Scheduling cluster tools for concurrent processing of two wafer types with PM sharing

We examine cyclic scheduling of single-armed and dual-armed cluster tools that concurrently process two wafer types by sharing a process module (PM). Because a PM is shared by two different wafers, the backward and swap sequences, which are prevalently used for single-armed and dual-armed tools without such complexity, respectively, are not effective. We therefore propose new sequences, called alternating backward and alternating swap sequences, for steady cycles of single-armed and dual-armed tools, respectively. We then develop optimality conditions for which the proposed sequences achieve the minimum cycle times in a fundamental cycle, and show that the optimality conditions hold for most practical cases. We also develop a condition for which a shared PM becomes the bottleneck and hence the PM sharing increases the cycle time. For general cycles, we propose heuristic scheduling methods that combine both the alternating backward (or swap) sequence and the conventional backward (or swap) sequence. Finally, we experimentally verify the efficiency and effectiveness of the proposed algorithm for dual-armed cluster tools.     

Scheduling dual-armed cluster tools with cleaning processes

¹ 1. The first author’s current address is the Department of Industrial Engineering and Operations Research, University of California, Berkeley, CA 94720, USA. Modern cluster tools tend to clean the chambers each time a specified number of wafers have been processed. Such cleaning processes are intended to reduce the quality risk due to the residual chemicals within the chambers. When such cleaning processes are introduced, the conventional scheduling methods for cluster tools, such as the swap sequence for dual-armed cluster tools, are no longer effective. Therefore, we examine a cyclic scheduling problem for a dual-armed cluster tool that performs cleaning processes periodically. We first identify sufficient conditions for which the conventional backward and swap sequences give the minimum cycle time. We then develop a systematic way of generating all feasible robot task sequences, for which the cycle times can be computed. However, the number of feasible robot task sequences increases too fast to be enumerated as the cleaning cycle and the number of process steps increase. To address the complexity problem, we propose two heuristic scheduling strategies and compare them with the conventional scheduling methods and the lower bound of each schedule, respectively, to verify their effectiveness experimentally.     

Optimal scheduling for sequentially connected cluster tools with dual-armed robots and a single input and output module

We examine a cyclic scheduling problem of sequentially connected cluster tools with a single input and output module, which includes multi-cluster tools and linear cluster tools. Every component tool has a dual-armed robot, and chambers are parallelised for a long process step. An intermediate buffer between each pair of adjacent component tools has a limited capacity, and all processed wafers should return to the input and output module. To examine the scheduling problem, we first compute workloads of the process steps and robots to obtain a lower bound on the tool cycle time. We then identify a rule of assigning the chambers to the process steps that makes the tool cycle time independent of the order of using the parallel chambers. We also propose a simple robot task sequence which is modified from the well-known swap sequence for each component tool. We prove that the modified swap sequence is optimal when one of the process steps, not a robot, is the bottleneck. We also present a scheduling strategy which controls robot task timings to deal with interference of wafer flows between each pair of adjacent component tools. Finally, we perform numerical experiments to show the performance of the proposed sequence.     

基于eM-Plant求解具有晶圆重入过程的单臂组合设备最优调度的仿真方法

在半导体组合设备的晶圆加工过程中,常常要求晶圆重入,原子层沉积加工(ALD)就是典型的晶圆加工过程。对这样的系统实施调度是一个挑战性的问题。研究单臂组合设备的调度问题,目的在于求解ALD重入过程的最优调度。建立了系统的Petri网模型,利用该模型,可以通过比较非常有限的若干调度的生产节拍而得到其重入过程的最优调度。基于所建立的Petri网模型,开发出了基于eM-Plant的仿真系统,可以有效地获得给定调度的生产节拍。给定系统的参数可以有效地求得重入过程的最优调度。文中应用例子来说明所提出的方法的应用。     

Modelling and scheduling analysis of multi-cluster tools with residency constraints based on time constraint sets

This paper is dedicated to the scheduling problem of multi-cluster tools with process module residency constraints and multiple wafer product types. The problem is formulated as a non-linear programming model based on a set of time constraint sets. An effective algorithm called the time constraint sets based (TCSB) algorithm is presented as a new method to schedule the transport modules to minimise the makespan of a number of wafers. In approach, time constraint sets are maintained for all the resources and necessary operations to exploit the remaining production capacities during the scheduling process. To validate the proposed algorithm on a broader basis, a series of simulation experiments are designed to compare our TCSB algorithm with the benchmark with regard to cluster factor, configuration flexibilities and the variation of the processing times and residency constraint times. The results indicate that the proposed TCSB algorithm gives optimal or near optimal scheduling solutions in most cases.     

