Supportive Vector Regression (SVR) Hybridization and Evolutionary Genetic Algorithm in Modeling for Prediction / With Practical Application

Abstract

The forecasting of electrical loads critically supports energy management policy decisions. The aim of this study is to develop methods for predicting electrical load, as vector regression has been applied to predict electrical load, and the quality and stability of the SVR model depends greatly on the selection of optimal parameters, and the study proposes a new approach: Evolutionary SVR algorithms that solve problems of optimizing all parameters (SVR) To evaluate the values ​​of three typical parameters, the supporting vector regression was combined with the genetic algorithm and chaotic genomic algorithm (SVR), and the electrical load 1980-2019 was used as a dataset, and compared the results (GASVR) with (CGASVR) to choose the best form for predicting the electrical load where the results show that more superior and efficient CGASVR model based on MSE, MAE, MAPE and MPE criteria to predict validity.