Real Manufacturing System Research Articles

Performance measurement of legacy equipment through its connection to the cloud

Goal: Objective focuses on development and application of an architecture for measuring the performance of legacy equipment through its connection to the cloud. Methodology: The research was approached by the Design Science Research (DSR) method, which focuses on the development and evaluation of the application of the artifact (legacy equipment performance measurement system) through its technological base. Results: Both in the development and in the evaluation of the application of the system architecture in a real manufacturing system, the results were satisfactory through the accuracy obtained of 86% in the measurement carried out. Regarding equipment performance measurement, the measured index achieved an OEE of 27% efficiency, which is considered low compared to the average of companies with world-class manufacturing. It is concluded that the results made it possible to verify the reliability of the information generated by the architecture of the application system and to measure the performance of the equipment through its applicability in a real manufacturing scenario with a focus on manufacturing management. Search limitations: The development of this research is limited to the application architecture for measuring performance for bread production line. Practical Implications: The cloud architecture contextualizes the use of IIoT technologies (sensors, devices, among others) through cloud application architecture and how this has been transforming the industry and helping in the management of manufacturing operations. Originality / Value: This research underscores the efficacy of a customized integration approach for leveraging existing technologies. While cloud-based Industrial Internet of Things (IIoT) solutions are readily available, this work transcends the limitations of off-the-shelf options by tailoring the architecture to the specific requirements of a real-world production line. This focus on customization enables the architecture to be adapted to a broad spectrum of legacy equipment and production lines across diverse industries, thereby unlocking the full potential of IIoT technologies. Consequently, this research demonstrates the replicability of the proposed methodology for various manufacturing scenarios.

Effect of Layout Discretization on the Performance of Zone Control-Based Multi-AGV Traffic Management Systems

Automatic Guided Vehicles (AGVs) are widely used in flexible manufacturing systems for material handling inside the factory. Traffic management strategies, required to guarantee a conflict-free operation of the overall fleet, discretize the workspace of the AGVs and use the resulting graph model for route planning and execution. In zone control approaches, AGVs move from node to node on a permit basis, with limitations on the allowed number of AGVs at a time in each area of the graph to prevent and/or resolve deadlocks and conflicts. Hence, for an optimal implementation of traffic controllers in real manufacturing systems, it is essential to understand how the layout discretization influences the performance of the AGV network. This paper analyzes its effect in grid-like shaped workspaces by using a representative zone control algorithm and a recently developed improvement of it. Realistic numerical experiments on different layouts reveal that denser discretizations do not yield faster executions or increase in throughput, while lower control periods in the permit system entail significant performance uplifts.

Automated simulation modeling: ensuring resilience and flexibility in Industry 4.0 manufacturing systems

Industry 4.0 brings a revolution in the way to create and think of products. Consumer needs change rapidly and production systems must adapt quickly to market changes by changing processes, resources, and configurations. In this context, as real manufacturing systems are subject to constant change, simulations can easily become obsolete. This reduces the lifetime of the simulation model, causes repeated disruption in industrial planning support, and consequently continuous maintenance by experts. Within this framework, the novelty of the following research is the development of a flexible and resilient simulation model able to adapt according to its physical counterpart. Through advanced modeling, based on the object-oriented programming paradigm, an automated simulation model has been developed. This model adapts based on input data, ensuring accuracy and real-time planning support. The research is applied to a real case study of the production system of a world leader in the energy technology sector. The results show that the applied simulation tool can analyze different configurations and enhance production planning.

A Reinforcement Learning Driven Artificial Bee Colony Algorithm for Distributed Heterogeneous No-Wait Flowshop Scheduling Problem With Sequence-Dependent Setup Times

