Mixed Line Optimization Based On The Dual-population Parallel Genetic Algorithm

E-commerce has an important role in the contemporary service industry. And the subject that how to manage and utilize the process in the FC (short for Fulfillment Center) has drawn a lot attention from the E-commerce enterprises. In the FC, the response speed and quality of the Reverse logistics department’s is a critical factor not only affects customer experience but also the operation costs. The research background of this article is based on a Kaizen Event of the C-R (short for Customer-Return) Department from a E-commerce enterprise. The goal of the Kaizen Event is to utilize and balance the whole process line within the department. On the contemporary, the C-R department has a difficulty to raise up the total UPH duel to the fact of unevenness of the whole process line. We are to eliminate the’Bottle-neck’station by re-layout as well as re-plan the work elements within different stations, ultimately, achieve the goal of roll up the UPH along with balancing the process.The optimize work of this article is based on the outcome of the Kaizen project and set up the multi-objective model on mixed assembly line balancing principle. What’s more, this paper uses the theory of dual-population genetic algorithm and collaborative parallel algorithm to solve the problem. This paper has significant features and the characteristics in the following four perspectives:In the research background: the C-R department processes numerous kinds of goods and what’s more, the object of this paper can make an appointment schedule just according to its requirement which means the optimization has a more comprehensive consideration. In the analysis method: This paper combines the qualitative with the quantitative optimization ways to achieve the perfect layout solution which could reduce the "Bottle-neck" station. Moreover, gets the best work elements assignment and product sequencing plan based on the mixed line balancing model. In the model set:By goal separation and the independent population evolution, this paper solved the multi-objective function in a way which can avoid the inappropriate target weights set. In the algorithm and programming:In order to seek for the global optimal solution as well as to higher the searching speed, this paper uses the dual-population genetic algorithm optimization principle and the coarse-grained parallel optimization theory. First of all, evaluates the optimal performance of the multi-objectives exchange theory with the other four. Secondly, seeks for influence of parameters like population scale and generations etc. At last, this paper validates the optimal solution from two different dimensions further to guarantee the performance of the optimal solution.