Power system short-term load prediction method based on RBF-AR model

Abstract

The invention discloses a power system short-term load prediction method based on an RBF-AR model. The power system short-term load prediction method comprises the following steps: firstly, modeling ashort-term load of a power system by adopting an RBF-AR model; and then, adopting a stochastic gradient descent (SG) optimization algorithm to perform online optimization on parameters of the RBF-ARmodel, and selecting the optimal order of the RBF-AR model finally used for predicting the short-term load of the power system through a defined minimum information criterion. The RBF-AR model designed by the invention can effectively avoid the problems of more parameters, difficulty in estimation and poor real-time performance due to direct use of the RBF model, and the designed stochastic gradient descent (SG) algorithm for online optimization of RBF-AR model parameters has faster real-time performance than an offline estimation algorithm, and is more suitable for short-term load predictionof a power system.

Description

technical field [0001] The invention relates to the technical field of short-term load forecasting of electric power systems, in particular to a method for short-term load forecasting of electric power systems based on an RBF-AR model. Background technique [0002] Short-term load forecasting is of great significance to the reasonable dispatching of the power system. Accurate power load forecasting can effectively reduce the cost of power grid operation and significantly improve the reliability and economy of power system operation. Compared with long-term load forecasting, short-term load forecasting is mainly used to arrange power generation planning, and has higher real-time requirements. It often has the characteristics of fast change speed, large environmental impact, and strong nonlinearity. With the urgent requirements of power grid transformation and the rapid development of advanced intelligent algorithms, the research on advanced forecasting algorithms for short-te...

AI Technical Summary

Problems solved by technology

However, in practical applications, directly using the neural network model to predict the short-term load of the power system often has problems such as many parameters of the model, difficulty in accurate estimation, and poor real-time performance.

Images