Noisy Immune Optimization for Chance-Constrained Programming Problems
This work puts forward a parameter-less and practical immune optimization mechanism in noisy environments to deal with single-objective chance-constrained programming problems without prior noisy information. In this practical mechanism, an adaptive sampling scheme and a new concept of reliability-dominance are established to evaluate individuals, while three immune operators borrowed from several simplified immune metaphors in the immune system and the idea of fitness inheritance are utilized to evolve the current population, in order to weaken noisy influence to the optimized quality. Under the mechanism, three kinds of algorithms are obtained through changing its mutation rule. Experimental results show that the mechanism can achieve satisfactory performances including the quality of optimization, noise compensation and performance efficiency.