A Multi-objective Cross-training Plan Based on NSGA-II

Abstract

The previous research on cross-training mainly focused on productive efficiency. However, enhancing labor’s satisfaction of tasks is as much important as improving production performance. This paper addresses a new cross-training policy for an assembly cell from the point of view of humanization. A multi-objective 0-1 integer programming model is presented to implement the cross-training policy for an assembly cell. The first objective works on getting to maximize average satisfaction degree, and the second objective seeks to minimize average paid salary, while determining which labors should be cross-trained on which tasks. Non-dominated sorting genetic algorithm (NSGA-II) is developed to solve the model. A series of computational experiments are proceeded to explore the impact of three factors on cross-training, including labor’s preference structure, labor’s salary structure, and task redundancy. The results indicate that the balanced preference structure is better than the extreme one, the non-uniform salary structure is better than the uniform one, and the smaller task redundancy is better than the bigger one under various scenarios in the paper. Those insights can be used to direct the managers of human resource to choose the candidates for cross-training