Real Assembly Line Research Articles

An exact constraint programming method for the multi-manned assembly line balancing problem with assignment restrictions

This paper proposes an exact Constraint Programming (CP) method with an extensive focus on real-world constraints for the Multi-manned Assembly Line Balancing Problem with Assignment Restrictions (MALBP-AR). We perform an in-depth literature review to gather examples from real assembly lines and organize the AR regarding tasks, stations, workers, and mounting positions. Our study classifies the AR related to transformed resources and provides a general and unified model. We explore the concept of variable workplaces to dynamically assign workers to mounting positions and aggregate no overlap restrictions to avoid interference between workers. The classic MALBP model is extended by gradually incrementing the number of restrictions. The model variant found in the literature is herein called Partial MALBP-AR. Compared to the previous state-of-the-art Tabu Search Algorithm (TSA) for this problem, the Partial MALBP-AR found twelve additional optimality proofs. Besides the relevant results regarding solution quality, the CP method also has a satisfactory CPU performance. We also propose an entirely new set of AR and test these practical conditions with the so-called Extended MALBP-AR. Such an extended model, which covers all the AR presented here, reached optimality within the computational time limit for 36 out of 38 instances. The worst-case gap for an open instance is 8.20%. The results show a trade-off between the number of deemed restrictions and the computational performance. However, considering the detailed set of AR, we can obtain more representative solutions regarding the final balancing implementation compared to theoretical cases. The method can be used to design experiments, turning certain constraints on and off and allowing managers to evaluate different resource allocation scenarios.

A Robust Optimization Approach for the Resource Investment Problem of an Aircraft Final Assembly Line

In an aircraft final-assembly line (AFAL), test processes account for over 20% of all processes. And the processing times of test processes are usually uncertain owing to the fact that possible test failure brings about additional rounds of tests, which has a considerable impact on the resource investment of the AFAL. This paper studies a robust resource investment problem to minimize the resource investment cost for all realized processing times within budget uncertainty sets. A robust optimization model is formulated to determine the resource investment strategy. The robust model is reformulated and then solved by an exact solution approach based on an enhanced column and constraint generation algorithm featuring two novel acceleration techniques. Computational results demonstrate the effectiveness of our approach in solving realistic problem. Sensitivity analyses provide valuable managerial insights to real production. <i>Note to Practitioners</i>&#x2014;The uncertainty of processing time has a considerable impact on the resource investment of the aircraft final-assembly line, since longer processing time incurs more resource investment than expected and consequently delay the whole assembly process. In this article, we present a robust model to deal with the uncertainty by finding an optimal robust resource investment strategy. The model that captures the characteristics of the real assembly line is computationally tractable and solved by an enhanced algorithm that is very efficient for practical application. The proposed approach allows the shop floor managers to make a satisfactory tradeoff between the total investment cost and the completion time, the total investment cost and the levels of robustness and uncertainty by adjusting the values of related parameters. In future, we will explore more effective algorithms for solving larger instances and consider other uncertainties like resource disruption.

SafeFac: Video-based smart safety monitoring for preventing industrial work accidents

This work presents SafeFac, an intelligent camera-based system for managing the safety of factory environments. In SafeFac a set of cameras installed on the assembly line are used to capture images of workers that approach the machinery under hazardous situations to alert system managers and halt the line if needed. Given a challenging set of practical application-level requirements such as multi-camera support and low response latency, SafeFac exploits a YOLOv3-based light-weight human object detection. To address the latency–accuracy tradeoff, SafeFac incorporates a set of algorithms as pre- and post-processing modules and a novel adaptive camera scheduling scheme.Our evaluation with a video dataset containing more that 113,000 frames from real assembly line activity shows that SafeFac achieves high precision (99.93%) and recall (96.44%), and SafeFac successfully satisfies such challenging requirements as a ready-for-deployment system for safe factory management.

