A multi-series grey forecasting model based on neural network improved by genetic algorithm

Traditional Grey GM(1,1) Model had its defect when it was applied to forecast relative data series. The relationship between different data series can´t be reflected properly. In order to solve the problem, artificial neural network (ANN) is combined to forecast multi-series data. Then the network optimization is aided by improved genetic algorithm (GA). So the network weights and thresholds were self-adaptively evolved. Then a hybrid grey model combined with ANN and GA was put forward. Based on Matlab program, the simulation example shows that the hybrid algorithm improves the forecasting precision. It can provide effective help for forecasting work.