Non-dominated sorting on performance indicators for evolutionary many-objective optimization

Much attention has been paid to evolutionary multi-objective optimization approaches efficiently solve real-world engineering problems with multiple conflicting objectives. However, the loss of selection pressure and non-uniformity in distribution Pareto optimal solutions objective space can impede both dominance-based decomposition-based optimizers when solving many-objective problems. In this work, we circumvent issue by exploiting two performance indicators, use these an optimizer’s environmental via non-dominated sorting. This effectively converts original problem into a bi-objective one. Our convergence criterion tries balance individuals different parts space. The angle between on is adopted measure diversity each individual. Using be separated layers easily, which often not possible for representation. proposed method evaluated DTLZ benchmark up 30 objectives, MaF test suite 10, 15, 20 experimental results show that our competitive compared six recently algorithms, especially large number