Augmented Reality Authoring of Digital Twins: Design, Implementation and Evaluation in an Industry 4.0 Context

This paper deals with Digital Twins (DTs) for Industry 4.0 factories, and their implementation in the context of a reconfigurable factory. This context implies a modification of the layout of the workstations during production, and thus requires a live update of the digital twins according to these modifications. We needed this update done by the operators directly on the workstations using an AR authoring tool. A literature review helped us to determine the criteria that a tool should fulfill in order to achieve this goal. The most important criteria are that the tool should be suitable for use by operators not trained in AR, that the learning curve should be short, and that it should be usable in a reconfigurable factory context. We created a DT containing all the necessary factory data and 3D models of the workstation interaction zones of a real assembly line. We then developed a tool enabling operators to match the DTs with their physical twin (PT) in AR, as well as to update their position in case of a reconfiguration. The experimentation we carried out confirms our analysis and shows us that it is possible to deploy a DT in a factory quite simply if the positioning of the DTs is done by direct manipulation (the 3D objects are co-located with the operator’s hand) with the help of an AR display device.

Augmented reality on industrial assembly line: Impact on effectiveness and mental workload

Studies examining the potential of augmented reality (AR) to improve assembly tasks are often unrepresentative of real assembly line conditions and assess mental workload only through subjective measurements and leads to conflicting results. We proposed a study directly carried out in industrial settings, to compare the impact of AR-based instructions to computerized instructions, on assembly effectiveness (completion time and errors) and mental workload using objective (eye tracking), subjective (NASA-TLX) and behavioral measurements (dual task paradigm). According to our results, AR did not improve effectiveness (increased assembly times and no decrease in assembly errors). Two out of three measurements indicated that AR led to more mental workload for simple assembly workstation, but equated computer instructions for complex workstation. Our data also suggest that, AR users were less able to detect external events (danger, alert), which may play an important role in the occurrence of work accidents.

A practical and synchronized data acquisition network architecture for industrial robot predictive maintenance in manufacturing assembly lines

This manuscript presents a methodology and a practical implementation of a network architecture for industrial robot data acquisition and predictive maintenance . We propose a non-intrusive and scalable robot signal extraction architecture, easily applicable in real manufacturing assembly lines. The novelty of the paper lies in the fact that it is the first proposal of a network architecture which is specially designed to address the predictive maintenance of industrial robots in real production environments. All the infrastructure needed for the implementation of the architecture is comprised of traditional well-known industrial assets. We synchronize the data acquisition with the execution of robot routines using common Programmable Logic Controllers (PLC) to obtain comparable data batches. A network architecture that acquires comparable and structured data over time, is a crucial step to advance towards an effective predictive maintenance of these complex systems, in terms of effectively detecting time dependent degradation. We implement the architecture in a real automotive manufacturing assembly line and show the potential of the solution to detect robot joint failures in real world scenarios. The architecture is therefore specially interesting for industrial practitioners and maintenance personnel. Finally, we test the feasibility of using one-class novelty detection models for robot health status degradation assessment using data of a real robot failure. To the best of our knowledge, this is the first contribution that uses robot torque signals of a real production line failure to train one-class models. • A non-intrusive and scalable robot data acquisition architecture is proposed for real manufacturing assembly lines. • The network acquires synchronized comparable robot operational data over time. • The architecture is able to detect a significant deviation in a faulty ABB IRB 6400r robot data. • One-class novelty detection models detect deviations in real operational conditions.

Optimization of Industry 4.0 Implementation Selection Process towards Enhancement of a Manual Assembly Line

Last year’s developments are characterized by a dramatic drop in customer demand leading to stiff competition and more challenges that each enterprise needs to cope with in a globalized market. Production in low-mix/high-volume batches is replaced with low-volume/high-variety production, which demands excessive information flow throughout production facilities. To cope with the excessive information flow, this production paradigm requires the integration of new advanced technology within production that enables the transformation of production towards smart production, i.e., towards Industry 4.0. The procedure that helps the decision-makers to select the most appropriate I4.0 technology to integrate within the current assembly line considering the expected outcomes of KPIs are not significantly been the subject of the research in the literature. Therefore, this research proposes a conceptual procedure that focus on the current state of the individual assembly line and proposes the technology to implement. The proposed solution is aligned with the expected strategic goals of the company since procedure takes into consideration value from the end-user perspective, current production plans, scheduling, throughput, and other relevant manufacturing metrics. The validation of the method was conducted on a real assembly line. The results of the validation study emphasize the importance of the individual approach for each assembly line since the preferences of the user as well as his diversified needs and possibilities affect the optimal technology selection.

