Research On Dynamic Job Shop Scheduling For Die & Mould Corporations

Moulds/dies are important technical equipment for forming industrial products. Mould & die industry is one of the generally recognized key industries. In recent decades, mould & die industry has been developing rapidly as the national economy develops apace. However, the level of our national mould & die industry is still inferior to those in developed industrial country especially with regard to enterprise management. By investigating and researching several mould & die manufacturers and analyzing correlative data, it has been found that production management is the management bottleneck. For the production is driven by orders and its outstanding features are one-of-a-kind and experience-dependent, production in the shop floor is poor regular and changeful. Planning management is found to be the weakest link. Unfortunately, planning management is very intractable for mould & die manufacturers. Considering the condition, this thesis takes the research subject as dynamic job shop scheduling in mould & die manufacturers. Theoretical researches for some key technologies have been conducted systematacially. Based on results of the theoretical researches, a dynamic job shop scheduling system for mould & die manufacturers, which can be integrated with shop monitoring system, has been developed in this dissertation. Starting with modeling dynamic job shop scheduling, this thesis systematacially analyzed all the job-shop-scheduling-related properties of tasks, resources and machining by adapting object-oriented ideas. In order to explore effective algorithms to schedule job shop in real production system, a kind of relatively simple and special job shop scheduling problem, parallel machine scheduling problem, is put forward. A modified genetic algorithm(mGA)is given to solve the problem. In the mGA, two new concepts, Machine Hour Capacity Index and Redundant Machine Gene Pool, are presented; Monte Carlo selection is add to choice two chromosomes from father chromosomes and children chromosomes in crossover operator and a machine load-based heuristics(LBH) mutation operator is designed. Comparing to other existing algorithms, the mGA perform well with regard to the presented parallel machine scheduling problem. However, for the running time, the mGA seems not adequate to schedule job shop for real production system in mould & die manufacturers. For this reason, a dynamic job shop scheduling problem with operation relativities is put forward. Based on defining and describing operation relativities systematacially, optimal model and disjunctive graph model are established for the problem. Furthermore, a dispatching rule-based heuristic approach is advanced. In the heuristic approach, a new combinatorial dispatching rule, RAN(FCFS,ODD), is designed to compare with seven dispatching rules presented by previous literatures. A simulation research is conducted to validate the approach and the RAN(FCFS,ODD) dispatching rule. The results indicates that the approach is effective and RAN(FCFS,ODD) dispatching rule outperforms all the other seven dispatching rules with regard to the dynamic job shop scheduling problem with operation relativities. In order to put the heuristic algorithm in practice, this dissertation has researched some related key technologies systematacially in view of scheduling job shop for real production system in mould & die manufacturers. Considering practicability and time complexity of the advanced algorithm, a concept of time parameters conversion is presented and the correlative conversion algorithm is developed. With the aim of utilizing machining resources adequately and rationally, a fuzzy decision method is advanced to optimize machining resources for operations. By introducing into concepts such as "combinatorial part","resource group","resource using unit"and "processing time span", some unconventional scheduling problems, two kind of scheduling problems, grouped processing scheduling problems and multiprocessor scheduling problems, are simplified into conventional scheduling problems. In order to integrate dynamic job shop scheduling system and shop monitoring system effectively, correlative integration technologies, which include method of determining time span of daily-schedule and processing method of monitor data, are researched. Considering characteristics of dynamic job shop schedulingin mould & die manufacturers, a backward search algorithm is developed to get the critical paths of moulds/dies. Based on all the above-mentioned theories, methods and results, a dynamic job shop scheduling system for mould & die manufacturers is developed according to the ideas of object-oriented programming. The system is designed in C/S mode and programmed in Java. With the help of the system, in which the production features in mould & die manufacturers and effective integration with shop monitoring system are considered, users can get an original job shop schedule rapidly. Then, by adequately interacting with the system, they can get a feasible and satisfying job shop schedule. Implementation This system has been put into use in two mould corporations. The results indicate that the system is significant in strengthening orderliness of production in shop floor, enhancing utilance of resources, shortening manufacturing span of moulds/dies, predicting potential deferred moulds and helping to make appropriate decisions.