Research on MFBP Algorithm of Nerve Net Self-Adapting Controller

The nerve net self-adapting controller aiming at some AC servo system is designed on the basis of Mathematical Modeling. The controller includes two forward nerve nets with three layers. One is designed to identify the nerve net system named NNI and the other is to control the nerve net system named NNC.NNC use special learning algorithm to control system and at the same time improve its constringency pace through introducing new MFBP algorithm. The MFBP algorithm is programmed using MATLAB. The system’s Steady state error is no more than 0.2mil under the step response,speed response and accelerated response.According to the results of the simulation and experiments, the nerve net self-adapting controller has a high quality of static、dynamic performance and restrain the infection to system performance under hard circumstances like large inertia,variety load and huge change of torsion.