A grey forecasting model based on evolving neural network

Traditional GM (1,1) and Grey Model based on artificial neural network were described; the defects of them were discussed. In order to find an optimal network, genetic algorithm was applied. Aiming at the defects of the traditional GA, an improved GA was proposed. Then the network optimization was aided by improved GA, and network's connection weights were self-adaptively evolved. So a hybrid forecasting model was proposed: a grey forecasting model based on evolving neural network, which combines GM(1,1) with GA and ANN together. The simulations based on Matlab program show that this hybrid algorithm can offer a scientific and effective support for forecasting work.