基于模糊Petri网推理的半导体生产线动态调度研究

针对动态调度的特点,为了综合优化半导体生产线性能指标,尝试使用基于模糊Petri网推理的方法进行动态调度。首先分析了影响动态调度决策的生产线状态信息;然后建立了模糊Petri网形式化推理机,继而构建了面向半导体生产线的模糊Petri网推理模型;最后使用实际半导体生产线模型,将提出的方法与FIFO、EDD和CR策略进行了仿真比较。结果表明基于模糊Petri网推理的动态调度方法能够改善半导体生产线多种性能指标。     

Comparing the effect of different strategies of continuous improvement programmes on repair time to reduce lead time

To cope with uncertainty in semiconductor wafer fabrication facilities (fabs), scheduling methods are required to produce quick real-time responses. They should be well tuned to track the changes of a production environment to obtain good operational performance. This paper presents an efficient adaptive dispatching method (ADM) with parameters determined dynamically by real-time state information of fabs. ADM is composed of a dispatching rule considering both batch and non-batch processing machines to obtain improved fab-wide performance, several feature selection methods to determine key scheduling-related real-time state information, and a linear regression model to find the relations between the weighting parameters of the dispatching rule and the determined real-time state information. A real fab simulation model is used to demonstrate the proposed method. The simulation results show that ADM is adaptive to changing environment with better performance than a number of commonly used rules (such as FIFO, EDD, CR, LPT, LS, SRPT, and SPT) and an adaptive dispatching rule that considers only real-time ratio of hot jobs to the number of all jobs in a fab and the ratio of jobs with one third of photo steps left to the number of all jobs.     

A nonlinear scheduling rule incorporating fuzzy-neural remaining cycle time estimator for scheduling a semiconductor manufacturing factory—a simulation study

Scheduling of a semiconductor wafer fabrication system (SWFS) is complicated due to its re-entrant product flow, high uncertainties in operations, and rapidly changing products and technologies; thus dispatching rules have been widely used for real-time scheduling because they can provide a very quick and pretty good solution. However, deciding how to select appropriate rules is very difficult and seldom tackled. This paper describes an approach into the evaluation and optimization of dispatching rules by integrating the simulation and response surface methodology (RSM). In order to implement the proposed approach, a dynamic bottleneck dispatching (DBD) policy is designed, in which bottlenecks are detected in a timely way and adaptive dispatching decisions are made according to the real-time conditions. In addition, two case studies are carried out to demonstrate the approach. One case compares DBD to regular rules, such as CR + FIFO, EDD, SRPT, SPT, SPNB and Justice, a bottleneck dispatching method. Simulation results show that the DBD policy is superior to the other six methods. In another case study, the parameters of DBD are optimized by RSM and desirability function, and the result proves that the optimized DBD method can get even better performance.     

基于混沌粒子群算法的单台批加工设备调度

提出了一种基于混沌粒子群算法的双层调度方法。双层调度的外层基于加工时间最小的目标构建组批方法集,然后将其作为内层算法的搜索空间;双层调度的内层以加工时间最小为适应值函数,采用混沌粒子群算法求解批次排序的最优解,得到最优的组批方法及其排序。通过仿真验证了该算法在搜索时间和搜索精度的可行性。     

基于动态机器故障率的并联加工系统资源多目标调度

针对并联加工系统,为处理生产过程中因机器故障导致的加工资源动态调度问题,考虑系统资源负荷和故障率存在的动态相互制约关系,依据子周期划分的不同策略,对调度后的加工总时间、预防维修时间和系统可靠度进行量化研究,构建多目标动态调度模型。采用基于Pareto熵的多目标粒子群算法对模型进行求解,通过改进个体最优解选择策略,提升最优解选取的多样性;通过差熵来估计种群所处进化状态,改进算法的搜索能力。以某企业的加工系统为实例,利用该模型进行加工资源调度方案设计,通过模糊决策得到的多目标权重,选取最优调度方案,并进行多个参数对比,验证在加工资源调度过程中,集成考虑动态机器故障率和基于机器最大役龄约束划分子周期的可行性,不仅可以完成订单准时交付,还可以使企业有更好的柔性去应对可能到来的紧急订单。     

