A framework of design weakness detection through machine health monitoring for the evolutionary design optimization of multi-domain systems

Design of a multi-domain engineering system can be complicated due to its complex structure and dynamic coupling between domains. Ideally, designing a multi-domain system should be done in an integrated and concurrent manner, where dynamic interactions between domains in the entire system have to be considered simultaneously, throughout the design process. In recent years, researchers have made some progress in the integrated and optimal design of multi-domain systems. Dynamic modeling tools such as Bond Graphs and Linear Graphs have been considered for modeling multi-domain systems, which can facilitate the design process. In the process of design optimization, a rather challenging task is to concurrently satisfy multiple design objectives. Methods of evolutionary computing, genetic programming in particular, have received much attention in recent years for application in design optimization. These methods can be extended to evolutionary optimization, which may involve complex and non-analytic objective functions and a variety of design specifications. More recently, machine health monitoring system (MHMS) has been considered for integration into the scheme of design evolution even though no concrete developments have made in this regard. In this paper, a framework of design weakness detection through machine health monitoring for evolutionary design optimization of multi-domain system is proposed. MHMS is integrated with evolutionary design optimization to make the overall process of design evolution more effective and feasible from the practical point of view. Information form MHMS is utilized to detect the “sites” or “candidates” of design weakness, which will involve computation of a new measure that can reflect the quality of the current design. These candidates of design weakness are then provided to the process of evolutionary design optimization. On subsequent analysis, design improvements would be made only if these candida- es were found to be related to design weaknesses. Otherwise, the monitoring process will continue. Supervised design weakness detection is achieved through the integrated system of MHMS and evolutionary design optimization. In addition, a Design Expert System is employed to monitor and assist both design weakness detection and isolation, and feasible design selection.