BIPOP: A New Algorithm with Explicit Exploration/Exploitation Control for Dynamic Optimization Problems

摘要: Dynamic optimization problems (DOPs) have proven to be a realistic model of dynamic environments where the fitness function, problem parameters, and/or constraints are subject changes. Evolutionary algorithms (EAs) getting pride place in solving DOPs due their ability match with Nature evolution processes. Several approaches been presented over years enhance performance EAs locate moving optima landscape and avoid premature convergence. We address this chapter new bi-population EA augmented by memory past solutions validate it knapsack (DKP). suggest, through use two populations, conduct search different directions space: first population takes charge exploring while second is responsible for exploiting. Once an environment change detected, knowledge acquired from old stored order recall whenever same state reappears. illustrate our study presenting several experiments compare results those standard algorithms.