半导体生产线群体智能调度模型研究

模仿蚁群生态系统，提出了半导体生产线群体智能调度模型。该调度模型由集中优化控制MAS(C-MAS)与分布式动态控制人工蚁群MAS(D-MAS)构成，实现了优化控制与动态调度的优势互补。其中，C-MAS主要由仿真Agent与调度Agent构成；D-MAS由一个蚁后Agent和代表半导体生产线物理实体的多个工件蚂蚁Agent与多个资源蚂蚁Agent构成，蚂蚁Agent通过基于信息素的间接交互实现半导体生产线的动态调度，蚁后Agent用于实现C-MAS与D-MAS的紧密协作，从而获得半导体生产线优化的动态智能调度。给出了基于群体智能调度模型的智能动态调度算法，对群体智能调度模型进行了定性评价。     

Real-Time Manipulation of Alternative Routeings in Flexible Manufacturing Systems: A Simulation Study

This paper presents the results of a simulation study of a typical flexible manufacturing system (FMS) that has routeing flexibility. The objective is this study is to test the effectiveness of the dissimilarity maximisation method (DMM) for real-time FMS scheduling. DMM is an alternative process plan selection method developed for routeing selection in off-line FMS sched-uling. An integrated framework that consists of a computer simulation model, which mimics a physical system, a C++ module, and a linear program solver is used to evaluate the effects of various operational control rules on the system performance. The hypothetical FMS employed in this study consists of seven machining centres, a loading and an unloading area, and six different part types. Owing to the existence of identical machining centres in the system, the part types have alternative routeings. For selecting an incoming part and later routeing it to a machining centre for its next operation, three control rules, namely, first-in first-out/first available (FIFO/FA), equal probability loading (EPL), and dissimilarity maximisation method/first-in first-out (DMM/ FIFO) are used. In this study, DMM is 1. Used as a real-time decision-making tool to select routeings for the parts that are in the system. 2. Tested and benchmarked against FIFO/FA and EPL. The results show that DMM/FIFO outperforms FIFO/FA and EPL on system throughput. Other measures such as average waiting time, average transportation time, and percentage utilisation rates are also investigated to provide insights for the effectiveness of the DMM rule for real-time FMS control applications.     

Job shop scheduling in real-time cases

A real-time scheduling algorithm is proposed, that is, to first make a fuzzy classification for the operations of jobs in real-time and then, according to their fuzzy sort, to schedule them with the heuristic. The heuristic is obtained by training a neural network offline with the genetic algorithm. Based on these ideas a real-time scheduler is built with neuro-fuzzy network (NFN). Finally the simulation for the real-time scheduling and the rescheduling are made. The results show that the real-time scheduling algorithm is effective and highly efficient compared to the first in and first out (FIFO) and the Lagrangian relaxation (LR) method.     

一种改进的半导体生产线批处理机调度策略研究

研究了半导体生产线上有加工类型限制的并行批处理机组调度问题,其目标是最小化总加权拖期交货损失,从而最大程度地满足客户的按期交货要求.在研究了ATC-BATC和DBDH规则的基础上,提出了一种改进的ATC-BATC调度规则,通过反馈控制局部在制品数量来改善系统性能,并通过仿真证明该规则能更有效地减小总加权拖期交货损失.     

