Genetic algorithm based support layout optimization for machining fixtures considering process dynamics

The fixture-workpiece system endures elastic deformation throughout the metal cutting process. The deformation causes error to the machined features. Deriving the optimal support layout is critical to minimize the workpiece deformation. Moreover, the support layout also impacts chatter stability of the fixture-workpiece system. This research presents a support layout optimization scheme considering the process dynamics.; The overall development is modular and amenable to automatic formulation. The optimization program is built around a state-space model of the fixture-workpiece system, called the "Black Box". The state-space representation is derived from the truncated modal model of the workpiece. The Black Box evaluates two key criteria for a given support layout: peak dynamic response at machining regions and marginal stability of the system.; Considering multi-modal topography of the solution space, a genetic algorithm based optimization module is developed. An efficient encoding-decoding technique is developed to relax the constraints due to the discontinuous candidate regions for support placement. Experimental evaluation shows that the algorithm is able to find near optimal solution consistently.; A "Lookup" strategy is developed to prevent repeated evaluation of same support layout. Experiments show that the Lookup strategy reduces solution time by 38% to 90%.; A further development is made by introducing a procedural method to incorporate the fixture designer's knowledge at the beginning of the search process. The technique, called "Regional Constraint", allows the designer to impose placement constraints on one or more of the fixture elements. The experimental evaluation of this technique showed 75% improvement in solution time. The solution quality is also improved.; Taken together, this research proposes use of the genetic algorithm with the enhancements modules to perform the support layout optimization in the dynamic domain.