Certain algebraic criteria for design of hybrid neural network models with applications in renewable energy forecasting

Abstract

In line to meet the energy exigency, renewable energies like wind and solar receive remarkable popularity, expeditious enlargement of power generation from the wind and solar energy entails acute forecasting of wind speed and solar irradiance, therefore, it has been an intensive research field in recent and past years. The importance pertaining to the planning and control of a solar farm, wind farm, and energy system are based on the forecast of the spot on the solar irradiance and wind speed forecasting. In the past decades, numerous researchers suggested various approaches for the wind speed and solar irradiance forecasting models, but still, an exact wind speed and solar irradiance prediction are of high thrust. There is a possibility of over fitting or under fitting occurrence due to the random selection of hidden neurons in Artificial Neural Networks (ANN) model. This research work is concerned with development of novel and hybrid forecasting models for the chosen applications and formulating suitable hidden neurons for the proposed various artificial neural networks to address the existing lacuna. This thesis presents comparative performance analysis of the wind speed forecasting application based on the six artificial neural network models, namely Back Propagation Neural Network (BPN), Multi-Layer Perceptron Neural Network (MLPN), Radial Basis Function Neural Network (RBFN), ELMAN Neural Network (EN), Improved Back Propagation Neural Network (IBPN) and Recursive Radial Basis Function Neural Network (RRBFN) with 6 inputs based forecasting models to forecast different time scale wind speed forecasting newline