基于Petri网原油处理短期生产计划可调度性分析

由于原油运作短期调度的复杂性,现有的建模过程会忽略一些关键约束,导致不可行解。为了避免这个问题,从控制理论的角度研究短期调度问题,将短期调度中的一个运作决策作为一个控制。基于这个观点,对厂区油罐的数量以及它们的容量、厂区油罐中不同类型原油的库存、管道的运输速率存在的情况,利用Petri网分析其可行性。最后,利用一个实例对短期调度的可行性条件进行验证。     

Production scheduling using adaptable fuzzy logic with genetic algorithms

Production scheduling for a flexible manufacturing environment must satisfy multiple conflicting criteria. Whilst estimation and modelling of capacity is facilitated by commercially available tools, the actual release strategy of orders into the system is still subject to considerable research as improved solutions over conventional dispatching heuristics are sought. An order release mechanism incorporating an adaptable fuzzy logic system enhanced by genetic algorithms is proposed. Through the use of fuzzy logic, the system can consider multiple criteria and rapidly determine solutions of consistently high quality. Adaptability ensures that the solution quality is maintained throughout the life of the system. The subsequent application of a genetic algorithm follows an efficient optimization path, since the initial solution derived through fuzzy logic is known to be good. The system developed, using the combined methodology, was tested on a discrete event simulation model and showed measurable benefits in schedule performance against commonly implemented dispatching heuristics.     

Combined scheduling algorithm for re-entrant batch-processing machines in semiconductor wafer manufacturing

In this paper, a new combined scheduling algorithm is proposed to address the problem of minimising total weighted tardiness on re-entrant batch-processing machines (RBPMs) with incompatible job families in the semiconductor wafer fabrication system (SWFS). The general combined scheduling algorithm forms batches according to parameters from the real-time scheduling simulation platform (ReS²), and then sequences batches through slack-based mixed integer linear programming model (S-MILP), which is defined as batch-oriented combined scheduling algorithm. The new combined scheduling algorithm obtains families’ parameters from ReS² and then sequences these families through modified S-MILP, which is defined as family-oriented combined scheduling algorithm. With rolling horizon control strategy, two combined scheduling algorithms can update RBPMs scheduling continually. The experiments are implemented on ReS² of SWFS and ILOG CPLEX, respectively. The results demonstrate the effectiveness of our proposed methods.     

Multi-step ART1 algorithm for recognition of defect patterns on semiconductor wafers

The integrated circuits (ICs) on wafers are highly vulnerable to defects generated during the semiconductor manufacturing process. The spatial patterns of locally clustered defects are likely to contain information related to the defect generating mechanism. For the purpose of yield management, we propose a multi-step adaptive resonance theory (ART1) algorithm in order to accurately recognise the defect patterns scattered over a wafer. The proposed algorithm consists of a new similarity measure, based on the p-norm ratio and run-length encoding technique and pre-processing procedure: the variable resolution array and zooming strategy. The performance of the algorithm is evaluated based on the statistical models for four types of simulated defect patterns, each of which typically occurs during fabrication of ICs: random patterns by a spatial homogeneous Poisson process, ellipsoid patterns by a multivariate normal, curvilinear patterns by a principal curve, and ring patterns by a spherical shell. Computational testing results show that the proposed algorithm provides high accuracy and robustness in detecting IC defects, regardless of the types of defect patterns residing on the wafer.     

Manufacturing system development framework using a data models driven approach

We report on a data models driven approach for the development of a manufacturing system framework. Relational data schemes are used for the creation of data models. A modular approach is adopted for the creation of complex manufacturing system configurations. Each module is created by selecting the appropriate resources from the data models and is represented as a standard module template in the module library for future use. Several modules can be integrated together through an automated guided vehicles system to model a complex manufacturing system. Each module is autonomous in creating and executing its plans, but cooperative with other modules to realize the system's overall goals. For effective control of parts in the system, an event-based simulation strategy is implemented. The developed algorithm can resolve the situations arising due to any deadlock or conflict during a simulation run. System data tables are regularly updated as and when an event occurs in the system. The updated information includes machine status, part location and status, and capacity status, etc. of all parts and resources. A case study comprising eight part types with five different flow patterns in a three-module system is taken to show the effectiveness of the proposed approach.     

