention, including:

[0062] S1. Obtain the historical comprehensive load data and weather forecast data of the integrated energy system; wherein, the historical comprehensive load data is composed of historical load data of several load types;

[0063] In a preferred embodiment of the present invention, the comprehensive load data of the integrated energy system is mainly composed of cooling load, heating load, and electrical load. However, due to the energy conversion, distribution, and organic coordination involved in the integrated energy system, the data of various loads in the comprehensive load data of the integrated energy system are intertwined and doped with each other. The weather forecast data is obtained through the weather forecast information released by relevant institutions, including: temperature, humidity, weather type, such as: rainy, cloudy, sunny, etc.

[0064] S2. Perform decoupling analysis on the historical comprehensive load data to generate the historical load data of each load type and a joint distribution function for characterizing the correlation of each load type, and determine a target historical load data from the historical load data;

[0065] In a preferred embodiment of the present invention, since the data of various loads in the comprehensive load data are intertwined and doped with each other and cannot be directly used for load forecasting, it is necessary to decouple the load data of various loads to avoid the mutual influence of various load data during subsequent load forecasting, resulting in inaccurate forecasting results. After decoupling, select the historical load data of a load type as a target historical load data.

[0066] It can be understood that a historical load data with the highest integrity can be selected from several historical load data as the target historical load data, or the historical load data of the load type with the highest correlation with other load types can be used as the target historical load data, or considering both integrity and correlation, a target historical load data can be selected.

[0067] Preferably, the performing decoupling analysis on the historical comprehensive load data to generate the historical load data of each load type and a joint distribution function for characterizing the correlation of each load type includes:

[0068] S21. Decouple the historical comprehensive load data to generate the historical load data of each load type;

[0069] S22. Generate the marginal probability density function and Copula function corresponding to each historical load data according to the historical load data;

[0070] S23. Use the maximum likelihood estimation method to generate the static correlation coefficient between the load types according to each Copula function;

[0071] S24. Construct an evolution equation using the edge probability density function, and calculate the time-varying correlation coefficient between the load types according to the evolution equation;

[0072] S25. Generate a joint distribution function for characterizing the correlation of each load type according to the time-varying correlation coefficient and the static correlation coefficient.

[0073] In a preferred embodiment of the present invention, each historical load data is normalized to the [0, 1] interval to ensure data unity; the kernel density estimation function is calculated using the maximum likelihood estimation (MLK) method to determine the edge density function of the multi-energy flow load sequence; to obtain the static correlation coefficient, the extreme value point of the likelihood function is calculated using the Copula function of the edge density; to obtain the dynamic correlation coefficient, the likelihood function is constructed using the dynamic Copula distribution, and the corresponding evolution equation parameters (ω, α, and β) are considered; the obtained maximum likelihood estimation value and the corresponding evolution equation parameters are substituted into the evolution equation parameters to calculate the required time-varying correlation coefficient.

[0074] Specifically, first, the historical load data, such as the load type corresponding to the electrical load as the target historical load data, and the cooling load and heating load are normalized to the [0, 1] interval to ensure data unity, forming the L P , L C sequence. Further, according to the basic theory of the Copula function, if the L P , L C sequence follows a certain distribution, then the functions f P (x n ), f C (x n ) are the edge probability distribution functions of L P , L C at the x n sampling point, and f(x n , y n ) is the joint distribution function. Then there exists a Copula function such that:

[0075]

[0076] Therefore, when studying the load correlation, its joint distribution function can be equivalently studied as the edge distribution functions of its components and the Copula function describing the correlation of N random variables, which can effectively reduce the complexity of the multivariate random probability model. Further, the maximum likelihood estimation method based on non-parametric kernel density (MLK) is used to estimate the correlation coefficient of the Copula function, forming the following equation:

[0077]

[0078] Among them, fP (x1) and f C (x2) represent the probability density functions of L P and L C respectively. K w is the kernel function, and T is the length of the variable sequence. Using the MLK method, the marginal distributions in the likelihood function can be replaced with the probability density functions f P (x1) and f C (x2) to obtain the static correlation coefficient S(θ) and the extreme points of the likelihood function as shown in the following formula, where c[f P (x1), f C (x2)] represents the Copula function.

