Experiments with a neural controller
暂无分享,去创建一个
A neural controller is proposed that combines a simple feedback controller with a multilayer perceptron feedforward controller. Diffusive backpropagation of feedback error and genetic algorithms are investigated as learning tools. The genetic algorithm is found to be particularly effective. The experimental results presented demonstrate that the proposed neural controller can cope well with sudden changes in system dynamics, strongly nonlinear systems, and intrinsic time delays without prior knowledge of the system being controlled. It can also effectively interpret and use auxiliary information it is given about the system. The production of an inexpensive compact hardware implementation is discussed
[1] A. Sideris,et al. A multilayered neural network controller , 1988, IEEE Control Systems Magazine.
[2] Trevor J. Hall,et al. Learning by diffusion for multilayer perceptron , 1989 .
[3] Mitsuo Kawato,et al. Feedback-error-learning neural network for trajectory control of a robotic manipulator , 1988, Neural Networks.
[4] D. E. Rumelhart,et al. Learning internal representations by back-propagating errors , 1986 .