Nonlinear combination forecasting approach based on WNN-RAGA

Building WNN nonlinear combination models by combining wavelet transform with neural network to supply a gap of single forecasting method, optimizing network parameters of WNN study by RAGA's abilities of whole optimization to resolve the problems which are appeared in the late phase of WNN network study algorithm. These problems contains that convergence rate is low, local least value is existed and training results are instable. The forecasting results of instance reveals a favorable forecasting capability based on WNN-RAGA nonlinear combination forecasting.