Complex Job Shop Research Articles

Automated model generation framework for material flow simulations of production systems

Owing to increasingly important drivers for change, such as automation, digitisation, and dynamic demand patterns, simulation models of production systems become outdated rapidly. At the same time, building simulation models often requires much time, cost, and expertise, especially when dealing with complex job shop production systems. To address these challenges, an automated simulation model generation (ASMG) framework for material flow simulation of production systems is presented. This framework contains multiple approaches to infer routeing, control and temporal aspects from event-based data. To achieve this, methods from process mining (PM) and machine learning (ML) are applied. Additionally, the suitability of Coloured Petri Nets (CPNs) to serve as conceptual and operational simulation models is examined. The inferred simulation models have high validity when compared to the real system concerning the KPIs machine utilisation, throughput, and work in process. It is shown, that most model elements can be inferred very well, in particular process routes, processing times, and resource selection rules. This proof of concept serves as a foundation for research on detection approaches inferring further model elements such as setup times accurately.

Complex Job Shop Simulation “CoJoSim”—A Reference Model for Simulating Semiconductor Manufacturing

The manufacturing industry is facing increasing volatility, uncertainty, complexity, and ambiguity, while still requiring high delivery reliability to meet customer demands. This is especially challenging for complex job shops in the semiconductor industry, where the manufacturing process is highly intricate, making it difficult to predict the consequences of changes. Although simulation has proven to be an effective tool for optimizing manufacturing processes, reference data sets and models often produce disparate and incomparable results. CoJoSim is introduced in this article as a reference model for semiconductor manufacturing, along with an associated reference implementation that accelerates the implementation and application of the reference model. CoJoSim can serve as a testbed and gold standard for other implementations. Using CoJoSim, different dispatching rules are evaluated to demonstrate an improvement of almost 15 percentage points in adherence to delivery dates compared to the reference. Findings emphasize the importance of optimizing setup time, particularly in products with high variance, as it significantly impacts adherence to delivery dates and throughput. Moving forward, future applications of CoJoSim will evaluate additional dispatching rules and use cases. Combining CoJoSim with dispatching methods that integrate manufacturing and supply networks to optimize production planning and control through reinforcement-learning-based agents is also planned. In conclusion, CoJoSim provides a reliable and effective tool for optimizing semiconductor manufacturing and can serve as a benchmark for future implementations.

A new algorithm of the scheduling of a flexible manufacturing system based on genetic algorithm

In the flexible manufacturing system, a reasonable production scheduling is crucial in shortening the processing completion time and improving the equipment utilization. Traditional manual scheduling cannot effectively solve the complex workshop scheduling problems and cannot provide a scheduling solution that meets the requirements in a short period of time, which can lead to a decrease in processing efficiency. Aiming at the complex job shop scheduling problem, the genetic algorithm is used to find the optimal scheduling solution in this study, taking the number of overdue jobs, the total overdue time, the job completion time, the comprehensive load rate and the maximum load rate of the machine tool as the performance indicators of the scheduling algorithm. The chromosomes are designed as process gene chain and equipment gene chain to improve the diversity and the robustness to scheduling problems of chromosome through crossover, variation, selection and other processes. The impact of different parameter settings on the performance indicators of each scheduling algorithm is researched by adjusting the four algorithm-related parameters, and there has been a certain improvement in the results of the scheduling problems. This study provides a reference for the design and optimization of production scheduling algorithm based on genetic algorithm.

Real-time scheduling simulation optimisation of job shop in a production-logistics collaborative environment

ABSTRACT In a complex and dynamic job shop containing logistics factor, schedule needs to be generated rapidly, so the real-time scheduling method is more suitable for such scenario. Such method takes advantage of local information within a short time due to the rapid changes of information under uncertain environment. Therefore, how to make use of the future information by prediction while ensuring the robustness of schedule is a valuable problem. To solve it, firstly, a new real-time scheduling model and algorithm is proposed. There is a new kind of release moment of task information which can give AGVs the longest time to prepare for the task than existing research. Secondly, a real-time information update mechanism is designed to increase schedule’s robustness. Finally, a large-scale and dynamic job shop simulation experimental platform is developed. Dynamic factors include the random insertion of orders and failures of equipment. Results show that the method proposed outperforms existing research in terms of customer satisfaction, equipment utilisation and energy consumption. The robustness of schedule can also be acceptable. This paper also finds a rule that in job shop with the large proportion of logistics transportation time, the above method can achieve more competitive results.

