Based on the study of Probabilistic Neural Networks (PNN), a new clustering method is proposed to optimize the parameters of neural network, perfectly respond to the fault importance and the distribution of fault sample, reasonably determine the number of t hidden neurons and a optimized structure and a high training efficiency are achieved.
By comparing the training efficiency of the two algorithms, it is concluded that the BP neural network algorithm based on adaptive learning rate is more efficient for simulating processing mapping relations in die casting.
Take the highest training efficiency as the target function, an assessment system for conceptual design of an advanced training aircraft has been established by using the multivariate SIMPLEX optimization method.
This paper, through the analysis of the ground air defense weapon system, sets up a factor system of the combat training efficiency, uses a judgment matrix rectification method in improving the determination of weights, and comprehensively evaluates the efficiency of the ground air defense weapon system combat training example by adopting the method of AHP combined with judgment matrix rectification.
Experiments indicate that, compared with BP network model, the training efficiency and precision of GA-ENN model are obviously improved, and the model is effective. It lays the basis of real-time properties identification in intelligent control of deep drawing process.
The neural networks is adopted to forecast Shanghai stock indexes. The result of emulation indicates that the neural networks has better global convergence and higher training efficiency and forecasting precision.