SMT生产线双机系统的贴装过程优化与系统实现

发布时间：2018-07-10 11:14

本文选题：贴片机路径优化 + 启发式算法　；参考：《北京交通大学》2017年硕士论文

【摘要】：随着智能化时代的到来,电子产品对贴装生产的要求越来越高,传统的贴片机优化策略已满足不了生产优化的需要,故有必要提出新的优化算法满足当前生产的需求。本文以某企业的SMT生产线优化为研究对象,针对目前拱架式结构贴片机双机系统的生产负荷分配不均、单机贴装行走路径较长与上料器排布欠优的问题,提出了针对生产线的双机分配整数规划模型,并构建了针对单机的直接启发式算法,采用模块集成方式结合上述算法实现对整条生产线的优化。首先,本文深入研究了拱架式结构的贴片机的结构特点和贴装过程,针对贴片机的单机优化问题展开了研究,分别采用遗传模拟退火算法、分区精确求解算法和直接启发式算法对贴片机单机优化求解。经过求解对比实验发现,分区精确求解结果要比遗传模拟退火算法的优化程度提高约19%,但分区精确求解有集成性差、求解时间长、对分区启发依赖性强的缺点;直接启发式较分区精确求解的优化程度相差约为5%,但直接启发式有求解时间短、算法集成性好的优点;同样,在不同规模下直接启发式比遗传模拟退火算法有求解结果更优、求解效率高、实用性更强的特点。通过算法间的比较,在贴片机单机优化方面选取直接启发式算法。其次,在对SMT生产线的双机任务分配的研究中,建立了针对双机的任务分配模型;由于数学模型含非线性因素,在原模型的基础上实现了模型线性化,且给出了线性化无损证明;给出了双机分配的求解测试与分析,表明了该方法的实用性。最终,本文实现了对优化结果的可视化,并对可视化模块与算法模块进行封装打包,且开发了软件的人机交互界面。其中,软件集成实现了输入文件的标准化,调用了 Cplex求解双机任务分配的整数规划模型,采用了直接启发式算法对已分配的单机优化问题求解,并实现了各机求解结果的可视化。软件试运行采取不同规模的PCB数据测试,经过对软件运行结果的分析,表明本文方法能在双机系统中实现较为均衡的任务分配,并且对任务分配后的单机贴装路径能实现很好的优化。
[Abstract]:With the advent of the age of intelligence, the demand for the production of electronic products is becoming more and more high. The traditional patch machine optimization strategy can not meet the needs of production optimization. Therefore, it is necessary to put forward a new optimization algorithm to meet the needs of the current production. This paper takes the SMT production line of an enterprise as the research object, aiming at the present arch type structure patch. The production load distribution of the machine and double machine system is not equal, the single machine mounting path is long and the feeder is not excellent. A two machine allocation integer programming model for the production line is put forward, and a direct heuristic algorithm for single machine is constructed, and the optimization of the whole production line is realized by combining the module integration method with the above algorithm. First, the optimization of the whole production line is realized. In this paper, the structure characteristics and the mounting process of the patch machine of the arch type structure are deeply studied. In view of the single machine optimization problem of the patch machine, the genetic simulated annealing algorithm is used, the partition precision solving algorithm and the direct heuristic algorithm are used to solve the single machine in the single machine. The optimization degree of the genetic simulated annealing algorithm is about 19% higher than that of the genetic simulated annealing algorithm, but the exact solution has the disadvantage of poor integration, long time and strong dependence on the partition heuristic. The difference between the direct heuristic and the optimal solution is about 5%, but the direct heuristic has the advantages of short time and good integration of the algorithm. The direct heuristic is better than the genetic simulated annealing algorithm, which is more efficient and more practical. Through the comparison of the algorithms, the direct heuristic algorithm is selected for the single machine optimization of the patch machine. Secondly, in the study of the dual task allocation of the SMT production line, a task allocation model for the double machine is established. Because of the nonlinear factors in the mathematical model, the model linearization is realized on the basis of the original model, and the linearized nondestructive proof is given. The solution test and analysis of the double machine allocation is given, which shows the practicability of the method. Finally, this paper realizes the visualization of the optimization results and encapsulates the visual module and the algorithm module. The software integration has developed the human-computer interaction interface of the software. Among them, the software integration realizes the standardization of the input files, calls the integer programming model of Cplex to solve the dual task assignment, and uses the direct heuristic algorithm to solve the distributed single machine optimization problem and realizes the visualization of the results of each machine. The scale of PCB data test, after the analysis of the results of the software operation, shows that this method can achieve a more balanced task allocation in the dual machine system, and the single machine placement path after the task allocation can be well optimized.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN405

【参考文献】