The Effect of Data Pre-Processing Techniques on the Accuracy of Medium-Term Solar Power Forecasting

The increasing global emphasis on renewable energy has led to a significant increase in the use of photovoltaic power stations. Accurate solar power forecasting is essential for integrating the intermittent energy source into the grid. This ensures stability and the optimisation of energy management. Current solar power forecasting methods face challenges such as medium-term forecasting accuracy. This project aims to enhance the accuracy of medium-term photovoltaic power forecasting in Australia. This is achieved by investigating the impact of specific pre-processing techniques on multiple machine learning models. A key finding in the project is the superior performance of deleting the rows of missing data. This method improves the accuracy of medium-term power forecasts across multiple machine learning models and datasets.These findings enhance medium-term solar power forecasting accuracy, contributing to improved grid stability and energy management as reliance on intermittent renewable sources like photovoltaic continue to grow.