Short-period combined load prediction method

Abstract

The invention discloses a short-period combined load prediction method which comprises the following steps: predicting a load by using at least two different single model prediction methods according to initial load data of a power grid to obtain different load data prediction results; weighting and summing the prediction results obtained by using the prediction methods to obtain a combined prediction result; determining optimal weights obtained by using the prediction methods by adopting a genetic algorithm, so as to obtain the weight with smallest error sum of squares as a solution, and finally weighting and summing predicting values of the prediction methods according to a weight coefficient of the solution to obtain a load prediction result. According to the method, the defects of high dependency of a conventional combined prediction model on an initial value, non convergence and the like are overcome, the prediction accuracy is high, the algorithm is simple, and the universality is relatively good.

Description

technical field [0001] The invention relates to power system load forecasting technology, in particular to a short-term combined load forecasting method, belonging to the fields of power system planning and economic dispatch. Background technique [0002] In the fields of power system planning, economic dispatch, stable operation and optimal control, load forecasting is of great significance, which determines the reasonable arrangement of power generation, transmission and distribution. The accuracy of load forecasting is directly related to the safe and economical operation of the power system, the development of the national economy and many other aspects. Traditional power load forecasting methods include Kalman filter, Box-Jenkins method, regression method, decomposition model, and climate (mainly temperature, followed by humidity, wind speed, etc.) identification methods. [0003] Due to different modeling mechanisms and starting points, there are different load foreca...

AI Technical Summary

Problems solved by technology

Although the existing literature has given several methods for estimating the optimal weight coefficients, they often fail to obtain better results in practice because of the strong dependence of the algorithm on the initial value, slow convergence, and easy to fall into local minimum. application

Images