Design of Reliability Monitoring on Solar Power Plants Using Artificial Neural Network

The use of solar energy as electrical energy in Indonesia is still relatively small, accounting for 0.15GWp of a total of 207.8GWp capacity.The sustainability of renewable energy in Indonesia is an important aspect towards national energy security and one of the efforts is to create a real-time reliability monitoring system.Reliability is an important factor in the longevity of the components of a power plant.The purpose of this research is to create a real-time reliability monitoring system for a solar power plant (SPP) which is developed through two stages, modeling the SPP system and designing a reliability system.The SPP system model was built using an artificial neural network (ANN) which uses time and previous measurement data as input and three lags as variations.Measurement data is divided into training data and testing data with a ratio of 70:30.The ANN model for the reliability system is developed based on failure detection, which occurs when the predicted value deviates from the actual measurement by more than five percent.Each occurrence of failure will be detected and entered into Benard's approximation to determine the value of reliability.The results of the ANN model for the most accurate prediction system are achieved by using time and lag-3 voltage data as inputs.The resulting MSE value is 8.96310 -3 for training data and 9.53310 -3 for testing data.The SPP reliability value was 0.683 in the first year and continued to decrease by 0.376 in the second year.Preventive maintenance needs to be done when the reliability reaches 0.55 with a maintenance interval of 21 months.