Multi-Manned Assembly Line Balancing: Workforce Synchronization for Big Data Sets through Simulated Annealing

The assembly of large and complex products such as cars, trucks, and white goods typically involves a huge amount of production resources such as workers, pieces of equipment, and layout areas. In this context, multi-manned workstations commonly characterize these assembly lines. The simultaneous operators’ activity in the same assembly station suggests considering compatibility/incompatibility between the different mounting positions, equipment sharing, and worker cooperation. The management of all these aspects significantly increases the balancing problem complexity due to the determination of the start/end times of each task. This paper proposes a new mixed-integer programming model to simultaneously optimize the line efficiency, the line length, and the workload smoothness. A customized procedure based on a simulated annealing algorithm is developed to effectively solve this problem. The aforementioned procedure is applied to the balancing of the real assembly line of European sports car manufacturers distinguished by 665 tasks and numerous synchronization constraints. The experimental results present remarkable performances obtained by the proposed procedure both in terms of solution quality and computation time. The proposed approach is the practical reference for efficient multi-manned assembly line design, task assignment, equipment allocation, and mounting position management in the considered industrial fields.

Mathematical modeling to standardize times in assembly processes: Application to four case studies

Purpose: This paper proposes model-based standard times estimates, using multiple linear regression, nonlinear optimization, and fuzzy systems in four real cases assembly lines. The work includes a description of the models and a comparison of their performance with values obtained using the conventional chronometer method. These models allow estimating standard times without reconducting field studies.Design/methodology/approach: For the development of the time study, the methodology applied by the International Labour Organization (ILO) was used as a baseline. This methodology is structured in three phases: selection of the case study, registration of the process by direct observation, and calculation/estimation of the standard time. The selected case studies belong to real assembly lines of motorcycles, television sets, printed circuit boards, and bicycles.Findings: In the motorcycle’s assembly case, the study allowed constructing seven linear regression models to estimate standard times for assembling the front parts, and seven linear regression models to predict standard times for the rear parts of the different motorcycle types. Compared to the classical chronometer method, the results obtained never exceeded 10%. Regarding the case studies of assembling TV sets and PCBs, the study considered the construction of nonlinear optimization models that allow making appropriate predictions of the standard times in their assembly lines. Finally, for the bicycle assembly line, a fuzzy logic model to represent the standard time was constructed and validated.Research limitations/implications: For reasons of confidentiality of information, this work omitted the names of companies, services, and models of manufactured products.Originality/value: The literature consulted does not refer to the representation of standard time on assembly lines using mathematical models. The construction of these models with empirical data from actual assembly lines was a valuable aid to the companies involved in supporting activity planning.

Reinforcement Learning Based Production Control of Semi-automated Manufacturing Systems

In an environment which is marked by an increasing speed of changes, industrial companies have to be able to quickly adapt to new market demands and innovative technologies. This leads to a need for continuous adaption of existing production systems and the optimization of their production control. To tackle this problem digitalization of production systems has become essential for new and existing systems. Digital twins based on simulations of real production systems allow the simplification of analysis processes and, thus, a better understanding of the systems, which leads to broad optimization possibilities. In parallel, machine learning methods can be integrated to process the numerical data and discover new production control strategies. In this work, these two methods are combined to derive a production control logic in a semi-automated production system based on the chaku-chaku principle. A reinforcement learning method is integrated into the digital twin to autonomously learn a superior production control logic for the distribution of tasks between the different workers on a production line.By analyzing the influence of different reward shaping and hyper-parameter optimization on the quality and stability of the results obtained, the use of a well-configured policy-based algorithm enables an efficient management of the workers and the deduction of an optimal production control logic for the production system. The algorithm manages to define a control logic that leads to an increase in productivity while having a stable task assignment so that a transfer to daily business is possible. The approach is validated in the digital twin of a real assembly line of an automotive supplier.The results obtained suggest a new approach to optimizing production control in production lines. Production control shall be centered directly on the workers’ routines and controlled by artificial intelligence infused with a global overview of the entire production system.

