Maximum joint entropy criterion-based least square support vector machine power prediction method

Abstract

The invention discloses a maximum joint entropy criterion-based least square support vector machine power prediction method. The method comprises the following steps of: 1, constructing an input dataset via same-term electricity consumptions of past three years and corresponding monthly average temperatures, and constructing an electricity sale quantity prediction model for to-be-predicted monthsby using a least square support vector machine method; 2, carrying out data preprocessing; 3, normalizing the input data set; 4, aiming at the history electricity consumptions, carrying out electricity sale quantity prediction by selecting a least square support vector machine model; 5, optimizing key parameters <sigma>0 and <sigma> of the model so as to determine an optimal parameter; 6, optimizing the key parameters <sigma>0 and <sigma>; 7, selecting a monthly relative error and an annual average error as evaluation indexes of the prediction result and calculating prediction precision at the time; and 8, carrying out prediction for at least 3-5 times and taking an average value. The method is capable of carrying out electricity sale quantity prediction on users.

Description

technical field [0001] The invention belongs to the technical field of power system electricity sales forecasting, and relates to a least squares support vector machine power forecasting method based on the maximum correlation entropy criterion. technical background [0002] Since the long-term transaction contract power of the power system is the planned value, and the supply and demand of power are affected by many random factors, it is difficult to predict accurately. Therefore, the high-precision power sales forecasting method has very critical theoretical significance and engineering value. [0003] In the actual electricity sales forecasting scheme, the current electricity sales companies generally use the historical electricity consumption data collected on the user side as the forecast basis. However, due to the generally low reliability of these data, the structure of historical electricity consumption data on the user side is not neat enough, coupled with the influ...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to propose a least squares support vector machine electricity forecasting method based on the maximum correlation entropy criterion, which solves the problem that the accuracy of the electricity sales forecast model is not high under the existing technical conditions, and it is difficult to meet the requirements of electricity sales transactions

Images