| The function of the buffer is to store work in process and finished products temporarily,and at the same time,it is gradually becoming a part of each manufacturing company that cannot be completely eliminated because of its role in reducing impact and improving balance.However,with the increasing competition in the market and the improvement of machine intelligence making the internal and external structures of manufacturing enterprises complex and the random event disturbance diversified,and the combinatorial optimization problem in the optimal allocation of buffer capacity in the large scale manufacturing networks is becoming more and more significant,it is difficult to deal with the problem by conventional methods.Based on this,the following work is developed in this paper with the complex manufacturing network as the research object and with a certain total buffer capacity as the constraint and the maximum system productivity as the goal:1.Taking the hybrid production line as the model,the program and technical features related to the system aggregation method,system decomposition method and similarity conversion method in the buffer allocation problem are learned,analyzed and mastered to lay the foundation for the final solution of the buffer optimization allocation problem of the large scale manufacturing network.2.An improved performance analysis method for non-similar manufacturing network is proposed for the performance analysis problem of non-similar manufacturing network.The method is based on the idea of overlapping decomposition to introduce virtual machines and a manufacturing network containing typical manufacturing unit assembly lines and parallel lines is structurally decomposed into a set of linear production lines in a modular way.Meanwhile,to ensure the equivalence of the production system before and after decomposition,the parameters of the virtual machine and the performance parameters of each linear production line are solved by using the improved forward and backward iterative aggregation technique,which is different from the traditional aggregation technique,and finally the productivity of the main production line when the system is in steady state is taken as the productivity of the whole original system.The effectiveness and accuracy of the method are demonstrated by simulation experiments.3.Based on the contents of the previous part 1 and 2,a large-scale non-similar manufacturing network buffer optimization allocation technique is proposed.In order to reduce the difficulty of solving the scaled manufacturing system,this technique combines with the patent of the group,namely the recursive modeling technique,and the methodological ideas in part 2 above,divides the original system into several subsystems with different orders according to certain principles,and then establishes a recursive model,and then synchronizes solving the optimal solutions of each subsystem to finally determine the approximate optimal solution of the original manufacturing network system.Also,an improved adaptive genetic algorithm based on simulated annealing is proposed in subsystem search,which introduces a genetic operation based on simulated annealing and an adaptive mechanism to increase the diversity of populations to improve the global search capability;a statistic-based machine clustering algorithm is also proposed in building a recursive model of manufacturing network,and finally,it is validated using simulation experiments.4.To further enhance the usefulness of the above manufacturing network system performance analysis method and the buffer capacity optimization allocation technique of large-scale manufacturing networks and the efficiency and convenience in practical applications.Based on the previous paper,the graphical user interface(GUI)in Matlab was chosen to design and develop the corresponding software module in order to facilitate the user to input the relevant parameters and directly derive the assignment results.Finally,it was tested using Flexsim. |
PDF Full Text Request