An integrated simulated annealing-based method for robust multiresponse process optimisation

作者：Sibalija Tatjana V^*; Majstorovic Vidosav D

来源：International Journal of Advanced Manufacturing Technology, 2012, 59(9-12): 1227-1244.

DOI：10.1007/s00170-011-3572-3

• Simulated annealing
• Multiple responses
• Optimisation
• Robust parameter design

This paper presents a generic optimisation methodology for the selection of optimal process parameters in multiresponse processes that consists of two stages: The first stage is based on a quality loss function and multivariate statistical methods in order to adequately present responses, uncorrelate and synthesise them into a single process performance measure; the second stage uses artificial neural networks to build a process model and a simulated annealing (SA) algorithm to find the optimal process parameter conditions. The initial point of the SA algorithm is generated in such a way as to improve the convergence to the actual global optimum. The most important SA algorithm parameters are varied to assess the algorithm's robustness in terms of the effect of the algorithm setting on the quality of the final solution (optimal process parameters and the corresponding process performance measure). The results are benchmarked to genetic algorithm (GA) performance within the proposed methodology, and the quality of a final solution, the robustness of the optimisation algorithm, the speed of a convergence to the optimum and the computational time are evaluated. Four case studies are presented to illustrate the effectiveness of the proposed methodology in comparison to several commonly used approaches from the literature and also to the GA-based performance.

• 出版日期2012-4