Neural network parallel modelling of broad domain nonlinear continuous mapping

First, we quantificationally discuss the necessity of blocking and parallel modelling of broad domain nonlinear continuous mapping based on neural network. Then, we give the effective neural network blocking and parallel modelling method to raise modelling quality and to shorten modelling time. The result is the embodiment and development of Tu Xuyan´s modelling ideal of decompound-compound for large-scale system. Experiments for nonlinear continuous mapping of 3D Mexican Straw Hat at broad domain and so on show that the NN block and parallel model is more precise than the direct NN model, and it is also faster to build the model.