[0079] S(θ) = ∑Inc[f P (x1), f C (x2)];

[0080]

[0081] Since the historical load data is a time series variable and the correlation between different load types at different times is different, to calculate the dynamic time-varying correlation coefficient sequence, a dynamic Copula function must be considered. In addition, the dynamic distribution parameters related to the variables and their corresponding evolution equations must be used to obtain the likelihood function. By constructing the dynamic Clayton copula function θ t characterizes the correlation between dynamic time series, as shown in the following formula, where ω, β, α are all evolution parameters and Λ is a restrictive function.

[0082]

[0083] Finally, according to the time-varying correlation coefficient and the static correlation coefficient, a joint distribution function for characterizing the correlation of each load type is generated.

[0084] S3. Perform load forecasting based on the weather forecast data and the target historical load data to generate target forecast load data;

[0085] Preferably, the performing load forecasting based on the weather forecast data and the target historical load data to generate target forecast load data includes:

[0086] S31. Perform modal decomposition on the target historical load data to generate modal components of the target historical load data;

[0087] Preferably, the performing modal decomposition on the target historical load data to generate modal components of the target historical load data includes:

[0088] S311. Perform constrained variation on the target historical load data to generate a constrained variation function of the target historical load data;

[0089] S312. Use the Lagrange multiplier and the quadratic penalty function to convert the constrained variation function into an unconstrained variation function;

[0090] S313. Perform iterative optimization on the unconstrained variation function to generate the modal components and the central frequencies of the target historical load data.

[0091] In a preferred embodiment of the present invention, the constraint condition of the constrained variation is that the sum of all modes is equal to the original signal, the decomposition sequence is modal components with a finite bandwidth having a central frequency, and the sum of the estimated bandwidths of each mode is minimized. The constrained variation expression is:

[0092]

[0093] where, u m = {u1, u2,..., u m} are the modal components for decomposing a certain type of load power sequence, is the square expression of the two-norm, δ(t) is the impulse function, is the first-order partial derivative with respect to time t, f(t) is a certain type of load power sequence, and m is the number of components at different time scales.

[0094] The expression of the unconstrained variation problem is:

[0095]

[0096] where, u m (t) = {u1(t), u2(t),..., u m (t)} are the modal components for decomposing a certain type of load power sequence at time t, ω m = {ω1, ω2,..., ω m} are the central frequencies of the modal components, α is the introduced quadratic penalty factor parameter, and λ(t) is the Lagrange multiplier at time t.

[0097] The expressions of the modal components and the central frequencies of the target historical load data are:

[0098]

[0099] where, τ is the noise tolerance, satisfying the fidelity of the decomposition of the target historical load data, and are the Fourier transforms of ums + 1(t), ui(t), f(t), and λ(t) respectively, is the L2 distance between the current frequency and the center frequency obtained in the s-th iteration, and α is the introduced quadratic penalty factor parameter. is the center frequency of each modal component after the (s + 1)-th iteration. is the Lagrange multiplier after the (s + 1)-th iteration, ui(t) is the i-th modal component at time t, and ums+1(t) is each modal component after the (s + 1)-th iteration.

[0100] S32. Input the weather forecast data and the modal components into a preset load prediction model, so that the load prediction model generates target predicted load data according to the modal components; wherein, the load prediction model is jointly constructed by several support vector machines for load prediction.

[0101] Preferably, the construction of the load prediction model includes:

[0102] S321. Obtain several training load data of the load type and the corresponding weather data, and construct a training data set.

[0103] S322. Construct several support vector machines as base predictors, and assign a weight value to each base predictor.

[0104] S323. Combine several base predictors according to the weight values to construct an initial load prediction model.

[0105] S324. Repeatedly perform iterative training operations on the initial load prediction model using the training data set until the load prediction model is generated.

[0106] Preferably, the iterative training operation includes:

[0107] S01. Obtain the model to be optimized; initially, the model to be optimized is the initial load prediction model.

