Agent And CA-based Research On The Shop Job Scheduling In Multi-Variety And Multi-Process Manufacturing Company

Shop scheduling problem in multi-variety and multi-process manufacturing company is a typical NP-hard problem. As modeling tools which can solve complicated problem, Cellular Automata and Multi-Agent System both have their own virtue and flaw, and also are complementary. So it is to some extent valuable and feasible to model, simulate, analyze and optimize the shop scheduling problem by combining the two methods.As the markets compete fiercely, the production processes of product have to be adjusted continually, according to the customer’s requirements, to personalize the customization, and therefore the scheduling part became especially important. Multi-variety and multi-process production system is a production method between single-variety of mass production and multi-variety of small-batch production, and it is a completely discrete and complicate system. As a part of the multi-variety and multi-process production system, shop scheduling problem in multi-variety and multi-process manufacturing company is as complicated as the ordinary shop scheduling problem, maybe even more complicated, however the existed solution can’t get a satisfactory answer.According to the characteristics of Cellular Automata and Multi-Agent System, a way of combining the two methods is put forward to apply in the shop scheduling problem in multi-variety and multi-process manufacturing company. The major work and results are as follow:1. Original characteristics of production logistics system in multi-variety and multi-process manufacturing company are studied. The shop scheduling module is analyzed, and some details are confirmed;2. A primary Cellular Automata model is set up to describe the shop scheduling system in multi-variety and multi-process manufacturing company. The shop scheduling system is simulated as a two-dimension network space, and the work station, scheduling rule and production information are used to describe the system;3. As the limit of CA and it is complementary with MAS, the CA model of shop scheduling problem in multi-variety and multi-process manufacturing company is extended. Import Agent and rule-decide unit, to set up an Agent-Cellular Automata whole model and its multi-goal function. Tuba search method is put forward to realize the scheduling process, solving a multi-goal and dynamic scheduling problem;4. Simulation frame and the simulation model of shop scheduling system in multi-variety and multi-process manufacturing company are set up, basing on the Swarm simulation platform. And the attribute and the method of each Agent are analyzed;5. The simulation research is done, taking a job shop in SN Company as a research object. And then according to the relationship among agents and their attribute and the ACA model evolution, the code is designed, and the simulation is operated. At last the simulation result is compared to that of the GA method.The result proved that the model and the simulation raised in this paper are to some extent feasible. However, due to the limit of time, people and material, there still is some flaw in the research. In the future, the discussion of multi-agent system theory could be further.