1　Estimation of the minimum PV power generation within a spot using clustering based on the clear-sky index and a neural network

Abstract

As the adoption of renewable energy continues to expand, securing supply-demand balancing capacity in the power system has become a critical challenge. Solar power generation is highly dependent on weather conditions, leading to significant short-term fluctuations that can affect the stability of the power system. This study aims to assess short-term fluctuations in solar power generation to ensure sufficient adjustment capacity within the power system.

Specifically, we cluster the solar power generation data, estimated from weather data at a 5-minute granularity, based on the clearness index using the x-means method. Subsequently, we use neural networks within each cluster to predict the minimum power generation over a 30-minute period at a 1-minute granularity.

This approach provides insights into the characteristics of short-term fluctuations in solar power generation and aims to contribute to ensuring flexibility in the supply-demand adjustment market. It is expected that this method will contribute to improving the overall stability of the power system.