Optimization Strategy For Production Scheduling With Learning/Deteriorating Effects

As the pillar of our national economy, the manufacturing industry has been faced many problems in the fierce market, such as, the downturn in exports caused by the weakly recovery of global economic, the increasingly lack of domestic natural resources induced by the rise of raw material prices, diverse requirements of the customers, extensive development model and "three high and one low" (high input, high consumption, high pollution, low efficiency) causing the waste of resources and so on. How to coordinate the production time, production resources, production environment, and other factors to improve the profit margin, shorten the production cycle, improve the credibility of the enterprise in the case of limited resources, is an important condition by which China’s manufacturing industry can get an invincible position in the domestic and international market competition. Production scheduling can effectively solve the optimal allocation of resources. It is an important way for enterprises to improve productivity, save costs and improve economic efficiency and competitiveness under the existing conditions of production and operation. In this context, we establish the scheduling model closing to the actual world and design the optimization algorithm according to the problems encountered in the enterprise and the new market, which has a great impact on the research of theory and the practical application.Firstly, we introduce the background and significance of the research on production scheduling, and summarize the main research directions and the research status in the field of modern production scheduling.Secondly, based on the practical production background, the paper studies the problem of production scheduling with the phenomenon of learning and deterioraion. The paper puts forward two scheduling models, which are based on the learning effect and deteriorating jobs, namely exponential position dependent learning/deterioration model, and DeJong’s learning and deterioration model. The first model considers the effect of learning and deterioration, respectively. The model only considers the effect of learning is suitable for the production environment which is simple, Otherwise, the model only considers the effect of deterioration is applied to the production environment in which the process is more complex. The second model considers DeJong’s learning effect and deteriorating jobs simultaneously, which is suitable for most production environment, such as sustainable production can cause upgrading, however halting production may lead to recession. The models we proposed above can overcome the shortcomings of the existing research, and can be closer to the real production environment.Thirdly, we study the single machine and parallel machine scheduling problems with the models proposed above, the objectives are to minimize the makespan and the total completion time. We fist analyze the complexity of each problem. Then for the polynomial-time solvable problems, we give corresponding polynomial-time optimal algorithms for them, and for the NP-hard problems, we give fully polynomial time approximation schemes (FPTAS) for them.Finally, by analyzing the experiment examples of the FPTAS with MATLAB, we verify the feasibility and effectiveness of the approximation algorithm from the practical point of view.