Recognition of stability domain of nonlinear dynamical system using multilayer feed-forward artificial neural network approach

A method for obtaining the stability region of a nonlinear dynamical system using an artificial neural network is presented. This method takes advantage of many properties of adaptive networks. A design philosophy is presented. A Rumelhart feed-forward neural network with backward propagation of error algorithm is used to update the synaptic weights and thresholds until the desired stability region is recognized. An example of a nonlinear system is presented to illustrate the method.<<ETX>>