An Efficient Feasibility Robust Optimization Method Using a Sensitivity Region Concept
We present a new robust optimization method that ensures feasibility of an optimized design when there are uncontrollable variations in design parameters. This method is developed based on the notion of a sensitivity region, which is a measure of how far a feasible design is from the boundary of a feasible domain in the parameter variation space. As the design moves further inside the feasible domain, and thus becoming more feasibly robust, the sensitivity region becomes larger. Our method is not sampling-based so it does not require a presumed probability distribution as input and is efficient in terms of function evaluations. In addition, our method does not use gradient approximation and thus is applicable to problems having non-differentiable constraint functions and large parameter variations. As a demonstration, we applied our method to an engineering example, the design of a control valve actuator linkage. In this example, we show that the method is efficient and the optimum design obtained is robust.