A combination of a dynamic Diagonal Recurrent Neural Network (DRNN) PID controller with a traditional PID controller applying in the PMSM servo system is designed in this paper. The servo system with traditional PID controller can't achieve a perfect performance under variable parameters and torque disturbance. DRNN has a good learning ability with a simple and recurrent structure, so it is suitable for controlling complicated servo system. In DRNN the matrix is transformed to the diagonal matrix, which greatly simplifies the computation, so it is very suitable for the real-time control system. A dynamic BP (DBP) algorithm is used in the DRNN controller to achieve a fast convergency. Simulation results show the compound control method can improve the dynamic response performance and enhance the static precision compared to the traditional PID controller.
[1]
Jamshid Ghaboussi,et al.
Neural networks for structural control of a benchmark problem, active tendon system
,
1998
.
[2]
Kwang Y. Lee,et al.
Diagonal recurrent neural network based control using adaptive learning rates
,
1992,
[1992] Proceedings of the 31st IEEE Conference on Decision and Control.
[3]
Tommy W. S. Chow,et al.
A recurrent neural-network-based real-time learning control strategy applying to nonlinear systems with unknown dynamics
,
1998,
IEEE Trans. Ind. Electron..
[4]
Yancheng Liu,et al.
The Design of Position Estimator for PMSM by Using Diagonal Recurrent Neural Network
,
2007,
2007 IEEE International Conference on Automation and Logistics.