The distributed heterogeneous factory system is a typical scenario in the manufacturing industry. A distributed heterogeneous no-wait flowshop scheduling problem with sequence-dependent setup times (DHNWFSP-SDST) is studied in this paper. The differences in factory configuration and transportation time are considered in DHNWFSP-SDST. A mixed-integer linear programming (MILP) model is constructed and an artificial bee colony algorithm (ABC) with Q-learning (QABC) is proposed to address the DHNWFSP-SDST. Heuristic methods named NEH_H and DHHS are designed to construct potential initial candidates for the population. The neighborhood structures based on the job blocks are introduced in QABC to explore the solution space during the evolution processes. The Q-learning mechanism is employed to select neighborhood structures via empirical knowledge in the operation processes. The speed-up methods to accelerate the evaluation of the obtained neighborhood are designed to reduce the computation time of the QABC. The experimental results show that the QABC is a potential algorithm to address the DHNWFSP-SDST. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —Distributed manufacturing under the heterogeneous environment generally exists in real manufacturing systems. The scheduling problem refers to the reasonable arrangement of production orders to optimize certain indicators under limited time, resources, and computing costs. Distributed heterogeneous no-wait flowshop scheduling problem with setup times is an important industrial scheduling problem in distributed heterogeneous manufacturing systems. The problem takes into account the differences in factory configuration and transportation time in distributed regions. The problem is a kind of NP-hard problem as the solution space is huge. A reinforcement learning driven artificial bee colony algorithm is designed to address the problem. Heuristic methods are utilized to construct potential initial solutions. Neighborhood structures are designed to further optimize the initial solution. The effective empirical guidance is provided by Q-learning for the selection of the neighborhood structure of the algorithm to avoid invalid search. The experimental results show that the QABC obtains a high-quality scheduling scheme in a reasonable time.

Automated digital twin generation of manufacturing systems with complex material flows: graph model completion

Industry 4.0 determined the emergence of technologies that enable data-driven production planning and control approaches. A digital model can be used to make decisions based on the current state of a manufacturing system, and its efficacy strictly depends on the capability to correctly represent the physical counterpart at any time. Automated model generation techniques such as process mining can significantly accelerate the development of digital twins for manufacturing systems. However, complex production environments are characterized by the convergence of different material and information flows. The corresponding data logs present multiple part identifiers, resulting in the wrong finding of the system structure with traditional process mining techniques. This paper describes the problem of discovering manufacturing systems with complex material flows, such as assembly lines. An algorithm is proposed for the proper digital model generation, aided by the new concept of object-centric process mining. The proposed approach has been applied successfully to two test cases and a real manufacturing system. The results show the applicability of the proposed technique to realistic settings.

Learning human-process interaction in manual manufacturing job shops through indoor positioning systems

Nowadays, manufacturing systems are increasingly embracing the Industry 4.0 paradigm. Therefore, manual and low-standardized manufacturing environments are often digitized through Industrial Internet of Things technologies to quantitatively assess and investigate the role of the human factor from multiple points of view. This approach is commonly known as Operator 4.0. In such a scenario, this manuscript proposes an original digital architecture to monitor the efficiency and the social sustainability of labor-intensive manufacturing job shops. While the anonymous spatio-temporal trajectories of tagged workers are acquired through an ultrawide band radio network, machine learning algorithms autonomously detect the human-process interactions with strategic industrial entities upon developing industrial key performing indicators. The proposed architecture is tested and validated in a real manual manufacturing system. In detail, the performing accuracies of the machine learning-based software provide industrial plant supervisors with several production metrics to identify the hidden weaknesses and bottlenecks of the monitored manufacturing system. Such digital assessment may trigger a re-organization of the considered process to, for instance, enhance the allocation of the material in storage areas while fairly re-balancing the distances traveled by workers for picking activities.

가상 센서 기반 재공 위치 추정 방법 연구

This paper presents the estimation methodology of position of WIP(Work-In-Process). Since the monitoring digital twin synchronizes with a real manufacturing system, it is essential for the CPS (Cyber-Physical System) simulation or on-line simulation. A CPS simulator needs to start a simulation execution for any given time ‘t’, which requires the exact ‘snap-shot’ of the real manufacturing system at time ‘t’. The snapshot is the total state of the manufacturing system including all states of manufacturing facilities as well as all position of WIPs. This paper proposes the estimation methodology of position of WIPs which can provide the snapshot of a manufacturing system, and the proposed methodology is based on the concept of virtual sensors. If all WIPs in the system have virtual sensors attached, the movement of the WIP is defined according to other virtual sensor data and the mechanism of the facility. To verify the performance of these estimation methods, positions are compared for the same WIP in simulation and monitoring.

