An Artificial Neural Network Based Forecasting Strategy for Estimating Weather parameters: Application for Sizing Stand-alone Renewable Power System

摘要

At present, fossil fuels (Non-renewable sources) are the main source for generation of Electricity in India. For nature friendly electric power generation, it is the need of time that we should develop highly efficient and Reliable power system based on these sustainable energy sources that can produce total energy demand. Therefore main focus of electric power generation is from photovoltaic(PV), Hydro and wind power systems. Renewable energy systems need prior weather information to plan their control activities and to manage power scheduling and dispatching problems efficiently. This paper proposes an Artificial Neural Network based weather prediction model that uses hourly weather data for training of neural networks and this model provides accurate short-term weather information using back-propagation algorithmic rule such as leven-berg Marquardt & Bayesian regularization, with standard NARX feed forward network built in MATLAB. This method provides very good results to predict up to one week ahead of weather information containing various parameters like solar irradiance, Temperature, wind speed and relative humidity.