Self-growth learning-based machine scheduler to minimize setup time and tardiness in OLED display semiconductor manufacturing

The Wafer Probing Scheduling Problem (WPSP) is a practical generalization of the parallel-machine scheduling problem, which has many real-world applications, particularly, in the Integrated Circuit (IC) manufacturing industry. In the wafer probing factories, the jobs are clustered by their product types, which must be processed on groups of identical parallel machines and be completed before the due dates. The job processing time depends on the product type, and the machine setup time is sequentially dependent on the orders of jobs processed. Since the wafer probing scheduling problem involves constraints on job clusters, job-cluster dependent processing time, due dates, machine capacity, and sequentially dependent setup time, it is more difficult to solve than the classical parallel-machine scheduling problem. In this paper, we consider the WPSP and formulate the WPSP as an integer programming problem to minimize the total machine workload. We demonstrate the applicability of the integer programming model by solving a real-world example taken from a wafer probing shop floor in an IC manufacturing factory.

A branch-and-bound algorithm for the early/tardy machine scheduling problem with a common due-date and sequence-dependent setup time

This paper addresses job scheduling problems with parallel machines. To satisfy customers better in a manufacturing company, meeting due dates has been an important performance metric. Besides the numerous other factors affecting due date satisfaction, the splitting of a job through parallel machines can contribute to the reduction of production lead time, resulting in less job tardiness against their due dates. Thus, this paper presents heuristic algorithms for minimizing total tardiness of jobs to meet their due dates in a manufacturing shop with identically functioning machines. The algorithms take into account job splitting and sequence-dependent major/minor setup times. The performance of the proposed heuristics is compared with that of past three algorithms in the literature.

Machine scheduling is one of the most studied problems due to its technical challenges and prevalence in real life. In the literature, continuous- and discrete-time formulations are the two most known formulations for scheduling problems. However, continuous-time formulations often suffer from weak linear relaxations, while discrete-time formulations struggle with large numbers of variables. In contrast, the bucket-indexed formulation is an alternative that mitigates both issues by working with partial time discretization. We propose a mixed-integer linear programming model based on a bucket-indexed formulation to solve a nonpreemptive scheduling problem of identical parallel machines considering release dates, deadlines, precedence, eligibility, and machine availability constraints. We evaluate the proposed formulation against real-world instances comprising more than 400 jobs and 100 machines, comparing its performance against equivalent continuous- and discrete-time formulations. Remarkably, our formulation can be solved to optimality for all instances, outperforming both continuous- and discrete-time formulations.

The multi-stage wafer probing scheduling problem (M-WPSP) with reentry is a practical variation of the parallel-machine scheduling problem. Since the M-WPSP involves multiple product families, to be processed on multiple stages, with various job due dates, ready times, reentry, serial and batch operations, sequential-dependent setup time, it is more difficult to solve than the classical parallel-machine scheduling problems. In this paper, we consider two strategies to solve the M-WPSP with reentry, where the total machine workload must be minimized. These two strategies incorporate a global planning mechanism, in advance, to determine the required stage due date of job at each process stage to prevent the due date problems occurring at the final stage. The sequential strategy schedules the jobs at the required stages according to the sequence of manufacturing process. The parallel strategy is designed specifically for the reentrant characteristic. To evaluate the efficiency of the proposed strategies, a set of test problems involving four critical factors, the product family ratio, the temperature-change consideration, the tightness of due dates, and the ready time, are designed to test the quality of solutions under two levels of workload.

Single-machine scheduling against due dates with past-sequence-dependent setup times

This paper studies a multi-stage and parallel-machine scheduling problem with job splitting which is similar to the traditional hybrid flow shop scheduling (HFS) in the solar cell industry. The HFS has one common hypothesis, one job on one machine, among the research. Under the hypothesis, one order cannot be executed by numerous machines simultaneously. Therefore, multiprocessor task scheduling has been advocated by scholars. The machine allocation of each order should be scheduled in advance and then the optimal multiprocessor task scheduling in each stage is determined. However, machine allocation and production sequence decisions are highly interactive. As a result, this study, motivated from the solar cell industry, is going to explore these issues. The multi-stage and parallel-machine scheduling problem with job splitting simultaneously determines the optimal production sequence, multiprocessor task scheduling and machine configurations through dynamically splitting a job into several sublots to be processed on multiple machines. We formulate this problem as a mixed integer linear programming model considering practical characteristics and constraints. A hybrid-coded genetic algorithm is developed to find a near-optimal solution. A preliminary computational study indicates that the developed algorithm not only provides good quality solutions but outperforms the classic branch and bound method and the current heuristic in practice.

