4　Improving the accuracy of power generation prediction and deterioration diagnosis of photovoltaic power plants using machine learning

Abstract

In order to utilize solar power generation, which is increasingly being introduced in large quantities, as a main power source in the future, it is important to suppress the uncertainty of power generation amount and the loss of power generation opportunities. It is necessary to predict the amount of power generated and to provide early warning of contamination and deterioration, which in some cases was difficult to judge using conventional PR values. We present the results of evaluating the effectiveness of quantity prediction and diagnostic methods using actual data.