Automated Identification of a Mechatronic System Model Using Genetic Programming and Bond Graphs

Abstract

Modeling, identification (experimental modeling), and design of dynamic systems and the associated problem of controller design are common problems in the field of mechatronics. A typical mechatronic problem entails finding the best topology as well as parameter values of the desired solution. In view of dynamic interactions in a mechatronic system, which involves more than one domain, it is desirable to use concurrent and integrated methodology in the solution. The powerful search ability of genetic programming (GP) along with the domain independence and the open architecture of bond graph (BG) modeling can be integrated to develop an evolutionary mechatronic tool for identification of a complex mechatronic system. This paper extends this integrated approach to nonlinear mechatronic problems and develops a software tool for this purpose. It is illustrated how the developed technique and the corresponding software tool can be used in the automated synthesis and identification of a nonlinear mechatronic system. The performance of the software tool is validated by applying it to a nonlinear electrohydraulic manipulator, which falls into the class of multidomain systems. The results obtained from the application are quite encouraging, and form a rationale for the extension of the tool for concurrent and optimal design of mechatronic systems.