A Markov Chain model for the performance evaluation of manufacturing lines with general processing times

This paper presents a Markov Chain approximation to model stations in manufacturing lines with general distributed processing times. The proposed Markov Chain approximation enables the use of continuous flow models for the performance evaluation of serial lines with finite buffers and mixed manual – automated operations. Each station in the line can consist of a highly automated machine with deterministic processing times, or of a human operator performing manual operations with general distributed processing times. Stations with random processing times are modelled through a continuous time – discrete state Markov Chain characterized by an operational state with a deterministic processing time, and by an auxiliary down state used to stochastically dilate the overall completion time of a part on the station. The Markov Chain parameters are defined through moments fitting of the probability distribution of the processing time of the original station. The resulting Markov Chain represents the behavior of the station in isolation and is then used as input in the decomposition techniques, based on continuous flow models, for the performance evaluation of serial lines. The model has been applied in the analysis of the production performances of a real assembly line.

Real-Time CU-Net-Based Welding Quality Inspection Algorithm in Battery Production

In the production process of laser welding products, visual inspection is usually employed to recognize welding spot locations and diagnose their quality faults. However, commonly used algorithms fail to succeed in both reliability and computational efficiency, especially when applied to assembly line. In this article, a method based on deep learning algorithm and traditional computer vision (TCV) algorithm is proposed, which achieves quality inspection of laser welding spots in the process of battery production. First, compressed U-shape network (CU-net) is proposed to extract welding pads and welding spots. Then, a template-based method is proposed to confirm the validity of each welding spot. Finally, TCV heuristic algorithms are proposed to achieve three error detections, i.e., welding pad placed obliquely, electrode tab placed over highly, and welding spot welded through. Moreover, we build a Welding Spot Quality Inspection Dataset taken from real assembly line. Compared with other pipelines, including U-net, MaskRCNN, and PSPNet, CU-net shows a significant superiority in both processing speed and detection accuracy. The results of template-based method and TCV heuristic algorithms have shown high computational efficiency and ensured inspection accuracy. The inference time of the whole method is less than 100 ms with the implementation on NVIDIA 1060 and Intel i7-6700.

Application of Digitization Procedures of Production in Practice

Abstract The paper addresses application of digitazation of production processes, which is part of organizational digitization, also known as Industry 4.0. It deals with modernization and optimization of production systems by creating models in simulation software via digital engineering. For our analysis, 8 real assembly lines were used, with an average of 15 workplaces, which were integrated into one universal line by means of simulation. The aim of our research was to create a digital twin of the real assambly lien and to analyse effectiveness of the proposed modernization universal assembly line using the generated statistical data.

Multi-manned assembly line synchronization with compatible mounting positions, equipment sharing and workers cooperation

The assembly of large and complex products as cars typically involves a huge amount of production resources as workers, pieces of equipment and layout areas. Multi-manned workstations commonly characterize these assembly lines along with some additional features such as compatibility/incompatibility between the different mounting positions, equipment sharing and worker cooperation. The simultaneous management of all such aspects significantly increases the balancing problem complexity. This paper proposes a new mixed integer programming model for such multi-manned assembly line synchronization problem to simultaneously optimize the line efficiency, the line length and the workload smoothness. A customized procedure based on a simulated annealing algorithm is developed to effectively solve this problem. The aforementioned procedure is validated through an industrial case study of an Italian sports car manufacturer distinguished by 665 tasks and numerous synchronization constraints typical of real assembly lines.

An Analysis of the Assembly Line Modernization by Using Simulation Software

The article deals with the optimization and modernization of assembly systems by creating models in the simulation software. The creation of digital models is a current trend in enterprise digitization called Industry 4.0. The Tecnomatix Plant Simulation environment allows you to create a virtual model of a real assembly line with the input of its basic production parameters. To perform the analysis, 8 real assembly lines were used, with an average of 15 workplaces, which were integrated into one universal line by means of simulation. The aim was to analyse the effectiveness of the proposed modernization universal assembly line using the generated statistical data.

Throughput Rate Improvement in a Multiproduct Assembly Line Using Lean and Simulation Modeling and Analysis

It has been a real challenge to improve the throughput of multiproduct assembly lines due to the complexity of the multiproduct assembly processes. In this paper, we used pull simulation module to mimic a real engine assembly line where 114 different products are being assembled. Line balancing and controlling the work in process (WIP) were found to be the driving elements to improve the throughput rate. Developing effective design of experiments for the simulation modelling and analysis helped in validating the impact of the changes. Recommended solutions have helped the engine assembly line to increase its throughput rate by 14%.

