Solar Irradiance Forecasting Using Dynamic Ensemble Selection

Solar irradiance forecasting has been an essential topic in renewable energy generation. Forecasting is important task because it can improve the planning and operation of photovoltaic systems, resulting economic advantages. Traditionally, single models are employed this task. However, issues regarding selection inappropriate model, misspecification, or presence random fluctuations solar series result approach underperforming. This paper proposes a heterogeneous ensemble dynamic named HetDS, to forecast irradiance. For each unseen test pattern, HetDS chooses most suitable model based on pool seven well-known literature methods: ARIMA, support vector regression (SVR), multilayer perceptron neural network (MLP), extreme learning machine (ELM), deep belief (DBN), forest (RF), gradient boosting (GB). The experimental evaluation was performed with four data sets hourly measurements Brazil. proposed attained overall accuracy that superior terms five error metrics.