Short-term load prediction method and system of LSTM neural network

Abstract

The invention discloses a short-term load prediction method and system for an LSTM neural network, and relates to the field of power system load prediction research. The method comprises the followingsteps: step 1, decomposing short-term load data through a variational mode decomposition method to obtain a load component; step 2, obtaining LSTM neural network parameters; step 3, establishing a prediction model through the load component and LSTM neural network parameters; and step 4, inputting to-be-predicted data into the prediction model to obtain a prediction result. According to the invention, the problems of modal aliasing, false IMF and the like can be solved, the purpose of avoiding adverse influence factors on prediction precision is achieved, and then the defect of the current algorithm in load prediction accuracy is effectively overcome.

Description

technical field [0001] The invention relates to the research field of power system load forecasting, in particular to a short-term load forecasting method and system of LSTM neural network. Background technique [0002] Load forecasting is one of the key challenges in formulating power supply plans and balancing power supply and demand in power grids. It is the basic work of power market operations and an integral part of power planning. Power system load forecasting can be divided into ultra-short-term, short-term, medium-term and long-term forecasting according to the length of forecasting time. In the prior art, for the load forecasting method, the multiple linear regression method has the advantages of simple structure and fast prediction speed, but when describing more complex problems, the accuracy is low; the Kalman filter method, the algorithm can be very good Solve the noise problem in the data, but this will also cause the load with large changes to be screened ou...

AI Technical Summary

Problems solved by technology

In the prior art, for the load forecasting method, the multiple linear regression method has the advantages of simple structure and fast prediction speed, but when describing more complicated problems, the accuracy is low; the Kalman filter method, the algorithm can be very good Solve the noise problem in the data, but this will also cause the load with large changes to be screened out, which will have a certain impact on the prediction results; the gray theory method, which requires less data and is convenient for calculation, but does not take into account the correlation The relationship between factors, so the error is too large; support vector machine, its generalization ability is good, the prediction accuracy is high, but the calculation time is too long, it is difficult to predict large-scale data; random forest method can handle high-dimensional data, its The generalization error is small, but for noisy data, it is prone to overfitting

Images