The Model And Algorithms Of Supply Chain Scheduling Under JIT And MTO Environment

Based on today’s market environment faced by individualized, diversified and rapid requirements, the mode of make to order, timely delivery and supply chain management has become an important way to respond fast changing market, enhance customer satisfaction and reduce cost for enterprise or enterprise group. Therefore, to study several suppliers who have multi-centers, multi-depots and plenty of transport vehicles for severing major clients, under the objective of improving customer satisfaction and decreasing the overall distribution cost, from the viewpoint of the level of supply chain operations and scheduling, how to arrange customer orders and all vehicles reasonable and effective is of vital significance for supplier enterprise or customers.Combined with the characteristics of make to order and timely delivery, the models of job scheduling and vehicle scheduling is proposed, which has established multi-supplies, multi-centers, many customers, large orders confined with strict time window. The factors of this model are considered comprehensively. For customers, multiple clients, each customer order with strict time window and greater orders is taken into account, and consequently the suppliers’distribution centers must dispatch timely more than one customer. Analogously, for suppliers, multi-centers, multi-depots, and several transport vehicles and vehicles with backhauls is also pondered over. In this model, the objective function clearly gives the maximum customer satisfaction and minimum the overall distribution cost, and corresponding the constraints include assign customer orders, calculate the job time for each customer, and require the customer with time windows and vehicle scheduling and so on. Because of many factors to consider, moreover, the constraints are harsh, it should be noted that reduce factors and relax constraints is easy to transform the problem into a model with another application background. The model has very strong versatility.In Algorithm design, this consider problem is NP problems, there is no polynomial time exact algorithm, so using hybrid algorithm based on genetic algorithm to solve the mode. Because the model is concluding many factor, after many times attempts, failure to design a solution and the corresponding coding form which can satisfy all the details of the problem and make it feasible, finally, using hybrid algorithm to solve the model. Facing customer order scheduling problems, using a variety of genetic algorithms and multi-step adjustment algorithm, direct to the ordinary genetic’s shortcomings of poor diversity and infeasibility arise by problem complexity, design a improved genetic algorithm and multi_population genetic based on simulated annealing Algorithm with3-step adjustment; at the same time, using rational meaning assigned and two-step multi-strategy adjustment algorithm for the vehicle scheduling problem.The distribution of ready-mixed concrete is typical problem which has strict time window constraints, its distribution level not only relation to the cost of distribution enterprises, but also directly affect the cost of the construction project, project progress and project quality, etc., Due to the model environment is similar to the distribution of ready-mixed concrete, so in Chapter5, take concrete scheduling and distribution as an example, Run the programming in the MATLAB environment, iterated350generation, total running about3.5hours, scheduling total64vehicles, including its own vehicles49vehicles and hire vehicles15, complete674jobs for71clients in the time windows of04:00-15:00, including purchased241, distribute ready-mixed concrete5323m3, which all meet customers’ time windows requirements and continuous supplying. By tracking each vehicle and each distribution center, the graphical display of each distribution center and each vehicle is extremely busy. By contrasting, the three kinds of genetic algorithm has good convergence, but the multi_population genetic simulated annealing algorithm is much better than the improvement genetic algorithms and ordinary genetic algorithm on the convergence rate; about the quality of eventually optimal solution, The improvement genetic algorithm has a slight advantage genetic compare to simulated annealing algorithm, and the normal genetic algorithm is worst. In summary, the results show that the model has good universality, and the algorithms is the feasible and effective.