Intelligent Control Using Theta-Adaptive Neural Networks: A Physically Based Approach

A novel use of neural networks for parameter estimation in nonlinear identification and control problems is proposed. The neural network is used to identify the relation between system variables and parameters of a dynamic system. A block estimation and two different recursive estimation algorithms are presented. For the block estimation algorithm, a neural network is used to directly approximate the implicit function between the response and the parameters. On the other hand, the two recursive algorithms, where the neural networks are trained using two different performance indices, compute the parameter update rather than the estimate itself. Both the block and the recursive methods are useful for parameter estimation in systems where either the structure of the nonlinearities present are unknown or when the parameters occur nonlinearly. Analytical conditions under which successful estimation can be carried out are studied. How the algorithms can be applied to control of nonlinear systems with unknown parameters and the associated stability issues are also discussed.