On-line optimization of unknown non-linear systems using a correlation-integration method

This paper addresses a novel on-line optimization control strategy for unkn own non linear systems. The Volterra series is used as a non-linear operato r for the representation of unknown non-linear systems, which maps one func tion to another with a specific form. Based on this non linear operator, th e correlation-integration method is proposed using on-line measured input a nd output data. This method has the following properties: system model free , strong rejection to disturbances, good robustness and on-line adaptation. Some examples are used to demonstrate these properties.