Real-Time Output Feedback Neurolinearization

An adaptive input-output linearization method for general nonlinear systems isdeveloped without using states ofthe system. another key feature ofthis structure is the fact that, itdoes not need model ofthe system. in this scheme, neurolinearizer has few weights, so it is practicalin adaptive situations. online training ofneurolinearizer is compared to model predictive recurrenttraining. relationships between this controller and neural network based model reference adaptivecontroller are established. a cstr reactor 'and ph control in a neutralization process illustrateperformance of this method. simulation studies show a superior- performance with respect to a picontroller.