Anomaly Detection Mechanism for Solar Generation using Semi-supervision Learning Model

Solar is an important energy resource at present, and thus how to generate power efficiently by using solar is the crucial research topics in next generation power system. Among these research topics, managing and maintaining the solar panels for avoiding the situation which cannot generate power due to damage is also an interesting issue. Because the cost of developing the solar plant is expensive and needing the extra-cost to maintain solar, how to maintain the solar panels effectively is another important issue. In this study, an anomaly detection mechanism with using the semi-supervision learning model is proposed to pre-identify whether the solar panel will occur the abnormal events or not. In the anomaly detection mechanism, this study uses the clustering algorithm to filter the normal events, and then adopts the neuron network model, Autoencoder, to develop the classificator. This study takes the data collected from a 500kW solar power plant to train models and verify the feasibility of the proposed anomaly detection mechanism.