A Flexible Job Shop Pre/Re-schedule System Based On AGENT

Scheduling is a problem that how some tasks are arranged on the machines can achieve optimal objective function. The complexity of scheduling, uncertainty of scheduling environment and dynamicity of scheduling tasks make schedule more difficult. In order to address this situation, the scheduling system must be robust, scalable and adaptability. Flexible job shop scheduling problem reduce machine constraints, and increase complexity of problem, more close to actual work environment, have important theoretical and engineering significance.Pre/re-scheduling is a commonly used method of solving dynamic schedule problem, which combines the advantages of complete response schedule and robustness schedule and overcomes their shortcomings. It not only has the capacity of rapidly response to the dynamic external environment but also has a robust character. The thought of rescheduling the affected parts greatly reduces the conversion cost of reschedule in the implementation process, achieving the optimization of system and guaranteeing the stability of the system.The basic thought of multi-agent system is to decompose complex system and solve problem through negotiation mechanisms. It becomes an important method in the research of production scheduling system. Based on the study of multi-agent and multi-agent system, design flexible job shop scheduling system based on multi-agent, and map the physical entities of parts, steps and machines in the schedule system as agents, generate schedule program through negotiation among agents. In this sense, negotiation strategy plays a decisive role for schedule system performance. This paper extends the traditional contract net protocol, supports many to many negotiation mechanisms, and achieving the collaboration between agents.In this paper, we use discrete event simulation method, and achieve schedule system simulation in eclipse platform based java language. Multi-objective strategy is used in this paper, the mean tardiness as the main objective, makespan, the mean flow time and machine utility as a secondary objective, analysis schedule scheme in multi-angle. First, we produce schedule scheme through the negotiation strategy in pre-schedule, and compare it with the genetic algorithm by Jain to prove the effectiveness of multi-agent approach. Based on pre-schedule, we consider three types of interferences in dynamic schedule environment: machine failure, urgent part arrival and part cancellation, and respectively design interference solving strategy, simulating by pre-schedule objective and the departure of start time as objective and comparing with the traditional right-shift method to prove the superiority of this method. The multi-agent-based flexible job shop pre/re-schedule system in this paper better meet the requirements of dynamicity, integration, intelligence and practical in job shop schedule system.