半导体生产线动态调度方法研究

为了满足半导体生产线多目标优化要求,提出了多目标优化动态调度规则。它由正常状态调度规则、瓶颈设备较低在制品水平的调度规则、非瓶颈设备较高在制品水平的调度规则、多批加工设备调度规则以及紧急工件调度规则五种类型的调度规则组成。使用具有半导体生产线本质特征的简化模型,将多目标优化动态调度规则与先入先出法、最早交货期法、临界值法进行了比较,结果表明,多目标优化动态调度规则能够改善半导体生产线的整体性能,更好地优化模型的生产率、加工周期与在制品水平,最终提高准时交货率。     

A study of the flexible job shop scheduling problem with parallel machines and reentrant process

This paper develops a scheduling algorithm for the job shop scheduling problem with parallel machines and reentrant process. This algorithm includes two major modules: the machine selection module (MSM) and the operation scheduling module (OSM). An order has several jobs and each job has several operations in a hierarchical structure. The MSM helps an operation to select one of the parallel machines to process it. The OSM is then used to schedule the sequences and the timing of all operations assigned to each machine. A real-life weapons production factory is used as a case study to evaluate the performance of the proposed algorithm. Due to the high penalty of delays in military orders, the on-time delivery rate is the most important performance measure and then makespan is the next most important measure. Well-known performance measures in the scheduling literature, such as maximum lateness and average tardiness, are also evaluated. The simulation results demonstrate that the MSM and OSM using the combination of earliest due date (EDD), the operations’ lowest level code (LLC) of the bill of materials (BOM), and the longest processing time (LPT) outperforms the other scheduling methods.     

基于瓶颈分析的优先权调度算法研究

制造执行系统在制造系统中的广泛应用,为实现生产车间的动态实时调度提供了可能性,也对实时调度算法提出了更高的要求。调度算法必须在考虑求解速度的前提下来求得满意解。根据优化生产技术的解决车间调度问题的思想,基于转移瓶颈启发式算法,提出了一种基于瓶颈分析的优先权调度算法,用于解决制造执行系统中的动态实时调度问题。     

复合优先级控制策略研究及其在半导体生产线调度中的应用

复合优先级控制策略从提高制造系统综合性能的角度出发,利用复合优先级系数反映系统内的局部信息及全局信息对待加工任务优先级的影响;通过计算比较待加工任务的复合优先级系数,可确定哪个工件先进行加工,还可确定新工件投入生产线进行加工的时间。仿真结果表明,所提出的复合优先级控制策略具有综合优化系统性能指标的能力,可同时降低平均加工周期及其方差,减少在制品WIP数,提高生产率,缩短工件的等待时间。     

基于粒子群优化的开放式车间调度

开放式车间调度(OSP)是重要的调度问题,它在制造领域中的应用非常广泛。优化调度算法是调度理论的重要研究内容。基于人工智能的元启发式算法是解决该问题的常用方法。分析了一种新的元启发式算法——粒子群优化(PSO)在信息共享机制上的缺陷,提出新的基于群体智能的信息共享机制。在该信息共享机制的基础上, 设计新的基于PSO的元启发式调度算法——PSO-OSP。该算法利用问题的邻域知识指导局部搜索,可克服元启发式算法随机性引起的盲目搜索。该算法应用于开放式车间调度问题的标准测试实例。仿真结果显示,PSO-OSP算法在加快收敛速度的同时提高了开放式车间调度解的质量。     

改进蚁群算法对车辆配送二次调度的Matlab实现

货物配送中在保证时间窗的前提下,需要同一辆车进行二次或多次配送,以提高单车利用率,节约成本。针对基于时间轴的车辆调度模型,改进蚁群算法转移概率公式,并引入了上下界限来约束信息素的更新;对可行解进行有效合并,得出最佳运输车辆数,并对车辆行驶状态进行实时监控。最后通过对模型进行Matlab仿真试验,验证了算法的良好性能。     

Scheduling optimisation of flexible manufacturing systems using particle swarm optimisation algorithm