A Hybrid of Genetic Algorithm and Greedy Batch Strategy for Complex Job-Shop Scheduling Problem

A Hybrid of Genetic Algorithm and Greedy Batch Strategy for Complex Job-Shop Scheduling Problem

Self-Supervised Long-Short Term Memory Network for Solving Complex Job Shop Scheduling Problem

Self-Supervised Long-Short Term Memory Network for Solving Complex Job Shop Scheduling Problem

Design of autonomous production using deep neural network for complex job

Deep Neural Network (DNN) in recent era offers new opportunities to manage the production systems with increasing complexity. In this paper, DNN is trained with discrete event simulation and a process based association to make the job shop an autonomous one. This intelligent system operates well in complex environment with constrained time limits while making optimal decisions. The DNN is henceforth combined with constrained time limits for the process of production control. The system is implemented typically in semiconductor manufacturing industries on complex job shops of a wafer fab case. The DNN is trained in such a way that it operates in complex environment with timing constraints that ships the job in accurate way without flaws in operation. The DNN rewards the selection with reduced time constraint, which tends to operates with most critical batch list. The study therefore shows that the DNN manages well the timing constraints than the standard benchmark technique.

Multi-variate time-series for time constraint adherence prediction in complex job shops

One of the most complex and agile production environments is semiconductor manufacturing, especially wafer fabrication, as products require more than several hundred operations and remain in Work-In-Progress for months leading to complex job shops. Additionally, an increasingly competitive market environment, i.e. owing to Moore’s law, forces semiconductor companies to focus on operational excellence, resiliency and, hence, leads to product quality as a decisive factor. Product-specific time constraints comprising two or more, not necessarily consecutive, operations ensure product quality at an operational level and, thus, are an industry-specific challenge. Time constraint adherence is of utmost importance, since violations typically lead to scrapping entire lots and a deteriorating yield. Dispatching decisions that determine time constraint adherence are as a state of the art performed manually, which is stressful and error-prone. Therefore, this article presents a data-driven approach combining multi-variate time-series with centralized information to predict time constraint adherence probability in wafer fabrication to facilitate dispatching. Real-world data is analyzed and different statistical and machine learning models are evaluated.

An imperialist competition algorithm using a global search strategy for physical examination scheduling.

The outbreak of the novel coronavirus clearly highlights the importance of the need of effective physical examination scheduling. As treatment times for patients are uncertain, this remains a strongly NP-hard problem. Therefore, we introduce a complex flexible job shop scheduling model. In the process of physical examination for suspected patients, the physical examiner is considered a job, and the physical examination item and equipment correspond to an operation and a machine, respectively. We incorporate the processing time of the patient during the physical examination, the transportation time between equipment, and the setup time of the patient. A unique scheduling algorithm, called imperialist competition algorithm with global search strategy (ICA_GS) is developed for solving the physical examination scheduling problem. A local search strategy is embedded into ICA_GS for enhancing the searching behaviors, and a global search strategy is investigated to prevent falling into local optimality. Finally, the proposed algorithm is tested by simulating the execution of the physical examination scheduling processes, which verify that the proposed algorithm can better solve the physical examination scheduling problem.

Designing an adaptive production control system using reinforcement learning

Modern production systems face enormous challenges due to rising customer requirements resulting in complex production systems. The operational efficiency in the competitive industry is ensured by an adequate production control system that manages all operations in order to optimize key performance indicators. Currently, control systems are mostly based on static and model-based heuristics, requiring significant human domain knowledge and, hence, do not match the dynamic environment of manufacturing companies. Data-driven reinforcement learning (RL) showed compelling results in applications such as board and computer games as well as first production applications. This paper addresses the design of RL to create an adaptive production control system by the real-world example of order dispatching in a complex job shop. As RL algorithms are “black box” approaches, they inherently prohibit a comprehensive understanding. Furthermore, the experience with advanced RL algorithms is still limited to single successful applications, which limits the transferability of results. In this paper, we examine the performance of the state, action, and reward function RL design. When analyzing the results, we identify robust RL designs. This makes RL an advantageous control system for highly dynamic and complex production systems, mainly when domain knowledge is limited.