In this chapter, we consider some practical scheduling problems under unavailability constraints (breakdown periods, maintenance durations and/or setup times). Such problems can be met in different industrial environments and be associated to numerous real-life applications. This explains why many researchers have become interested in this subject. We aim to present the recent approaches proposed to solve these problems and to discuss their performances. This family of scheduling problems, addressed in this chapter, has been intensively studied (Kacem [8], Lee [17], Schmidt [24]). The studied criteria in this chapter are related to the flowtime minimization (the weighted and unweighted cases). The chapter is organized in two main parts. The first part focuses on the single machine scheduling problem (see Section 2). The second part is devoted to the parallel machine scheduling problem (see Section 3). In each part, we present the main contributions and explain their principles (complexity results, heuristic algorithms and their worstcase performance, existing approximation schemes, exact methods, branch-and-bound algorithms, dynamic programming, integer linear models, lower bounds. . .). Finally, Section 4 concludes the paper.

In this paper, we discuss scheduling problems in semiconductor manufacturing. Starting from describing the manufacturing process, we identify typical scheduling problems found in semiconductor manufacturing systems. We describe batch scheduling problems, parallel machine scheduling problems, job shop scheduling problems, scheduling problems with auxiliary resources, multiple orders per job scheduling problems, and scheduling problems related to cluster tools. We also present important solution techniques that are used to solve these scheduling problems by means of specific examples, and report on known implementations. Finally, we summarize some of the challenges in scheduling semiconductor manufacturing operations.

The development of a scheduling methodology for module processing in thin film transistor liquid crystal display manufacturing is presented. The problem is a parallel machine-scheduling problem with family set-up times and due dates. Two objectives are considered: total tardiness and the number of family set-ups. Two heuristics are developed based on the MULTIFIT method and tabu search. The experimental results show that the proposed algorithms give schedules that are very efficient in terms of both total tardiness and the number of family set-ups.

Scheduling uniform parallel dedicated machines with job splitting, sequence-dependent setup times, and multiple servers

A multi-population genetic algorithm to solve multi-objective scheduling problems for parallel machines

Identical parallel machine scheduling problem for minimizing the makespan is a very important production scheduling problem which has been proven to be NP-hard. The problem further compounds with additional constraints. Genetic algorithms (GA) have shown great advantages in solving the combinatorial optimization problem in view of its characteristic that has high efficiency and that is fit for practical application. In this chapter we present a spreadsheet based GA approach for minimizing the makespan for scheduling of a set of tasks for identical parallel machines and worker assignment to machines. The results obtained from the proposed approach are compared with two sets of benchmark problems consisting of 100 problems each. It has been demonstrated that the performance of proposed approach is superior to the results that have been obtained earlier. The proposed approach produces optimal solution for almost 95% of the problems demonstrating the effectiveness of the proposed approach. An empirical analysis of GA parameters has also been carried out to see the effect on the performance of the proposed algorithm.

In simple flow shop problems, each machine operation center includes just one machine. If at least one machine center includes more than one machine, the scheduling problem becomes a flexible flow shop problem (FFSP). Flexible flow shops are thus generalization of simple flow shops. Flexible flow shop scheduling problems have a special structure combining some elements of both the flow shop and the parallel machine scheduling problems. FFSP can be stated as finding a schedule for a general task graph to execute on a multiprocessor system so that the schedule length can be minimized. FFSP is known to be NP-hard. In this study, we present a particle swarm optimization (PSO) algorithm to solve FFSP. PSO is an effective algorithm which gives quality solutions in a reasonable computational time and consists of less numbers parameters as compared to the other evolutionary metaheuristics. Mutation, a commonly used operator in genetic algorithm, has been introduced in PSO so that trapping of solutions at local minima or premature convergence can be avoided. Logistic mapping is used to generate chaotic numbers in this paper. Use of chaotic numbers makes the algorithm converge fast towards near-optimal solution and hence reduce computational efforts further. The performance of schedules is evaluated in terms of total completion time or makespan (Cmax). The results are presented in terms of percentage deviation (PD) of the solution from the lower bound. The results are compared with different versions of genetic algorithm (GA) used for the purpose from open literature. The results indicate that the proposed PSO algorithm is quite effective in reducing makespan because average PD is observed as 2.961, whereas GA results in average percentage deviation of 3.559. Finally, influence of various PSO parameters on solution quality has been investigated.

This paper presents the comparative use of Simulated Annealing (SA) and Genetic Algorithm (GA) in a scheduling problem of unrelated parallel machines with set-up times. The problem accounts for allotting batched work parts to the unrelated parallel machines, where each batch is composed of a fixed number of identical jobs. Processing time of each job of a batch is determined according to the machine group as well as the batch group to which the job belongs. Major or minor set-up times are required between two subsequent batches, depending on the batch sequence but yet independent of machines. The objective of the problem is to minimise the Total Weighted Tardiness (TWT) for the unrelated parallel machine scheduling. SA and GA algorithms are proposed to obtain near-optimal solutions of the problem. The performance of proposed heuristics is compared through computational experiments with real data from dicing operations of a compound semiconductor manufacturing facility.