[0108] S02. Input the training data set into the model to be optimized, so that the model to be optimized outputs predicted load data to be evaluated, and the predicted data output by each base predictor.

[0109] S03. Generate the mean absolute error, mean square error, and mean absolute percentage error of the load data to be evaluated according to the training data set and the load data to be evaluated.

[0110] S04. Determine whether the mean absolute error, the mean square error, and the mean absolute percentage error all satisfy the corresponding preset thresholds respectively.

[0111] S05. If so, use the model to be optimized as the load forecasting model;

[0112] S06. If not, adjust the weights of the base predictors according to the load data to be evaluated and the prediction data, and optimize the model parameters of each base predictor according to the prediction data to generate the model to be optimized required for the next round of iterative training operation.

[0113] In a preferred embodiment of the present invention, an SVM predictor is used, as shown in the following formula:

[0114]

[0115] K(x i ,x j ) = exp(-(x i -x j ) / 2σ 2 );

[0116] Among them, x i (i = 1, 2,..., n) is the relevant input vector of the prediction model, y i = {-1, 1} is the output vector, c is the penalty factor, and the value of the penalty factor c affects the structural risk of the model. When the value of α i is too large, the structural risk is greater; when the value of α i is too small, the model will be overly simplistic. K(x i ,x j ) is the kernel function, and the kernel function is selected according to needs. In this study, the radial kernel function is selected. σ is the kernel parameter of the radial kernel function, and its value also affects the effect of the model. If the value is too small, the generalization ability of the model will be weakened, and if the value is too large, the model will overfit.

[0117] Furthermore, use AdaBoost to synthesize weak SVM predictors to obtain a strong predictor. Specifically, set the maximum number of AdaBoost iterations T, initialize the sample weights, and after T iterative loops, obtain T base predictors and T corresponding weight coefficients. Finally, superimpose the T base predictors according to the weight coefficients to obtain the final strong predictor.

[0118] ① Given a training sample set S = {(x1, y1),..., (x N ,y N )} with labels, preset the number of loop iterations as T, error ε * , linear SVM parameters δ1 = {C1, C2, C3,...} and non-linear SVM parameters δ2 = {(C1, σ1), (C2, σ2),...} using the radial kernel function, and let δ = δ1 ∪ δ2.

[0119] ② Use AdaBoost to find each base predictor h1, h2,..., h r and their weights α1, α2,..., α r . Assume ψ(x) = [α1h1(x), α2h2(x),..., α r h r (x)], then the training set S = {[ψ(x1), y1],..., [ψ(x N ), y N}.

[0120] ③ Obtain the parameter δ of the SVM through the following steps i ∈δ.

[0121] Train the training sample set S and use the cross-validation optimization method to determine its error ε. If ε < ε * , then set ε * = ε, δ * = δ i .

[0122] ④ Set the parameter of the SVM to δ * , and output the trained strong predictor.

[0123] Further, through the following formula, calculate the mean absolute error, mean square error, and mean absolute percentage error of the load data to be evaluated;

[0124]

[0125] where, yi and are the true value and predicted value at time i respectively.

[0126] S4. According to the target predicted load data and the correlation between the target load type corresponding to the target historical load data and each of the load types in the joint distribution function, generate predicted load data of several other load types except the target historical load data;

[0127] S5. Generate the predicted comprehensive load data of the integrated energy system according to the predicted load data and the target predicted load data.

[0128] This embodiment provides a load forecasting method for an integrated energy system. By decoupling the historical integrated load data of the integrated energy system according to load types, historical load data of each load type is generated, and a target historical load data is selected from them for load forecasting to generate target forecast load data. Then, according to the correlation between the target load type corresponding to the target load data and each of the load types, the target forecast load data is converted to generate forecast load data for each of the load types. Therefore, in the present invention, by only forecasting the load data of one of the load types, interference from other loads during forecasting is avoided. Instead, using the correlation between load types, the target forecast load data is converted to generate forecast load data for other load types, and then the forecast integrated load data of the integrated energy system is accurately forecasted.