Reinforcement learning for an intelligent and autonomous production control of complex job-shops under time constraints

Reinforcement learning (RL) offers promising opportunities to handle the ever-increasing complexity in managing modern production systems. We apply a Q-learning algorithm in combination with a process-based discrete-event simulation in order to train a self-learning, intelligent, and autonomous agent for the decision problem of order dispatching in a complex job shop with strict time constraints. For the first time, we combine RL in production control with strict time constraints. The simulation represents the characteristics of complex job shops typically found in semiconductor manufacturing. A real-world use case from a wafer fab is addressed with a developed and implemented framework. The performance of an RL approach and benchmark heuristics are compared. It is shown that RL can be successfully applied to manage order dispatching in a complex environment including time constraints. An RL-agent with a gain function rewarding the selection of the least critical order with respect to time-constraints beats heuristic rules strictly by picking the most critical lot first. Hence, this work demonstrates that a self-learning agent can successfully manage time constraints with the agent performing better than the traditional benchmark, a time-constraint heuristic combining due date deviations and a classical first-in-first-out approach.

A shifting bottleneck procedure with multiple objectives in a complex manufacturing environment

Scheduling problems have been studied extensively over the years. Broad ranges of focus areas exist, from optimization to heuristics, production planning to production sequencing, customized algorithms to general-purpose algorithms, and from simple machines to complex environments. In this research, a heuristic approach has been proposed to overcome the scheduling problem on a complex job shop found at a manufacturer of commercial building products. The research is aimed at sequencing production orders in near-real-time, primarily to minimize total tardiness, but also to reduce total setup time. A layered Shifting Bottleneck Procedure is employed, with the top layer determining release dates and due dates for individual jobs, and the bottom layer applying algorithms to individual work centers. The outcome of this research is a better production schedule than current methods with minimal computation cost. The proposed framework performs well and could be applied to other production areas.

Simulation Optimization on Complex Job Shop Scheduling with Non-Identical Job Sizes

This paper addresses the complex job shop scheduling problem with the consideration of non-identical job sizes. By simultaneously considering practical constraints of sequence dependent setup times, incompatible job families and job dependent batch processing time, we formulate this problem into a simulation optimization problem based on the disjunctive graph representation. In order to find scheduling policies that minimise the expectation of mean weighted tardiness, we propose a genetic programming based hyper heuristic to generate efficient dispatching rules. And then, based on the nested partition framework together with the optimal computing budget allocation technique, a hybrid rule selection algorithm is proposed for searching machine group specified rule combinations. Numerical results show that the proposed algorithms outperform benchmark algorithms in both solution quality and robustness.

Effect of sequencing rules on order release policies in a stochastic and dynamic job shop with sequence-dependent setup times – a simulation study

Order review and release in job shop with sequence-dependent setup times is one of the most complex job shop scheduling problems. This article analyses the effect of sequencing rules on different order release policies for performance measures i.e. mean throughput time, total setups, mean setup time, number of tardy jobs and makespan in a stochastic and dynamic job shop with sequence-dependent setup time. Four sequencing rules, viz. first-come-first-serve, planned release date, earliest due date, shortest processing time and five order review and release policies i.e. aggregate workload trigger, workcentre workload trigger, corrected workload trigger, upper bound release and lancaster university management school corrected order release are taken into consideration. A simulation model using Promodel® is developed for experimental purpose. Results indicate that for a given order review and release policy, there is an effect of sequencing rule in a stochastic and dynamic environment with consideration of sequence-dependent setup time and the best performing sequencing rule is different for different order review and release policies for a given performance measure.

Smart Manufacturing Scheduling With Edge Computing Using Multiclass Deep Q Network

Manufacturing is involved with complex job shop scheduling problems (JSP). In smart factories, edge computing supports computing resources at the edge of production in a distributed way to reduce response time of making production decisions. However, most works on JSP did not consider edge computing. Therefore, this paper proposes a smart manufacturing factory framework based on edge computing, and further investigates the JSP under such a framework. With recent success of some AI applications, the deep Q network (DQN), which combines deep learning and reinforcement learning, has showed its great computing power to solve complex problems. Therefore, we adjust the DQN with an edge computing framework to solve the JSP. Different from the classical DQN with only one decision, this paper extends the DQN to address the decisions of multiple edge devices. Simulation results show that the proposed method performs better than the other methods using only one dispatching rule.

The performance of priority dispatching rules in a complex job shop: A study on the Upper Mississippi River

Many studies have derived and tested dynamic job shop priority dispatching rules using discrete event simulation models in the context of idealized job shop experimental designs. This paper extends research on evaluating priority dispatching rules in a completely reactive dynamic job shop by testing the performance of eight selected rules in a simulation model of a complex and real dynamic job shop: the Upper Mississippi River Inland Navigation Transportation System (UMR). The UMR incorporates many real-world complexities such as sequence dependent and seasonally varying stochastic job processing times, both capacitated and un-capacitated servers, and heterogeneous jobs with seasonally varying, interdependent stochastic arrivals that can balk (opt-out) at using the system in response to anticipated poor levels of service. Employing two related but different metrics, mean flow times and the total value (“utility”) of jobs processed, the results show that rules that incorporate increasingly more systemic information generally perform better as system congestion increases, particularly when balking is not allowed. However, this is not the case when customer balking is allowed, particularly for value-based priority dispatching rules. This demonstrates that the balking decision of system users has a large impact on the performance of the rules and the expected utility (value) generated by the system, particularly at high levels of system congestion.

