Research On Hybrid Flow Shop Scheduling Using Ant Colony Algorithm Based On Petri Nets

The production scheduling is to carry out the plan of manufacturing schedule and mainly focus on how to allocate various resources properly in the workshop. The scheduling problem is a typical combinatorial optimization problem and belongs to NP-complete problems. Hybrid Flow Shop (HFS) manufacturing is the mode with mass production and varieties of products. Without changing the original production conditions, it allows the same production line or production unit to process the products of different types and different quantities, but the processing of products are similar. HFS has the high flexibility and is widely used in the real manufacturing enterprises. HFS system will involve scheduling problem when it needs to dispatch varieties of machines to process kinds of workpieces according to the specific artifacts. Therefore, to strengthen the research of hybrid flow shop production scheduling problem is of great significance for improving the production ability of Chinese enterprises.This thesis applies an automobile glass enterprise in Guangzhou as research background, elaborate the characteristics and research situation of HFS production. Then analyze several cases of HFS scheduling and point out it is a two-stage HFS scheduling problem with identical parallel machines and non equivalent parallel machines in this thesis. In order to improve the artificial scheduling of the laminated glass procuctionline, first according to the reality of workshop scheduling problem then append the buffer cache of limited capacity. Then propose a model which is suitable for HFS system and called Probability Petri Net (PPN). By adding the selection function in transitions and introducing tokens which carry the information of time and color, we can simulate the logic relationship of real time manufacturing system dynamically and build the petri net model. Next, the ant colony optimization algorithm is mainly researched, this thesis takes the strategy which the maximum and minimum values is limited to improve the convergence of the algorithm. Then introduce the algorithm into PPN model and select minimize makespan as the objective function. The scheduling scheme is constructed by scheduling the sequence of transitions in system, then use taguchi experiment to optimize the parameters of the algorithm. The program of PPN model and ant colony algorithm is written by using C#programming language in the environment of.Net Framework4.0. Finally, execute the simulations to verify the effectiveness of the algorithm.The ant colony optimization algorithm based on Petri net proposed by this thesis could obtains good effects on stability and convergence according to a large number of test results. The scheme reduces production cycle greatly and it verifies the rationality of this thesis. In the end, a summary of work in this thesis is made and the future work is also expected.