Comparative Analysis of Intelligently Tuned Support Vector Regression Models for Short Term Load Forecasting in Smart Grid Framework

A large amount of work has been taken place, if we talk about forecasting in the fields of power system. Various reforms in the existing techniques have proved to be helpful in providing guidance to researchers for developing efficient algorithms exhibiting greater accuracy. This paper presents three forecasting models viz. three-day-trained Support Vector Regression model and parameter optimized Support Vector Regression using Genetic Algorithm (SVRGA) and that using Particle Swarm Optimization (SVRPSO). Unlike existing models, these models accomplish accurate forecasting by optimizing the regularized structural risk function. The models make use of previous three days hourly load data for predicting next day hourly load. This paper performs a comparative study between GA and PSO on the grounds of optimization of the hyper-parameters of SVR model.