The moderating role of master production scheduling method on throughput in job shop systems

Accurately predicting throughput is a challenging task for managers of manufacturing companies and analytical models for this purpose have been limited to small and simple production systems. Motivated by a complex and real-world case from the automotive industry, this paper first examines how realistically the throughput of complex job shop systems can be predicted based on problem characteristics and different master production scheduling (MPS) approaches. Next, it investigates how different MPS approaches moderate the relationship between problem characteristics and throughput. To achieve these aims, we develop a mixed-effects model based on operational characteristics and the MPS development method to predict the system's throughput. The analyses are based on data from a real-world case in the automotive industry and two complex job shop systems in the literature. The experimental results indicate that the throughput of job shop systems can be predicted with a high level of accuracy (R2=0.756) based on the problem's characteristics and that the predictive model cross-validates well on a holdout sample. Moreover, we observed that the MPS method can moderate the impact of problem characteristics reflecting complexity, capacity shortages, and setup requirements on throughput, with moderation especially pronounced in the presence of capacity shortages.

Scheduling recurring radiotherapy appointments in an ion beam facility

Ion beam radiotherapy is a modern form of cancer treatment that is offered in specialized facilities. Treatment consists of multiple, almost daily irradiation appointments, followed by optional imaging and control assignments. The corresponding problem of scheduling these recurring radiotherapy treatment appointments can be classified as a complex job shop scheduling problem with custom constraints, such as recurring activities, optional activities, and special time window constraints. The objective is to minimize the operation time of the bottleneck resource, the particle beam, while simultaneously minimizing any penalties arising from violations of time window constraints. The authors model the problem mathematically and introduce various customized constraints. Three metaheuristic solution approaches—namely a genetic algorithm with tailor-made feasibility-preserving crossover operators, an iterated local search, and a combination of the two approaches—all perform well on both small and large problem instances. However, the simple combination of the two stand-alone algorithms leads to best results when applied to real-world inspired problem instances.

A neighborhood for complex job shop scheduling problems with regular objectives

Due to the limited applicability in practice of the classical job shop scheduling problem, many researchers have addressed more complex versions of this problem by including additional process features, such as time lags, setup times, and buffer limitations, and have pursued objectives that are more practically relevant than the makespan, such as total flow time and total weighted tardiness. However, most proposed solution approaches are tailored to the specific scheduling problem studied and are not applicable to more general settings. This article proposes a neighborhood that can be applied for a large class of job shop scheduling problems with regular objectives. Feasible neighbor solutions are generated by extracting a job from a given solution and reinserting it into a neighbor position. This neighbor generation in a sense extends the simple swapping of critical arcs, a mechanism that is widely used in the classical job shop but that is not applicable in more complex job shop problems. The neighborhood is embedded in a tabu search, and its performance is evaluated with an extensive experimental study using three standard job shop scheduling problems: the (classical) job shop, the job shop with sequence-dependent setup times, and the blocking job shop, combined with the following five regular objectives: makespan, total flow time, total squared flow time, total tardiness, and total weighted tardiness. The obtained results support the validity of the approach.

A batch-oblivious approach for Complex Job-Shop scheduling problems

We consider a Flexible Job-Shop scheduling problem with batching machines, reentrant flows, sequence dependent setup times and release dates while considering different regular objective functions. Semiconductor manufacturing is probably one of the most prominent practical applications of such a problem. Existing disjunctive graph approaches for this combined problem rely on dedicated nodes to explicitly represent batches. To facilitate modifications of the graph, our new modeling reduces this complexity by encoding batching decisions into edge weights. An important contribution is an original algorithm that takes batching decisions “on the fly” during graph traversals. This algorithm is complemented by an integrated move to resequence and reassign operations. This combination yields a rich neighborhood that we apply within a local search and a Simulated Annealing (SA) metaheuristic. The latter is embedded in a Greedy Randomized Adaptive Search Procedure (GRASP) which is the most efficient approach. Numerical results for benchmark instances of different problem types show the generality and applicability of our approach. The conciseness of our idea facilitates extensions towards further complex constraints needed in real-world applications.