多負載式無人搬運車系統之控制策略

Abstract

本研究主要探討諸如半導體無塵室等要求高精密度、低污染性，且具複雜製程的無人自動化混製系統中物料搬運的問題。在此類系統中，無人搬運車可用來遞送加工站相互間的工件，每一加工站前設有輸出、輸入暫存設備以供車輛載卸工件。由於此類生產設備之購置及操作成本相當高，輸出、輸入暫存設備之容量因而受到嚴格限制。為使環境保持潔淨且有足夠運輸量，以少輛數多負載式車輛取代多輛數單負載式的無人搬運車不失為一有效解決方案。如何控制系統中同時加工、運搬中的多個工件為此類生產系統特有的重要課題。本研究建立一套車輛及工件的指派控制法則使得系統免於發生鎖死的現象，亦即車輛每次卸貨都不須等待，而且甫完成加工的工件均可即刻移至輸出暫存區。在整個系統運作中，許多因子間的交互影響使得所形成之無鎖死環境的績效表現各有不同。這些因子諸如：各加工站輸出、輸入暫存容量的大小；車輛的負載量；使用的車輛數；系統的工作量；有否使用集中暫存區；系統中的總工件數量；設施佈置圖。本研究使用電腦模擬軟體AutoMod 來模擬在若干個上述因子不同的設定組合下，系統的運作狀況。各組合之模擬結果將依若干指標加以收集整理並比較，諸如：系統的完工率；一工件於車輛上運行一程的時間；加工機使用率；加工機被回堵的頻率；每次加工機被回堵的平均時間；平均系統中在製工件數；工件於輸出暫存區之平均等候數。 本論文之內容分成三個子題。一、針對多負載式無人搬運車系統，在一特定的工作量下，考慮車輛之空車運行時間、車隊之預期利用率以及總載貨時間，提出一解析模式，求得所需之最少的車輛數目。二、針對一個僅使用單部多負載式無人搬運車的作業環境，發展一套即時且具免鎖死特性的車輛控制策略。藉由取得現場中車輛、工件暫存區、加工機器之即時作業現況，安排車上之載運位置與各輸出入暫存空間之使用順序，有效防止車輛在搬運過程中形成鎖死現象。三、針對多輛多負載式無人搬運車之製造系統，加入對不同車輛之間的任務協調考量，推衍出另一套適用的多車控制策略。 經過電腦模擬實驗的結果分析顯示，本論文所提出的多負載式無人搬運車控制策略可針對系統的即時動態做出快速回應，其簡易性更提供了相當大的應用彈性。有效地為無人自動化工廠提供了一個免鎖死的製造環境。

This research concerns several issues of the material handling system in an unmanned automated job shop with the stringent requirement on precision, ultra-clean, and complicated processes, such as the operations of semiconductor wafer fabrication in a clean room. Automated guided vehicles are employed to deliver jobs between processing centers. An input queue and output queue devices are installed at each processing center for interfacing with vehicles. Smaller queue size is preferred since the specific manufacturing system is very expensive in construction and operation. To maintain the environment in cleanliness and to satisfy transport requirement, instead of many vehicles with single loading position, fewer vehicles with multiple loading positions are employed. How to control the material flow of concurrent multiple jobs in the system is a critical issue for this type of manufacturing system. This research developed a set of control rules for the jobs and vehicles in order to enable the manufacturing system with deadlock-free in real-time operation. The system is deadlock-free if vehicles never wait for unloading jobs as well as each job could be moved from the processing machine to the output queue right after the processing is completed. Several factors interface each other and result in different system performance while achieving a deadlock-free environment. Factors under consideration are: capacities of input and output queues at processing centers, vehicle loading capacity, number of vehicles employed, throughput requirement of the system, a central buffer device is existed or not, the number of jobs allowed concurrently in the system, and facilities layout of the system. Simulation software package, AutoMod is used as the tool to simulate each combinations of above factors. Several performance criteria are collected at each simulation run with different factor combinations; for instance, shop throughput, riding time of a job loaded on vehicle per trip, machine utilization, frequency of machine blockage occurring, average duration of machine blockage events, average work-in-process in the system, and average number of jobs on the output queues. This thesis composes three related subjects. The first subject is the analytical method for estimating the fleet size of a multi-load AGV system. The second subject is the development of a set of real-time deadlock-free control rules for a job shop system with a single multi-load vehicle. All the status of vehicle, queues, and processing centers are considered. The last subject deals with a set of control rules for the job shop system with multiple multi-load vehicles. As the simulation experiment shown that the control rules effectively respond to the system dynamic and provide a significant flexibility in application. And, a deadlock-free workshop is also verified successfully without using any other extra storage or limiting the number of concurrent jobs among the shop.