CN-121983952-A - Method, system, equipment and medium for predicting multi-energy load of comprehensive energy system

CN121983952ACN 121983952 ACN121983952 ACN 121983952ACN-121983952-A

Abstract

The invention discloses a method, a system, equipment and a medium for predicting the multipotency load of a comprehensive energy system, wherein the method comprises the following steps of acquiring meteorological factor data and load history data of the comprehensive energy system, and forming meteorological factors corresponding to Pearson correlation coefficients meeting the conditions into an input characteristic set; searching the optimal decomposition layer number and penalty factor of the variation modal decomposition in the parameter space, performing variation modal decomposition on various load sequences by utilizing the optimal decomposition layer number and penalty factor to obtain an intrinsic modal component, combining the input feature set and the intrinsic modal component to construct a prediction model, and verifying the prediction model until the prediction model meets a set standard. According to the invention, meteorological factors are screened through the pearson correlation coefficient, redundant factors with low load correlation are removed, so that the dimension of input features is reduced, and the calculation complexity of model training is reduced.

Inventors

MIAO YU
LI XINGJIE
SONG ZIHONG
YOU XINYU

Assignees

贵州电网有限责任公司

Dates

Publication Date: 20260505
Application Date: 20251218

Claims (10)

1. The method for predicting the multipotency load of the comprehensive energy system is characterized by comprising the following steps of: Acquiring meteorological factor data and load historical data of a comprehensive energy system, calculating pearson correlation coefficients between each meteorological factor data and various load historical data, analyzing the pearson correlation coefficients, and forming meteorological factors corresponding to the pearson correlation coefficients meeting the conditions into an input feature set; Searching the optimal decomposition layer number and penalty factor of the variation modal decomposition in the parameter space, and performing variation modal decomposition on various load sequences by utilizing the optimal decomposition layer number and penalty factor to obtain intrinsic modal components; Combining the input feature set with the intrinsic mode components to construct a prediction model; and verifying the prediction model until the prediction model meets the set standard.
2. The method for predicting the multipotency load of an integrated energy system according to claim 1, wherein the calculating step of the pearson correlation coefficient includes: Calculating the average value of each meteorological factor data sequence and the average value of each load historical data sequence; Calculating the dispersion of each data point and the mean value of each data point in each meteorological factor data sequence, and calculating the dispersion of each data point and the mean value of each data point in each load history data sequence; Multiplying the dispersion of the meteorological factor data sequence with the dispersion of the load history data sequence at the corresponding moment point by point, and summing all products to obtain a covariance molecular term; Summing the deviations of the weather factor data sequences after the point-by-point squaring, and then squaring to obtain standard deviation components of the weather factor data sequences; Dividing the covariance numerator term by the product of the standard deviation component of the meteorological factor data sequence and the standard deviation component of the load history data sequence to obtain the pearson correlation coefficient.
3. The method for predicting the multi-energy load of an integrated energy system according to claim 2, wherein the step of searching the parameter space for the optimal decomposition level and penalty factor for the variant modal decomposition comprises: Searching is carried out in a parameter space by taking envelope entropy minimization as an objective function, wherein the number of decomposition layers K epsilon [3,10] and the penalty factor alpha epsilon [1000,5000] are in the parameter space.
4. The method for predicting the multipotency load of an integrated energy system of claim 3, wherein the searching in the parameter space comprises: Initializing a gray wolf population, and randomly generating an initial solution of a decomposition layer number K and a penalty factor alpha; Performing variation modal decomposition on each component of the decomposition layer number K and the penalty factor alpha, and calculating a corresponding envelope entropy as a fitness value; Sequencing individual gray wolves according to the fitness value, and determining the positions of alpha wolves, beta wolves and delta wolves; updating the position of the gray wolf population, and carrying out iterative search until the maximum iteration times or envelope entropy convergence is reached; and outputting the optimal decomposition layer number K and the optimal penalty factor alpha corresponding to the minimum envelope entropy.
5. The method for predicting the multipotency load of an integrated energy system of claim 4, wherein the step of constructing a prediction model comprises: combining the input feature set with the intrinsic mode components to form input data of a prediction model; Designing and improving a sparrow searching algorithm, introducing a random walk mechanism into a sparrow position updating formula, and setting a step attenuation rule as ; Optimizing the kernel parameters of the least square support vector machine by adopting the improved sparrow search algorithm with the aim of minimizing the prediction error Wherein ; The kernel function of the least squares support vector machine adopts a radial basis function, and the expression is as follows: ; In the formula, For two data points And The kernel function value, sigma is the kernel parameter, Data points And Euclidean distance between them.
6. The method for predicting the multipotency load of an integrated energy system of claim 5, wherein said step of randomly walk the mechanism comprises: initializing iteration point, maximum iteration times and initial step length Convergence accuracy; Randomly generating n-dimensional direction vectors , wherein, ; Carrying out standardization processing on the direction vector D; triggering step attenuation when the objective function value is not improved by continuous X iterations, executing 。
7. The method for predicting the multi-energy load of an integrated energy system according to claim 6, wherein the calculation formula of the envelope entropy is: ; In the formula, For envelope entropy, N is the total number of envelope magnitudes, Is the ratio of the ith envelope magnitude to the sum of all envelope magnitudes.
8. A comprehensive energy system multipotency load prediction system, applying the method of any of claims 1-7, comprising: The feature screening module is used for acquiring weather factor data and load history data of the comprehensive energy system, calculating pearson correlation coefficients between each weather factor data and various load history data, analyzing the pearson correlation coefficients, and forming an input feature set by weather factors corresponding to the pearson correlation coefficients meeting the conditions; The parameter optimization decomposition module is used for searching the optimal decomposition layer number and penalty factors of the variation modal decomposition in the parameter space, and performing variation modal decomposition on various load sequences by utilizing the optimal decomposition layer number and penalty factors to obtain intrinsic modal components; the prediction model construction module is used for combining the input feature set with the intrinsic mode components to construct a prediction model; And the verification module is used for verifying the prediction model until the prediction model meets the set standard.
9. An electronic device, comprising: A memory and a processor; The memory is configured to store computer-executable instructions that, when executed by the processor, implement the steps of the integrated energy system multi-energy load prediction method of any one of claims 1 to 7.
10. A computer readable storage medium storing computer executable instructions which when executed by a processor perform the steps of the integrated energy system multi-energy load prediction method of any one of claims 1 to 7.