The increased use of flexible manufacturing systems (FMS) to efficiently provide customers with diversified products has created a significant set of operational challenges. Although extensive research has been conducted on design and operational problems of automated manufacturing systems, many problems remain unsolved. In particular, the scheduling task, the control problem during the operation, is of importance owing to the dynamic nature of the FMS such as flexible parts, tools and automated guided vehicle (AGV) routings. The FMS scheduling problem has been tackled by various traditional optimisation techniques. While these methods can give an optimal solution to small-scale problems, they are often inefficient when applied to larger-scale problems. In this work, different scheduling mechanisms are designed to generate optimum scheduling; these include non-traditional approaches such as genetic algorithm (GA), simulated annealing (SA) algorithm, memetic algorithm (MA) and particle swarm algorithm (PSA) by considering multiple objectives, i.e., minimising the idle time of the machine and minimising the total penalty cost for not meeting the deadline concurrently. The memetic algorithm presented here is essentially a genetic algorithm with an element of simulated annealing. The results of the different optimisation algorithms (memetic algorithm, genetic algorithm, simulated annealing, and particle swarm algorithm) are compared and conclusions are presented .     

Data-based scheduling framework and adaptive dispatching rule of complex manufacturing systems

Based on the analysis of the differences and relations between traditional and data-based scheduling methods for complex manufacturing systems, a data-based scheduling framework was proposed and discussed for its implementation into a semiconductor manufacturing system. The state-of-the-art research on the key technologies of data-based scheduling was then introduced together with their development trends. By taking a real wafer fabrication facility (fab) as an example, an adaptive dispatching rule (ADR) was developed. Firstly, a simulation system for the fab was developed, and study samples were generated by simulation. Then, the relations between the parameters of ADR and real-time running state of the fab were obtained by learning with an integration of a binary regression model, backward propagation neuro-network, and particle swarm optimization algorithm from these study samples to realize the adaptive regulations of these parameters of ADR. Finally, ADR was integrated with the simulation system. The simulation results showed that ADR had a positive effect on the operational performance of the fab. Its “move” performance was increased by 2.41 and 7.24 % for the cases of 70 % and 90 % workload, respectively.     

加工时间不确定的柔性作业车间鲁棒调度方法

针对中小批量环境下加工时间不确定的柔性作业车间调度问题,采用冗余处理方法构建了以最大完工时间为目标的鲁棒调度模型。为降低算法的搜索规模和提高算法的求解速度,提出了顺序搜索机制,并设计两阶段遗传算法,分阶段获取冗余状态和最优结果。采用某柔性生产线的数据进行正交试验,优化了算法关键参数,并构建了柔性生产线仿真模型,对调度结果的鲁棒性和优化目标性能进行了分析。结果表明,该算法在目标性能和鲁棒性上都显著优于标准遗传算法,能有效处理加工时间不确定的柔性作业车间调度问题。     

Intelligent scheduling of FMSs with inductive learning capability using neural networks

With the growing uncertainty and complexity in the manufacturing environment, most scheduling problems have been proven to be NP-complete and this can degrade the performance of conventional operations research (OR) techniques. This article presents a system-attribute-oriented knowledge-based scheduling system (SAOSS) with inductive learning capability. With the rich heritage from artificial intelligence (AI), SAOSS takes a multialgorithm paradigm which makes it more intelligent, flexible, and suitable than others for tackling complicated, dynamic scheduling problems.SAOSS employs an efficient and effective inductive learning method, a continuous iterative dichotomister 3 (CID3) algorithm, to induce decision rules for scheduling by converting corresponding decision trees into hidden layers of a self-generated neural network. Connection weights between hidden units imply the scheduling heuristics, which are then formulated into scheduling rules. An FMS scheduling problem is also given for illustration. The scheduling results show that the system-attribute-oriented knowledge-based approach is capable of addressing dynamic scheduling problems.     

基于改进粒子群算法的车间作业排序的优化设计

兼顾车间作业排序中的制造周期和机器利用率,建立了以最小化最大完工时间为主目标、以最大化机器利用率为从目标的优化模型。设计了引入自适应技术的惯性权重,使基本粒子群算法的学习因子可动态变化地改进粒子群算法,并用该改进后的算法对车间作业排序进行了优化设计。实例研究表明：改进后的粒子群算法在收敛速度和收敛可靠性上均优于未改进的粒子群算法,在求解车间作业排序问题的应用中具有更高的求解质量。