Local intensity in memetic algorithm
Zhiwei Zeng, Xiao-Min Hu, Min Li, Yu Luo · Proceedings of the Genetic and Evolutionary Computation Conference Companion · 2018
he research in capacitated arc routing problems (CARPs) is getting more and more attention for its wide applications in reality. Memetic algorithms (MAs) are promising in solving CARPs, but the intensity of local search (LS) in MAs should adapt to the evolution of the population for achieving the maximal effectiveness. In this paper, a novel MA based on adaptive adjustment of LS intensity (LIMA) is proposed. LIMA dynamically adjusts the probability of LS according to the distance between the individual of the population and the solution lower bound for further deepening the LS depth of potential solutions. It makes the algorithm spend as much time as possible on more potential solutions, so as to shorten the computation time and speed up the convergence. Experimental results show that LIMA has faster convergence speed and better global search ability than traditional MAs. Adaptively tuning the LS intensity of a MA has high potential for solving complex problems.1