Simultaneous prediction interval-based multiobjective solution approach for multiple quality characteristics optimization

Simultaneous optimization of multiple quality characteristics is a critical and difficult task for researchers and practitioners. This is primarily due to presence of correlation between responses and therefore tradeoff between them is inevitable. Thus, there is no single global optimal solution for such problems. Such problems are also known as multiple response optimization (MRO) problems. Among the different solution approaches proposed for MRO problems, Pareto front solution is one such alternative. However, there is no evidence of systematic work that addresses response uncertainty considering simultaneous prediction interval. This paper illustrates a systematic approach to generate Pareto solutions using multiobjective optimization (MOO) techniques for MRO problems considering appropriate simultaneous prediction intervals. The proposed solution approach is verified using two MRO case instances. Pareto solutions for MRO problems are generated using two MOO strategy [Non-dominated Sorting Genetic Algorithm (NSGA-II) and Multiobjective Particle Swarm Optimization (MOPSO)]. A comparative study shows that NSGA-II provides better Pareto fronts than MOPSO. The MRO solution quality of NSGA-II is also found to be encouraging for future research.