A novel multi-tasks chain scheduling algorithm based on capacity prediction to solve AGV dispatching problem in an intelligent manufacturing system

With the rapid development of automated guided vehicle (AGV), the AGV dispatching problem (AGVDP) becomes a major concern in the AGV system owing to its decisive impact on the productivity of the modern industry. However, based on the instantaneous assignment scheduling strategy to arrange tasks for AGVs, most AGVDP researches prove to be incomplete for taking no account of the remaining capacity of AGVs. Therefore, intended to optimize the AGVDP, this paper presents a multi-tasks chain scheduling algorithm (MTCSA) through the fusion of multi-tasks chain model and capacity prediction model based on support vector machine, which is verified by a real intelligent manufacturing system. The results reveal that the makespan and heavy load ratio aided by MTCSA can be reduced and improved by a maximum of 20.0% and 8.5% to the studied manufacturing system respectively, and thus further confirm the superior performance of the presented algorithm.

An efficient approach for solving a job shop scheduling problem with resources constraints: a case study iCIM 3000

In this work, we are interested in a job shop scheduling problem (JSSP) with resources availability constraints. The aim consists in scheduling a set of N jobs on M machines. To be processed in the system, each job needs an number of consumable resources that are available in a limited quantity. Solving such a problem means finding better jobs sequencing in order to minimise the maximum execution time. We suggest two different methods to solve the above-mentioned problem. We firstly propose a set of four heuristics based on priority rules. Then, we make call to genetic algorithm. Using a real job shop manufacturing system data, a large-scale experiment was performed in order to analyse the performance of the proposed methods. The studied system is called iCIM 3000. The simulation results reveal that the new proposed heuristics are better than genetic algorithms and achieve to good solutions in shorter time.

Credibility consideration for digital twins in manufacturing

Digital Twin has become an important technology for advanced manufacturing. However, to ensure that digital twins provide valuable decision support, the results generated by the digital twins must be trustworthy for real manufacturing systems. Model credibility assessment including Verification, Validation, and Uncertainty Quantification (VVUQ) techniques need to be applied throughout the life cycle of digital twins. Verification and Validation (V&V) activities are necessary to ensure that a digital twin meets its intended purpose and design goals used to establish its credibility. Uncertainty Quantification (UQ) produces a measure of performance that users can apply as part of a credibility assessment for a given digital twin. Credibility assessment of digital twins also includes factors beyond VVUQ. This paper discusses requirements of the digital twin credibility assessment, identifies potential uncertainty areas of digital twins, introduces a new digital-twin framework standard, proposes potential extension of the standard with credibility consideration, and discusses other ongoing relevant standards.

Nonlinear and Dotted Defect Detection with CNN for Multi-Vision-Based Mask Inspection.

This paper addresses the problem of nonlinear and dotted defect detection for multi-vision-based mask inspection systems in mask manufacturing lines. As the mask production amounts increased due to the spread of COVID-19 around the world, the mask inspection systems require more efficient defect detection algorithms. However, the traditional computer vision detection algorithms suffer from various types and very small sizes of the nonlinear and dotted defects on masks. This paper proposes a deep learning-based mask defect detection method, which includes a convolutional neural network (CNN) and efficient preprocessing. The proposed method was developed to be applied to real manufacturing systems, and thus all the training and inference processes were conducted with real data produced by real mask manufacturing systems. Experimental results show that the nonlinear and dotted defects were successfully detected by the proposed method, and its performance was higher than the previous method.

Privacy-preserving and sensor-fused framework for prognostic & health management in leased manufacturing system

Nowadays, increased investments and fierce competition have made the leased manufacturing paradigm being widely adopted. Advances in multi-sensor technologies and data-driven methodologies have also supported the original equipment manufacturer (OEM) to focus on outsourcing prognostic and health management (PHM) services for the leased manufacturing system. However, under this product-service paradigm, the degraded-related signal analysis for massive sensors of leased machines is always limited by privacy concerns, fusion lack, and varying correlation. To promote the security performance and analysis capability of PHM from the OEM to the lessee, this paper proposes a privacy-preserving and sensor-fused framework to achieve time-to-failure (TTF) updating and predictive maintenance (PdM) arrangement for the leased manufacturing system. We first aggregate degraded-related historical datasets through encrypted functional features to preserve the privacy signals of two parties. Next, informative sensors of each leased machine are identified by the penalized functional (log)-location-scale (LLS) regression, while keeping the signals private. Then, the real-time observed signals from the identified sensors on the lessee side are locally fused and cryptographically uploaded to the OEM for the adaptive functional LLS regression. The TTF distributions can thus be securely updated in real time and integrated to output PdM intervals for each leased machine. We further validate our privacy-preserving and sensor-fused framework on a real leased crankshaft manufacturing system, which shows our secure mechanism and fused prognostic can significantly ensure privacy protection, computational performance, fusion effectiveness, and PdM cost reduction.

