Diffuse Response Surface Model Based on Advancing Latin Hypercube Patterns for Reliability-Based Design Optimization

Abstract

Since variances in the input parameters of engineering systems cause subsequent variations in the product performance, Reliability-Based Design Optimization (RBDO) is getting a lot of attention recently. However, RBDO is computationally expensive. Therefore, the Response Surface Methodology (RSM) is often used to improve the computational efficiency in the solution of problems in RBDO. In this chapter, the Diffuse Approximation (DA), a variant of the well-known Moving Least Squares (MLS) approximation based on a progressive sampling pattern is used within a variant of the First Order Reliability Method (FORM). The proposed method simultaneously uses points in the standard normal space (U-space) as well as the physical space (X-space). At last, we investigate the optimization of the process parameters for Numerical Control (NC) milling of ultrahigh strength steel. The objective functions are tool life and material removal rate. The results show that the method proposed can decrease the number of `exact' function calculations needed and reduce the computation time. It is also helpful to adopt this new method for other engineering applications.