Description

Method, system, equipment and medium for predicting multi-energy load of comprehensive energy system Technical Field The invention relates to the technical field of energy system prediction, in particular to a method, a system, equipment and a medium for predicting a multi-energy load of a comprehensive energy system. Background By combining the advantages and disadvantages of multiple energy sources such as cold, heat, electricity, gas and the like, the cascade efficient configuration and comprehensive utilization of the energy sources are realized, and therefore the cascade efficient configuration and comprehensive utilization of the energy sources become one of the important directions of current energy source transformation. The accurate prediction of the multi-energy load is becoming a basic link of the scheduling operation of the modern comprehensive energy system, and the embodied prediction precision directly determines the economical efficiency and the safety of the system. However, due to the influence of various factors such as meteorological conditions, user behaviors and the like on various loads in the comprehensive energy system, the comprehensive energy system has the characteristics of coupling and nonlinearity, and great difficulty is brought to accurate prediction of the comprehensive energy system. The current commonly used prediction method mostly adopts all meteorological factors as input features, so that the input dimension is too high, the model training complexity is increased, the redundant features can interfere the prediction result, and the prediction precision is reduced. By means of the mode decomposition method of the variation, the complex load sequence can be skillfully decomposed into a plurality of relatively intrinsic mode components, so that the rule and the characteristics of load fluctuation are better revealed. However, the number of decomposition layers and penalty factors of the traditional variational modal decomposition are usually set by relying on manual experience, and lack of objective parameter selection criteria, so that modal aliasing or excessive decomposition is easy to occur. Disclosure of Invention The present invention has been made in view of the above-described problems occurring in the prior art. Therefore, the invention provides a method, a system, equipment and a medium for predicting the multi-energy load of a comprehensive energy system, which solve the problems that in the existing multi-energy load prediction, the input characteristic redundancy leads to overhigh dimensionality, the decomposition effect is unstable due to the fact that the decomposition parameters of a variation mode are selected depending on experience, and the adaptability of a single prediction model to load fluctuation is insufficient. In order to solve the technical problems, the invention provides the following technical scheme: The invention provides a multi-energy load prediction method of a comprehensive energy system, which comprises the following steps of obtaining meteorological factor data and load history data of the comprehensive energy system, calculating pearson correlation coefficients between each meteorological factor data and each load history data, analyzing the pearson correlation coefficients, forming an input feature set by meteorological factors corresponding to the pearson correlation coefficients meeting the conditions, searching an optimal decomposition layer number and penalty factors of variation modal decomposition in a parameter space, carrying out variation modal decomposition on each load sequence by utilizing the optimal decomposition layer number and penalty factors to obtain intrinsic modal components, combining the input feature set and the intrinsic modal components to construct a prediction model, and verifying the prediction model until the prediction model meets the set standard. The method comprises the steps of calculating the mean value of each meteorological factor data sequence and the mean value of each type of load history data sequence, calculating the dispersion of each data point in each type of load history data sequence and the mean value of each data point, multiplying the dispersion of the meteorological factor data sequence and the dispersion of the load history data sequence at corresponding moments point by point, summing all products to obtain covariance molecular terms, summing the dispersion of the meteorological factor data sequences point by point, squaring, summing the square again to obtain standard deviation components of the meteorological factor data sequence, summing the dispersion of the load history data sequence point by point, squaring again to obtain the standard deviation components of the load history data sequence, dividing the covariance molecular terms by the product of the standard deviation components of the meteorological factor data sequence and the standard deviation components of the load