A neural network-based technique for structural identification ofSISO systems

摘要

This paper presents a simple technique for the structuralidentification of single-input, single-output (SISO) dynamic systems,based on the use of a neural network. The network is trained torecognize some significant features of the process dynamics startingfrom a simplified representation of its unit step response, which inturn is obtained by a convenient I/O experiment. In addition, thenetwork classifies the process with respect to a convenient set ofpossible model structures, which represent the most common situationsarising when a process model needs to be identified for control purposes