ADAPTIVE PARAMETER ADJUSTMENT ALGORITHM OF THE BP MODEL BASED ON NEURAL NETWORK CONTROL

摘要

To fix the disadvantages of the traditional BP network, we propose a BP network algorithm, which use Network A to control the learning rate and variable parameter of Network B. The learning rate and variable parameter are both the dynamic output of the BP model of Network A. This not only improves the convergence rate of the network, but also avoids the negative possibility of the network being met with a local minimum and enables Network B to escape the flat region of error. The results of the simulation experiment show that, compared with the traditional BP model, the A-B network model has better convergence and stability. This paper purposes to improve the convergence and learning rate.