Find in Library

For characteristics of nonlinearity, time-varying volatility and nonstationarity to disturbance of hoisting system of PMLSM, an improved PMLSM speed-regulation system was designed on basis of established PMLSM d-q axis dynamic mathematical model and BP neural network PID control. The system modifies the fixed learning rate in BP neural network to a self-adaptive one, and adds momentum to reduce oscillation tendency in learning process, so it greatly improves convergence speed and avoid that network falls into a local minimum. The simulation result showed that using the improved BP neural network PID controller can make PMLSM speed-regulation system reduce adjusting time and overshoots greatly and accelerat response speed, and make the hoist system have better dynamic performance and robustness.