Forecasting of Renewable Power and Household Load using ANN Technique

For the past few decades, predicting the power output of the renewable sources and load demand has been a challenging endeavor in the power industry. Forecasts of wind speed may enhance the quality of wind energy production by making this unpredictable renewable resource more efficient and economically viable. PV power and wind turbine power depends upon the environmental parameters like irradiance, temperature and wind speed. To precisely forecast the irradiance and wind speed, researchers have employed a number of methods; every method has its own advantages and disadvantages. This study utilizes a dataset consisting of 26,280 samples, representing the hourly measurements of various environmental parameters over the past three years. The parameter to be forecasted is derived from a dataset, with an Artificial Neural Network (ANN) employed to make accurate predictions of this parameter. The ANN implemented in this study employs the Levenberg-Marquardt training algorithm technique and has a layer size of 30. The data undergoes preprocessing and a comprehensive analysis. 20% of the dataset is used for each validation and testing, while 60% of the dataset is used for training. The statistical study showed that the ANN accurately predicted wind speed, load demand, temperature and irradiance. According to regression analysis, the respective outcomes are 97%, 93%, 99% and 99%.