Research for Optimizing Rock Mass Mechanical Parameter Based on Improved Neural Network

In order to optimizing the rock mass mechanical parameters of Xi-Luodu hydropower station, the neural network is improved by coupling the partial least-squares regression. And so the coupling method has best explaining ability to the system, Solving the problem that neural network model is not stable and the calculation velocity is slow, and overcoming the bad effect of the many layers relativity among variables in system modeling. The results show that the improved neural network is the superiority to the only one method. The input layers of neural network decrease from 4 to 2, so to simplize network construction and strengthen network stability. The concise is 0.0002 when calculating times are less than 500, and 96% of forecast errors are less than 20%, and all predict errors are no more than 1%. So the improved CMAC neural network can provide a good idea to optimize rock mass mechanical parameters.