Information sharing in multiviewpoint injection moulding design and manufacturing

Global competitiveness challenges manufacturing industry to bring to market well designed and manufactured new products at competitive prices in as short a lead-time as possible. To achieve this, inputs are needed from experts in a multitude of disciplines as well as from customers and suppliers. While the use of design teams is achieving success, there is a need for modern software tools that support design and manufacture to be radically improved. Typically, design for manufacture software systems does not support the multiple views of information needed for information systems to be able to support multiple manufacturing perspectives. This paper proposes a linked, multiviewpoint, product model structure as a means of supporting the many views of product information necessary in team-based design and manufacturing systems. To test the ideas developed, an experimental product model has been implemented using an object-oriented database that focuses on views related to the design and manufacture of injection-moulded products. The experimental system explores the relationships between three views: mouldability, cavity design and cavity machining views.     

Closed loop control-based real-time dispatching heuristic on parallel batch machines with incompatible job families and dynamic arrivals

In this paper, a real-time closed loop control dispatching heuristic (RCLC) algorithm is proposed to address the scheduling problem of parallel batch machines with incompatible job families, limited waiting time constraints, re-entrant flow and dynamic arrivals in the diffusion and oxidation areas of a semiconductor wafer fabrication system (SWFS), which is known to be strongly NP-hard. The basis of this algorithm is the information of lots in the buffer when the parallel batch machines are idle and available. In RCLC, if the number of any family lots is less than the maximum batch size, the dispatching heuristic can be seen as a pull–pull–push–push (P⁴) strategy; otherwise, a genetic algorithm (GA). A look-itself strategy, P⁴ strategy and GA can build a closed loop control system. The experiments are implemented on the Petri nets-based real-time scheduling simulation platform of SWFS, and demonstrate the effectiveness of our proposed method.     

A self‐tuning control approach of mechanical manipulators

Abstract

The dynamic model of a manipulator system is a time‐varying highly nonlinear coupling equation set. When the moving speed increases or the payload, compared to its own weight, is no longer small, the performance of the conventional control schemes is not satisfactory for precision industrial application. Here a new adaptive control approach is developed for the manipulators to solve these problems. This algorithm directly uses a nonlinear dynamic model in the controller design to account for the nonlinear effects of the system. The least‐square time‐varying parameter identification scheme has been used to identify the change in configuration and payload. The simulation results show that this new approach has a very good trajectory tracking performance.     

Two-step approach to robust deadlock control in automated manufacturing systems with multiple resource failures

ABSTRACT

In an automated manufacturing system (AMS), resource failures are inevitable, which renders the existing deadlock control policies for AMSs without considering resource failures ineffective. For the AMSs with multiple unreliable resources, in this paper, a method is developed for robust deadlock resolution using the framework of Petri nets (PNs). The considered AMSs are modeled with PN models called system of simple sequential process with resources (S³PR). An unreliable S³PR (U-S³PR) is obtained by adding recovery subnets that model the resource failures and recovery procedures of the places where resource failures may happen. Based on the model, a two-step approach is developed to design a robust controller. At Step 1, we use a siphon-based deadlock control method to analyze the behavior with resource failures and propose an incomplete robust deadlock controller for a U-S³PR. At Step 2, a reachability-graph-analysis-based method is utilized to consummate the robust deadlock controller. Then, a robust liveness-enforcing supervisor is derived to make an unreliable S³PR live even if multiple types of resources break down. Finally, the proposed method is illustrated by using an example.     

Integration of input shaping technique with interpolation for vibration suppression of servo-feed drive system

Abstract

It is essential to suppress vibrations of the feed drive system in order to achieve high speed and high precision machining. To analyze the dynamic response of the feed drive system, a more complete model is derived in this paper which considers the bed, longitudinal and torsional modes. The dynamic model is represented by the block diagram and is incorporated with servo controls to form a closed-loop system. To alleviate vibrations of the servo-feed drive system, a full order modified input shaping with zero vibration (FMISZV) algorithm integrated into the CNC interpolator is developed. The performance of vibration suppression using the FMISZV is compared to those of conventional input shapers, such as zero vibration with derivation (ZVD) and conventional linear acceleration/deceleration (Acc/Dec) interpolator (CLAI). It is shown that the FMISZV not only has the notch filter effect, but also exhibits low-pass filter behavior for high frequency modes. Simulation and experimental results demonstrate that the FMISZV can outperform the ZVD and CLAI.     

