An application of a neural network to adaptive control of a servo system

A neural network was applied to adaptive control of a servo system. The neural network was used to learn inverse dynamics of a controlled system. Experimental works were done using an electro-hydraulic servo motor with a digital signal processor. Two types of neural networks were tested. Simulation studies were done to investigate effects of ordering of training data, choice of input signals fed to a neural network, and number of neurons in a hidden layer.