Integer based formulation for the simple assembly line balancing problem with multiple identical tasks

Assembly lines, especially those with welding procedures, can present several tasks with the same properties. These tasks can be treated as tasks with replicas, simplifying the problem. A Mixed Integer Linear Programming model is presented for the Simple Assembly Line Balancing Problem with Multiple Identical Tasks (or Repeated Tasks). Integer variables were used to define the number of identical tasks performed in each station. Along with variable reduction rules, the compact formulation presents only a fraction of the variables of equivalent binary models when several repeated tasks are present. Three instances inspired in real assembly lines and adapted benchmark problems with repeated tasks are used to compare the formulations. Using a universal solver, the integer formulation outperformed the binary formulation for the vast majority of instances and achieved competitive results in relation to the efficient procedure SALOME-2 (a dedicated algorithm based on branch-and-bound for Simple Assembly Line Balancing Problem). Grouping identical tasks proved to simplify the problem, allowing the procedure to solve larger instances.

Traveling worker assembly line (re)balancing problem: Model, reduction techniques, and real case studies

The assembly line balancing problem arises from equally dividing the workload among all workstations. Several solution methods explore different variants of the problem, but no model includes all characteristics real assembly lines might contain. This paper presents a mixed integer linear programming model that solves the Traveling Worker Assembly Line Balancing Problem (TWALBP). In this problem, the tasks’ balancing along with the assignment of workers to one or more workstations is determined for a given layout. The assignment flexibility is solved with a traveling salesman problem formulation integrated in the balancing model. Adapted standard datasets and three real case scenarios are used as benchmark sets. These scenarios present particularities such as human and robotic workers, assignment restrictions, zoning constraints, automatic and common tasks. The model successfully determines the tasks’ assignments and the routing of every worker for a layout aware optimization of assembly lines. Better quality balancing solutions were achieved allowing workers to perform tasks at multiple stations, showing a trade-off between assignment flexibility and movement time.

ErgoALWABP: a multiple-rule based constructive randomized search algorithm for solving assembly line worker assignment and balancing problem under ergonomic risk factors

This paper proposes a new type of assembly line worker assignment and balancing problem (ALWABP) which considers ergonomic risks. ALWABP occurs when task times vary according to the assigned worker. Although the operation time of a task is assumed to be fixed in classical assembly lines, it depends on the operator who executes the task in ALWABP. In ALWABP literature, the primary and secondary objectives are minimizing cycle time and balancing workload among workstations smoothly, respectively. When smoothing workload, only task times are taken into consideration in the relevant literature. However, degree of difficulty of tasks is also very important. Two workers executing two different stations with the same station time are assumed to be equally loaded according to the traditional perception. In fact, even they have the same station time; their workloads are different because of the different tasks they execute, in real life assembly line configurations. In order to close this gap between the real life and the literature, this study introduces an ALWABP problem with considering ergonomic risk factors (ErgoALWABP). Due to the complex nature of the problem, optimum seeking methods are not capable of solving it. So, the proposed problem is tackled with the multiple-rule based constructive randomized search approach. Also, OCcupational Repetitive Action method is used for making ergonomic risk assessment. Performance of the proposed solution procedure is compared with the relevant literature on benchmark data. Experimental results show that ergonomic environment could be improved when ergonomic risk factors are taken into account.

Sub-assembly detection and line balancing using fuzzy goal programming approach

This study suggests a mathematical modelling approach to sub-assembly detection and assembly line balancing problems. The sub-assembly detection problem involves two issues as follows: 1 determine which components of products are assembled in sub-assembly area; 2 determine which components of the products are assembled directly in the line. In assembly line balancing problem tasks are assigned to the workstations. The main motivation of this study is to determine the best compromised assembly line structure by solving the sub-assembly detection and the line balancing problems jointly. A multi-objective mathematical model is developed to formulate the above mentioned problems in which two conflicting costs of sub-assembly generation and assembly line labour cost are involved. Since the applications in real life are imprecise, fuzzy goal programming is adapted to the developed model to determining the vague goal. Then the model is implemented in a real life automotive manufacturers assembly line balancing.