An improved-MPGA and its application in OTSG load reduction characteristics optimization

Abstract

Multi-population genetic algorithm (MPGA) perfectly inherits the advantages of standard genetic algorithm (SGA), and improves the global search performance of SGA by introducing multiple populations. However, its optimization results depend on the initial solution and the search ability is weakened in the later calculation process. To overcome these disadvantages, an initial population generation method based on the idea of cluster analysis was improved to improve the diversity of the population. To balance global exploration ability and local search ability, different crossover operation values and different mutation operation values were assigned to different populations. The algorithm is tested by using classical functions of different dimensions and the results show that the improved algorithm has better robustness. What’s more, the Improved-MPGA is introduced to optimize parameters set in the reactor design control scheme. Parameters that can meet the requirements of reactor power overshoot and dimensionless evaluation index are obtained. And the results show that the load reduction characteristics of the Once-through Steam Generator (OTSG) under the optimal scheme are well improved compared with the design scheme.