Time-Series Energy Prediction using Hidden Markov Model for Smart Solar System

Solar energy is a one of the cleanest renewable energy resource, it does not affect greenhouse atmosphere. Presently, the photovoltaic solar energy characterizes as a third-biggest resource of renewable energy following to the hydro energy and wind energy. It results solar energy is fastest growing source of energy in the world. The management of photovoltaic systems is essential to increase the efficiency of solar system. The proposed system implemented in two phases, first is a panel level monitoring system and second is a solar power prediction system. We used Internet of Things for supervising solar power generation, it significantly enhance the performance, monitoring and maintenance of the solar plant. The solar monitoring system collects several constraints being evaluated by sensors to analyze the panel level performance. Monitoring system keeps track on solar power generation with environmental conditions, like temperature and humidity of particular location. This monitoring system forecast the faults as well as the panel's dead state with time parameters. Second phase works on the solar power prediction with the help of Hidden Markov Model. It gives accurate prediction considering correlation of the first value to next value in time-series. The output is then compared with existing prediction module that shows proposed module provides good accuracy of prediction.