论文信息 - Optimization techniques for the design of a neural predictive controller

Optimization techniques for the design of a neural predictive controller

Abstract In this paper, we present three different optimizing methods for the design of an external recurrent neural network based Smith predictive controller to compensate for large time-delays in nonlinear processes. These optimizing techniques are respectively the gradient descent method, the Newton-Raphson algorithm, and the method of Levenberg-Marquardt. The implementation of these algorithms is described. An application of these algorithms to a simulated digester process is also presented.

Yonghong Tan | Achiel R. Van Cauwenberghe | A. Cauwenberghe | Yonghong Tan

[1] R. De Keyser,et al. Auto-tuning PID control using neural predictor to compensate large time-delay , 1994, 1994 Proceedings of IEEE International Conference on Control and Applications.

[2] David A. Wismer,et al. Introduction to nonlinear optimization , 1978 .

[3] O Smith,et al. CLOSER CONTROL OF LOOPS WITH DEAD TIME , 1957 .

[4] Rüdiger Schmidt. Advances in Nonlinear Parameter Optimization , 1982 .

[5] Graham C. Goodwin,et al. Adaptive filtering prediction and control , 1984 .