[0129] See Figure 2 , which is a schematic structural diagram of a load forecasting device for an integrated energy system provided by an embodiment of the present invention, includes:

[0130] A data acquisition module, configured to acquire historical integrated load data and weather forecast data of the integrated energy system; wherein, the historical integrated load data is composed of historical load data of several load types;

[0131] A load decoupling module, configured to perform decoupling analysis on the historical integrated load data to generate historical load data of each load type and a joint distribution function for characterizing the correlation between each load type, and determine a target historical load data from the historical load data;

[0132] A load forecasting module, configured to perform load forecasting according to the weather forecast data and the target historical load data to generate target forecast load data;

[0133] A load conversion module, configured to generate forecast load data for several other load types except the target historical load data according to the target forecast load data and the correlation between the target load type corresponding to the target historical load data and each load type in the joint distribution function;

[0134] A load generation module, configured to generate the forecast integrated load data of the integrated energy system according to the forecast load data and the target forecast load data.

[0135] Preferably, the load decoupling module performs decoupling analysis on the historical integrated load data to generate historical load data of each load type and a joint distribution function for characterizing the correlation between each load type, including:

[0136] Decouple the historical comprehensive load data to generate the historical load data of each load type;

[0137] Generate the marginal probability density function and Copula function corresponding to each historical load data according to the historical load data;

[0138] Use the maximum likelihood estimation method to generate the static correlation coefficient between the load types according to each Copula function;

[0139] Construct an evolution equation using the marginal probability density function, and calculate the time-varying correlation coefficient between the load types according to the evolution equation;

[0140] Generate a joint distribution function for characterizing the correlation of each load type according to the time-varying correlation coefficient and the static correlation coefficient.

[0141] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the attached drawings of the device embodiments provided by the present invention, the connection relationship between modules indicates that they have a communication connection, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0142] Those skilled in the art can clearly understand that for the sake of convenience and brevity, the specific working process of the device described above can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0143] Another preferred embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a load forecasting method for a comprehensive energy system as described in any one of the above embodiments.

[0144] The terminal device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device may include, but is not limited to, a processor and a memory.

[0145] The so-called processor may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the terminal device and connects all parts of the entire terminal device through various interfaces and lines.

[0146] The memory can be used to store the computer program. The processor realizes various functions of the terminal device by running or executing the computer program stored in the memory and calling the data stored in the memory. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function, etc.; the data storage area can store data created according to the use of the mobile phone, etc. In addition, the memory may include high-speed random access memory and may also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, at least one magnetic disk storage device, flash device, or other volatile solid-state storage devices.

[0147] Another preferred embodiment of the present invention provides a storage medium. The storage medium is a computer-readable storage medium, and the computer program is stored in the computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0148] The above are the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A load forecasting method for an integrated energy system, characterized in that: include: Acquire historical comprehensive load data and weather forecast data of the comprehensive energy system; wherein the historical comprehensive load data is composed of historical load data of several load types; Performing decoupling analysis on the historical comprehensive load data to generate historical load data of each load type and a joint distribution function for characterizing the correlation of each load type, and determining a target historical load data from the historical load data; Perform load forecasting based on the weather forecast data and the target historical load data to generate target forecast load data; Generate forecast load data of several other load types in addition to the target historical load data according to the target forecast load data and the correlation between the target load type corresponding to the target historical load data and each load type in the joint distribution function; The predicted comprehensive load data of the integrated energy system is generated based on the predicted load data and the target predicted load data.

2. A load forecasting method for an integrated energy system according to claim 1, characterized in that: The decoupling analysis of the historical comprehensive load data to generate the historical load data of each load type and a joint distribution function for characterizing the correlation of each load type includes: Decoupling the historical comprehensive load data to generate historical load data of each load type; Generate a marginal probability density function and a Copula function corresponding to each of the historical load data according to the historical load data; Using the maximum likelihood estimation method to generate static correlation coefficients between the load types according to the Copula functions; Using the marginal probability density function, constructing an evolution equation, and calculating the time-varying correlation coefficient between the load types according to the evolution equation; A joint distribution function for characterizing the correlation of each load type is generated according to the time-varying correlation coefficient and the static correlation coefficient.

