Evolving the Tradeoffs between Pareto-Optimality and Robustness in Multi-Objective Evolutionary Algorithms

Many real-world applications involve the simultaneous optimization of several competing objectives and are susceptible to decision or environmental parameter variation which results in large or unacceptable performance variation. While several studies on robust optimization have been presented in the domain of singleobjective (SO) problems, the evolution of robust solutions is rarely studied in the context of evolutionary multi-objective optimization (EMOO). This chapter presents a robust multi-objective evolutionary algorithm for constrained multi-objective optimization. The proposed algorithm, incorporating the features of micro-GA which performs a local search for the worst case scenario of each candidate solution, the memory-based feature of tabu restriction to guide the evolutionary process and periodic re-evaluation of archived solutions to reduce uncertainty of evolved solutions, is capable of evolving the tradeoffs between Pareto optimality and robustness. The effectiveness of the algorithm is validated upon two benchmark with different properties and the I-beam design problem.