The Manufacturing Cell Formation Based On Queuing Network Model

How to supply customers with quality and quantity guarantee product in time is a problem for all enterprises, especially for some small-medium enterprises which adopt multiple products and small batch production mode. The problem is a key factor that restricts their development. Cellular manufacturing system(CMS) is a way to solve the problem. CMS incorporates the flexibility of job shops lines and high production rate of flow shop lines, so it has high flexibility and productiveness, that's the reason CMS could finish the task in short time. Cell formation(CF) is the first step to achieve CMS, and which is the most important step. Some key techniques are deeply discussed in this dissertation, like the method of determining the number of cells, the model for cell formation and the selection of optimum process route. The main work is described as follows:There is no certain approach to obtain the cell number of CMS. A way to obtain the cell number based on fuzzy cluster algorithm is proposed in this dissertation. Fistly, the application of fuzzy c-means algorithm(FCM) on the field of CF is investigated. The improvement of FCM did by predecessor is adopted. Then, the deficiencies of some validly functions are presented. A new validly function which considers some key factors is proposed. The valid function takes the essence of CMS into consideration. Lastly, the algorithm and valid function are tested by a couple of literature data.In order to solve CF problem, this dissertation resort to queuing network theory. The manufacturing system is equivalent to an open queuing network model. The machine in the network is regard as a service table of M/G/1 queuing system, which is more in line with the actual equipment diversification function. Supposing the volume of buffer in the queuing system is infinite, considering different parts have different arrival rates and process routes, different transport time, different equipments have a different service rates for different parts. Then a nonlinear multi objective optimization model is established based on minimized the parts' stay time in the queuing system, minimized the transport time for parts in the queuing network, minimized the cell load smoothness, and minimized the times of cross cell transport.A way to solve the model is obtained after incorporating the method of obtaining the cell number and the analysis results of objective functions. Then, apply genetic algorithm to optimizing the parts' stay time to select optimal processing route and apply NSGA-?algorithm to obtaining machine group. Lastly, assign part family into machine group suitably by minimizing the number of exceptional elements.Lastly, in order to verifying the application and effectiveness of the CF model, applying the model to a machining workshop of a power distribution equipment manufacturer in Lanzhou. The final results show the feasibility of the CF program.