Volume 10 Issue 4 ( December )

To guarantee maximum performance and reliability of solar plants, the probability of occurrence of improper operations should be very low. For that, a fault diagnosis expert system can be one of the solutions that can figure out the improper operations very precisely and suggest corrective actions. In this study, we will use a hybrid approach for fault detection, i.e., a knowledge-based and data-driven approach, i.e., fuzzy logic and machine learning. A fuzzy model will sense the real-time data and classify it into good or faulty. The fuzzy simulation model, thus, developed, will be used to generate a large amount of training data required for machine learning methods. Further clustering techniques are applied to the available database to determine which improper operation (fault) has the highest probability of occurrence.

Keywords: fuzzy logic; machine learning; clustering; classification.