Dynamic Constrained Time-Domain Robust Optimization with Noises

In the field of engineering optimization design, there are many optimization problems with time-varying objective functions. Tracking motion optimization (TMO) and time domain Robust optimization (ROOT) are two widely used methods to solve this problem. The purpose of TMO is to find the optimal solution before the environment changes, while the purpose of ROOT is to find a solution that can adapt to changes over a long period of time. However, both TMO and ROOT approaches have some defects. First of all, the existing engineering optimization problems have constraints and will change with time. Second, in the optimizing environment, the noises in most environments will affect the candidates. Therefore, a dynamic constrained time-domain robust optimization algorithm (DCRO/SCNO) considering noises and switching costs is proposed in this paper. Switching cost and robustness are considered as multi-objective optimization, and the idea of double elite is introduced to divide the population into three groups to collaborate for the problem of non-robustness solutions in noisy environment. Benchmark experiments show that DCRO/SCNO can find non-dominant solutions in different noise environments, which confirms the validity of the dual elite idea.