Real-time data-driven dynamic scheduling for flexible job shop with insufficient transportation resources using hybrid deep Q network

With the extensive application of automated guided vehicles in manufacturing system, production scheduling considering limited transportation resources becomes a difficult problem. At the same time, the real manufacturing system is prone to various disturbance events, which increase the complexity and uncertainty of shop floor. To this end, this paper addresses the dynamic flexible job shop scheduling problem with insufficient transportation resources (DFJSP-ITR) to minimize the makespan and total energy consumption. As a sequential decision-making problem, DFJSP-ITR can be modeled as a Markov decision process where the agent should determine the scheduling object and allocation of resources at each decision point. So this paper adopts deep reinforcement learning to solve DFJSP-ITR. In this paper, the multiobjective optimization model of DFJSP-ITR is established. Then, in order to make agent learn to choose the appropriate rule based on the production state at each decision point, a hybrid deep Q network (HDQN) is developed for this problem, which combines deep Q network with three extensions. Moreover, the shop floor state model is established at first, and then the decision point, generic state features, genetic-programming-based action space and reward function are designed. Based on these contents, the training method using HDQN and the strategy for facing new job insertions and machine breakdowns are proposed. Finally, comprehensive experiments are conducted, and the results show that HDQN has superiority and generality compared with current optimization-based approaches, and can effectively deal with disturbance events and unseen situations through learning.

Effectiveness of virtual reality in discrete event simulation models for manufacturing systems

Virtual reality (VR) and discrete event simulation (DES) have become fundamental technologies of industry 4.0. There is a growing interest in such technologies in both academic and industry sectors. Nevertheless, few studies have reported the use of them simultaneously. This study develops a DES model and proposes an integrated approach combining self-reported and objective measurements to determine the VR effectiveness, based on data collection and analysis of 72 volunteers randomly assigned either a VR system or a laptop computer. The model represents a real manufacturing system where the user can move around, interact with objects, modify variables, run different scenarios, and identify the best one based on the system’s key performance indicators. Results showed that experimentation time is significantly less when users interact with the VR interface, and there is no significant difference between the two interfaces when comparing the decision-making time. However, the VR users obtained a higher percentage of correct answers. Finally, self-reported feedback indicated users preferred the VR system, in addition to the fact that discomfort and presence questionnaires exhibit typical values of the VR experience, and the usability questionnaire yields higher values than the laptop computer.

Job Shop Sequencing in Manufacturing Plants by Timed Coloured Petri Nets and Particle Swarm Optimization

The increasing diversity of products that can be machined in a plant poses scheduling challenges in modern manufacturing environments. The problem, which is known in literature as Flexible Job Shop Scheduling, is NP-Hard in general and can be efficiently solved by meta-heuristic approaches. This paper models the production plant in a Timed Coloured Petri Nets (TCPN) framework to describe production systems including flexible Computerized Numerical Control machines. Then, an algorithm is introduced to simulate the TCPN and compute the system throughput. Finally, the throughput is maximized by optimizing the job type sequencing and the amount of units of each job type that enters the system by implementing a Particle Swarm Optimization algorithm. The proposed simulation and optimization approaches are applied to a real manufacturing system producing ophthalmic lenses in order to show the effectiveness and the benefits of the proposed method.

Concept of hybrid modeled digital twins and its application for an energy management of manufacturing systems

Digital twins are used to digitally map properties and states of real manufacturing systems. Based on relevant data-driven and physics-based digital models, digital twins are employed to monitor, analyze and predict aspects of the corresponding manufacturing system. Therefore, digital twins are used on different manufacturing levels, e.g. the factory, machine, or process level. This paper presents a concept for digital twins that are modeled by physics-based and data-driven models. Furthermore, this concept is validated with the help of a use case that controls an energy management system of a model-scale factory by finding the optimal time for charging and using the battery.