3. A load forecasting method for an integrated energy system as claimed in claim 2, characterized in that: The step of performing load forecasting according to the weather forecast data and the target historical load data to generate target forecast load data includes: Performing modal decomposition on the target historical load data to generate modal components of the target historical load data; The weather forecast data and the modal components are input into a preset load forecasting model so that the load forecasting model generates target forecast load data according to the modal components; wherein the load forecasting model is jointly constructed by a number of support vector machines for load forecasting.

4. A method for load forecasting of an integrated energy system according to claim 3, characterized in that: The performing modal decomposition on the target historical load data to generate modal components of the target historical load data includes: Performing constrained variation on the target historical load data to generate a constrained variation function of the target historical load data; Using Lagrange multipliers and quadratic penalty functions, the constrained variational function is converted into an unconstrained variational function; The unconstrained variational function is iteratively optimized to generate modal components and center frequencies of the target historical load data.

5. A load forecasting method for an integrated energy system according to claim 4, characterized in that: The construction of the load forecasting model includes: Acquire several training load data of the load type and corresponding weather data to construct a training data set; constructing a plurality of support vector machines as base predictors, and assigning a weight value to each of the base predictors; According to the weight value, a plurality of the base predictors are combined to construct an initial load forecasting model; The initial load forecasting model is repeatedly trained using the training data set until the load forecasting model is generated.

6. A method for load forecasting of an integrated energy system according to claim 5, characterized in that: The iterative training operation comprises: Obtaining a model to be optimized; wherein, initially, the model to be optimized is the initial load forecasting model; Inputting the training data set into the model to be optimized, so that the model to be optimized outputs predicted load data to be evaluated and predicted data output by each base predictor; Generate a mean absolute error, a mean square error, and a mean absolute percentage error of the load data to be evaluated according to the training data set and the load data to be evaluated; Determine whether the mean absolute error, the mean square error, and the mean absolute percentage error all meet corresponding preset thresholds; If yes, the model to be optimized is used as the load forecasting model; If not, the weights of the base predictors are adjusted according to the load data to be evaluated and the prediction data, and the model parameters of the base predictors are optimized according to the prediction data to generate the model to be optimized required for the next round of iterative training operation.

7. A load forecasting device for an integrated energy system, characterized in that: include: A data acquisition module is used to acquire historical comprehensive load data and weather forecast data of the integrated energy system; wherein the historical comprehensive load data is composed of historical load data of several load types; A load decoupling module, used for performing decoupling analysis on the historical comprehensive load data, generating historical load data of each load type and a joint distribution function for characterizing the correlation of each load type, and determining a target historical load data from the historical load data; A load forecasting module, used for performing load forecasting according to the weather forecast data and the target historical load data, and generating target forecast load data; A load conversion module, configured to generate forecast load data of several other load types in addition to the target historical load data according to the target forecast load data and the correlation between the target load type corresponding to the target historical load data and each load type in the joint distribution function; The load generation module is used to generate predicted comprehensive load data of the integrated energy system according to the predicted load data and the target predicted load data.

8. A load forecasting device for an integrated energy system as claimed in claim 7, characterized in that: The load decoupling module performs decoupling analysis on the historical comprehensive load data to generate the historical load data of each load type and a joint distribution function for characterizing the correlation of each load type, including: Decoupling the historical comprehensive load data to generate historical load data of each load type; Generate a marginal probability density function and a Copula function corresponding to each of the historical load data according to the historical load data; Using the maximum likelihood estimation method to generate static correlation coefficients between the load types according to the Copula functions; Using the marginal probability density function, constructing an evolution equation, and calculating the time-varying correlation coefficient between the load types according to the evolution equation; A joint distribution function for characterizing the correlation of each load type is generated according to the time-varying correlation coefficient and the static correlation coefficient.

9. A terminal device, characterized in that: It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, it implements a load forecasting method for an integrated energy system as described in any one of claims 1 to 6.

10. A storage medium, characterized in that: The storage medium includes a stored computer program, wherein when the computer program is executed, the device where the storage medium is located is controlled to execute a load forecasting method for an integrated energy system as described in any one of claims 1 to 6.

Images