Optimalization Global Horizontal Irradiance Based On Weather Data Using Hybrid model Modified Decomposition FeedForward Neural Network

A novel hybrid modeling for optimized predictive of global horizontal irradiance (GHI) based on weather data with a combination of modified decomposition and feedforward Neural Network model. The proposed hybrid modeling is based on weather data, especially for the optimization of global horizontal irradiance. This hybrid modeling is a modified decomposition algorithm and feedforward neural network method. The modified Decomposition FNN (MD-FNN) hybrid method is designed to forecast GHI 60 minutes based on weather data for Photovoltaic generation. The novelty of this combination model is taking into account based on weather data. The model modified decomposition as a preprocessing to the feedforward neural network model. In the error statistical index of the MD-FNN method, the mean absolute percentage error (MAPE) is 1 %. Then, we compare this hybrid prediction method with the actual data. The composition and simulation results show that the MD-FNN method can be calculated with optimal accuracy each hour.