An Adaptive Design for Six Sigma(ADFSS): a Simulated Annealing and Regression Analysis Embedded Approach

Design for six sigma （DFSS） is a powerful approach of designing products, processes, and services with the objective of meeting the needs of customers in a cost-effective maimer. DFSS activities are classified into four major phases viz. identify, design, optimize, and validate （IDOV）. And an adaptive design for six sigma （ADFSS） incorporating the traits of artifidai intelligence and statistical techniques is presented. In the identify phase of the ADFSS, fuzzy relation measures between customer attributes （CAs） and engineering characteristics （ECs） as well as fuzzy correlation measures among ECs are determined with the aid of two fuzzy logic controllers （FLCs）. These two measures are then used to establish the cumulative impact factor for ECs. In the next phase （ i. e. design phase）, a transfer function is developed with the aid of robust multiple nonlinear regression analysis. Furthermore, 1this transfer function is optimized with the simulated annealing （ SA ） algorithm in the optimize phase. In the validate phase, t-test is conducted for the validation of the design resulted in earlier phase. Finally, a case study of a hypothetical writing instrument is simulated to test the efficacy of the proposed ADFSS.

Multi-objective Optimization Design of Inset-surface Permanent Magnet Machine Considering Deterministic and Robust Performances

The inset-surface permanent magnet(ISPM)machine can achieve the desired electromagnetic performance according to the traditional deterministic design.However,the reliability and quality of the machine may be affected by the essential manufacturing tolerances and unavoidable noise factors in mass production.To address this weakness,a comprehensive multi-objective optimization design method is proposed,in which robust optimization is performed after the deterministic design.The response surface method is first adopted to establish the optimization objective equation.Afterward,the sample points are obtained via Monte Carlo simulation considering the design-variable uncertainty.The Design for Six Sigma approach is adopted to ensure the robustness of the design model.Furthermore,the barebones multi-objective particle swarm optimization algorithm is used to obtain a compromise solution.A prototype is manufactured to evaluate the effectiveness of the proposed method.According to the finite-element analysis and experimental tests,the electromagnetic performance and reliability of the machine are significantly enhanced with the proposed method.