A lab-scale manufacturing system environment to investigate data-driven production control approaches

Controlling production and release of material into a manufacturing system effectively can lower work-in-progress inventory and cycle time while ensuring the desired throughput. With the extensive data collected from manufacturing systems, developing an effective real-time control policy helps achieving this goal. Validating new control methods using the real manufacturing systems may not be possible before implementation. Similarly, using simulation models can result in overlooking critical aspects of the performance of a new control method. In order to overcome these shortcomings, using a lab-scale physical model of a given manufacturing system can be beneficial. We discuss the construction and the usage of a lab-scale physical model to investigate the implementation of a data-driven production control policy in a production/inventory system. As a data-driven production control policy, the marking-dependent threshold policy is used. This policy leverages the partial information gathered from the demand and production processes by using joint simulation and optimization to determine the optimal thresholds. We illustrate the construction of the lab-scale model by using LEGO Technic parts and controlling the model with the marking-dependent policy with the data collected from the system. By collecting data directly from the lab-scale production/inventory system, we show how and why the analytical modeling of the system can be erroneous in predicting the dynamics of the system and how it can be improved. These errors affect optimization of the system using these models adversely. In comparison, the data-driven method presented in this study is considerably less prone to be affected by the differences between the physical system and its analytical representation. These experiments show that using a lab-scale manufacturing system environment is very useful to investigate different data-driven control policies before their implementation and the marking-dependent threshold policy is an effective data-driven policy to optimize material flow in manufacturing systems.

A matrix analytic approach for Bayesian network modeling and inference of a manufacturing system

A shared digital twin of a manufacturing system is valuable to provide maintenance services in a collaborative manner. An accurate analytical model of the Bayesian network is crucial to depict endogenous failure mechanisms in the digital twin. Nodes’ connections in Bayesian networks correspond to a range of linear, bilinear or multilinear mappings over finite-state variables. The conditional probability table (CPT) of a child node can be represented as a k-dimensional tensor if it has (k-1) parent nodes. A new matrix analytic approach is proposed for Bayesian network inference based on the theory of semi-tensor product. The matrix representation of probabilistic networks is firstly studied, and a specific parameter training algorithm is constructed based on the matrix model. Bayesian network inference algorithms, including both forward and backward reasoning, are then presented for reliability analysis and fault diagnosis. A real manufacturing system is applied to verify the proposed approach. This matrix analytic approach helps to study the Bayesian network’s mathematical properties, and it is proved to be convenient and efficient in probability network modeling and inference.

A Five-Step Approach to Planning Data-Driven Digital Twins for Discrete Manufacturing Systems

A digital twin of a manufacturing system is a digital copy of the physical manufacturing system that consists of various digital models at multiple scales and levels. Digital twins that communicate with their physical counterparts throughout their lifecycle are the basis for data-driven factories. The problem with developing digital models that form the digital twin is that they operate with large amounts of heterogeneous data. Since the models represent simplifications of the physical world, managing the heterogeneous data and linking the data with the digital twin represent a challenge. The paper proposes a five-step approach to planning data-driven digital twins of manufacturing systems and their processes. The approach guides the user from breaking down the system and the underlying building blocks of the processes into four groups. The development of a digital model includes predefined necessary parameters that allow a digital model connecting with a real manufacturing system. The connection enables the control of the real manufacturing system and allows the creation of the digital twin. Presentation and visualization of a system functioning based on the digital twin for different participants is presented in the last step. The suitability of the approach for the industrial environment is illustrated using the case study of planning the digital twin for material logistics of the manufacturing system.

Robust Engineering for the Design of Resilient Manufacturing Systems

As the industrial requirements change rapidly due to the drastic evolution of technology, the necessity of quickly investigating potential system alternatives towards a more efficient manufacturing system design arises more intensely than ever. Production system simulation has proven to be a powerful tool for designing and evaluating a manufacturing system due to its low cost, quick analysis, low risk and meaningful insight that it may provide, improving the understanding of the influence of each component. In this research work, the design and evaluation of a real manufacturing system using Discrete Event Simulation (DES), based on real data obtained from the copper industry is presented. The current production system is modelled, and the real production data are analyzed and connected. The impact identification of the individual parameters on the response of the system is accomplished towards the selection of the proper configurations for near-optimum outcome. Further to that, different simulation scenarios based on the Design of Experiments (DOE) are studied towards the optimization of the production, under predefined product analogies.