This research extends the sub-population genetic algorithm and combines it with a global archive and an adaptive strategy to solve the multi-objective parallel scheduling problems. In this approach, the global archive is applied within each subpopulation and once a better Pareto solution is identified, other subpopulations are able to employ this Pareto solution to further guide the searching direction. In addition, the crossover and mutation rates are continuously adapted according to the performance of the current generation. As a result, the convergence and diversity of the evolutionary processes can be maintained in a very efficient manner. Intensive experimental results indicate that the sub-population genetic algorithm combing the global archive and the adaptive strategy outperforms NSGA II and SPEA II approaches.