Simple quantum information algorithms in cavity QED

Abstract

Cavity quantum electrodynamics has already proven to be a system capable of demonstrating basic tools in quantum information theory. By combining these tools, we show how simple quantum information protocols could be implemented. We will focus on the examples of the Grover search algorithm and quantum cloning.     

Deadlock avoidance algorithm for flexible manufacturing systems by calculating effective free space of circuits†

Modern flexible manufacturing systems (FMS) are highly automated and flexible in which raw parts of various types are processed concurrently. Deadlock issue arises easily in these systems due to shared equipment usage and high production flexibility. It was indicated in Wysk et al . (Wysk, R., Yang, N. and Joshi, S., Detection of deadlocks in flexible manufacturing cells. IEEE Trans. Rob. Auto., 1991, 7, 853–859.) that when a manufacturing system is modelled by a digraph, existence of circuits in such a graph is a necessary condition for deadlock. Deadlock avoidance for systems with free choice in part routing has been well studied in the Petri net formalism, however, we have not found that it has been studied systematically in the digraph formalism. Because of choices introduced, part flow dynamics become much more complex. Based on our previous work Zhang et al . (Zhang, W., Judd, R.P. and Paul P., Evaluating order of circuits for deadlock avoidance in a flexible manufacturing system. Int. J. Prod. Res., 2006, 44, 5247–5259.), this paper presents a deadlock avoidance algorithm for FMS, which allows free choices in part routing by calculation of effective free space of circuits of the digraph model. The algorithm is highly permissive since the effective free space calculation captures more parts flow dynamics, especially when there exist multiple knots in the digraph model. And it runs in polynomial time once the set of circuits of the digraph is computed offline. Simulation results on selected examples are given.     

Design of flexible manpower line walk cycles for a fixed number of operators

Flexible manpower lines (FMLs) are a form of flow process line in which operators are allocated ‘walk cycles’, i.e. a repetitive sequence in which to load and unload machine tools. The effective design of such lines is normally achieved with the expectation that operators without full walk cycles, i.e. those that do not require a full Takt time to accomplish, can complete their walk cycles at an adjoining FML. However, an alternative FML design strategy is possible in cases where no adjoining FML exists or it is not possible for operators to move between work areas. This strategy involves determining the minimum Takt time and the associated operator walk cycles at which the FML can operate under a fixed number of operators. To solve this type of problem, a genetic algorithm that make use of a novel crossover operator has been developed that can design FMLs. The genetic algorithm is capable of generating, for a specific Takt time and fixed number of operators, FMLs with high-quality, near-optimal operator walk cycles. Solutions for the fixed manpower case were then identified by performing multiple genetic algorithm runs to find the best walk cycles at various Takt times.     

Modeling and Change Detection for Count-Weighted Multilayer Networks

Abstract

In a typical network with a set of individuals, it is common to have multiple types of interactions between two individuals. In practice, these interactions are usually sparse and correlated, which is not sufficiently accounted for in the literature. This article proposes a multilayer weighted stochastic block model (MZIP-SBM) based on a multivariate zero-inflated Poisson (MZIP) distribution to characterize the sparse and correlated multilayer interactions of individuals. A variational-EM algorithm is developed to estimate the parameters in this model. We further propose a monitoring statistic based on the score test of MZIP-SBM model parameters for change detection in multilayer networks. The proposed model and monitoring scheme are validated using extensive simulation studies and the case study from Enron E-mail network.     

MODELING AND NONLINEAR CONTROL OF A CONTINUOUS KCI-NaCI CRYSTALLIZER

ABSTRACT

A controller is developed by combining the extended linear quadratic matrix control (EQDMC) and neural network algorithms. The dynamic neural network scheme is used to identify the process and generate a nonlinear model. The control algorithm is applied to a multi-input multi-output (MIMO) evaporative cooling KCl-NaCl-H₂O crystallizer. Closed loop responses of the system using the proposed algorithm and those of PID controllers are compared. It is shown that in all cases, the response of the proposed controller to step changes in setpoints is faster than the PID controllers.