The Multi-Objective Problems In Mixed-model Assembly Lines Sequencing Based On Scatter Search Approach

With the globalization of economy, demands of customers tend to be more and more diversified and personalized. Manufacturing industry has been shifted from the traditional model of high-volume production to small and medium volume production of multi—species model. Mixed model assembly line is one of the effective ways to achieve mass-produce for multi-variety and small and medium size batch. Mixed model assembly line sequencing problem is one branch of the scheduling problems, and a key step to achieving mixed production. It can optimize the production sequence to ensure a balanced production, shorten the delivery time, reduce inventory and improve the competitiveness of enterprises.In order to solve practical problems, shorten the assembly time and improve production efficiency, a mathematical model is setup which aims at minimizing advance or delay loss of delivery the parts consumption rate, minimizing over-time and idle-time and total setup cost. Because such problems have been proved to be a combinatorial optimization problem in NP problem, therefore, in order to get a high quality and stable solution, Scatter Search algorithm is designed to solve the problem.In order to meet the application demands of the case company, the weight coefficient method is adopted to change multi-objective model into a single-objective model. An advanced Scatter Search approach is proposed and a new diversification generation method based on Genetic Algorithm is presented to generate a set of potentially diverse and high-quality initial solutions. Accordingly an update method of reference set, subset generation method, solution combination method and improvement method are designed.There is no effective way to determine the weight coefficient and the solution gotten by weight coefficient method doesn't always belong to the non-dominated solution set. However, Multi-objective evolutionary algorithm can provide a non-dominated solution set of options for decision makers. Therefore, a multi-objective scatter search algorithm based on Pareto is proposed. Based on the flexible framework of scatter search, some parts of the scatter search Algorithm are redesigned to get the non-dominated solution set according to the demands of multi-objective problems.The proposed two algorithms were verified in this thesis. By comparing with the HA, GA and other algorithms, ASS method not only maintains the diversity of the population, but also keeps a higher quality for the satisfactory solution. By comparing with multi-objective genetic algorithm results, the non-dominated solution set which was produced by Multi-objective scatter search algorithm can be a good approximation of the Pareto front, and also has good dispersion.Finally, a summary of the paper is given and the further research tends are outlined.