Search

In the past decade the complexity of production systems has grown exponentially Manufacturing companies are under the influence of social economic and political trends that urge them to accelerate product development timelines while simultaneously achieving greater flexibility and resource efficiency Amid these challenges manufacturing companies commonly grapple with the task of defining key value streams within their production processes This challenge is especially relevant for high variety low volume businesses where value streams may include a high number of products and components The identification of these key value streams offers a pathway to improve decision making shedding light on the primary sources of complexity that impede operational efficiency Central to this endeavour is a comprehensive grasp of production flows In the context of Industry 4 0 the expanded accessibility of data sourced from factory operations opens doors to novel prospects facilitating a more profound understanding of production flows Following the long standing principles of Production Flow Analysis which have played a pivotal role in defining value streams over the years we present an approach for discerning these value streams by harnessing production data through process mining and machine learning techniques To demonstrate the practical viability of our proposed approach we present a case study that exemplifies its application in real world scenarios

The convergence of information technology with production technology, coupled with the escalating intricacy of production processes and products, is leading to new demands on the education of future engineers in mechanical engineering and related disciplines. New job profiles are arising and require interdisciplinary cooperation in the fields of information technology and the Internet of Things, machine learning and domain knowledge in order to enable data-based decisions for monitoring and improving products and processes in industrial production. The core of competence building has a strong focus on applying theoretical knowledge and information in order to make it tangible and thus enable people to learn. The contribution shows the conception and implementation of transnationally connected cyber-physical brewing labs, which were set up as learning factories for students and industry partners at TU Dortmund University and the University of Technology Sydney. The focus is on Industry 4.0 technologies, such as shared data space, condition monitoring of machines and assets in the Internet of Things, and the application of machine learning for product and process optimization. This article discusses the derivation of competence profiles, roles and the development of targeted theoretical and practical learning modules. It provides an overview of the use in various formats at both sites. The evolution of the brewing labs in existing research activities is also discussed. Finally, an outlook on future activities is given.

In recent times, learning by demonstration has seen tremendous progress in robotic assembly operations. One of themost prominent trajectory-level taskmodels applied isDynamic Movement Primitives (DMP). However, it lacks the ability to tackle complex operations as often encountered in industrial assembly. Augmenting low-level models with a highlevel framework in which different movement segments are deliberately parameterised is considered promising for such scenarios. This paper investigates the combination of trajectory-level DMPs with Methods-Time Measurement (MTM). We demonstrate how theMTM-1 system is utilised to establish distinguished DMP models for five of its basic elements, paving the way to benefitting from the sophisticated MTM system. The evaluation of the framework is conducted on a generic pick and place operation. Compared to a one-model-fits-all DMP approach for the whole task, the proposed method shows the advantage of appropriate temporal scaling, accuracy levelling and force consideration at adequate times.

Besides the sequence itself also additional factors serving as moderator variables affect the value of scheduling objectives. For mixed-model assembly lines, especially number and heterogeneity of different products, their volume mix proportions, average workload of the jobs to process and the degree of grouping of identical jobs within the sequence play a major role. By means of a simulation study based on data from a real unpaced mixed-model assembly line in the automotive industry, this work analyzes the impact of these moderating variables on various scheduling objectives. The analyzed scheduling objectives encompass flow-related objectives like mean flow time, productivity-related objectives like makespan, customer-related objectives like mean earliness, the supplier-related objective part usage rate variation and the human-related objectives mean learning effect and mean deterioration effect per job. Simulation scenarios are defined that differ regarding number and heterogeneity of products from three homogeneous to seven more heterogeneous products. Within every simulation scenario the volume mix proportions of the products, and inherently also the average workload of jobs, are systematically varied. Every simulation scenario is analyzed for five sequence types differing in the degree of grouping of identical jobs. For almost all scheduling objectives, strong dependencies on the volume mix proportions can be perceived, particularly for mean flow time. Homogeneous volume mixes with a dominating product in the mix often lead to other objective values compared to heterogeneous volume mixes that allow using alternation effects between various products in a sequence. Concerning degree of grouping, while some scheduling objectives like part usage rate variation are always strongly affected by the degree of grouping for every volume mix, other objectives like throughput show strong dependence only for some mixes and makespan does not even show any tendency. Average

In highly automated manufacturing systems running 24/7, preventive maintenance activities need to be executed during production times. Flexible job shops with several identical machines generally bear the potential to compensate temporary machine unavailability times caused by preventive maintenance without a considerable increase of the makespan due to machine paralleling. However, this requires sophisticated scheduling of manufacturing jobs and the assignment of maintenance activities over the scheduling horizon. This work, therefore, introduces a mixed-integer program that models both job scheduling and maintenance activity assignment. A local search algorithm is developed to solve both problems in an integrated way. Numerical studies are carried out based on data from a real flexible job shop in the automotive industry with 78 machines and four products. The results show that joint job scheduling and maintenance activity assignment is required to obtain a makespan similar to the makespan without maintenance consideration. Non-sophisticated scheduling of maintenance windows can increase the makespan by more than 20 percentage points compared to sophisticated scheduling. Besides, we show that only a limited amount of maintenance activities can be compensated and that higher unavailability times will inevitably lead to detrimental effects on the makespan unless maintenance worker capacity is increased.

The ongoing digitization of online learning resources has led to a proliferation of collaboration platforms for specific areas of application and disciplines. Simultaneously, the need for collaboration in the field of data analysis and knowledge management increases, especially for manufacturing companies. In this paper, collaborative and competency-based requirements for applying industrial data analytics are derived into tangible specifications for implementing a collaboration platform. The currently absent requirements of Industrial Data Science (IDS) projects are determined and then translated into platform-specific functions. Using the practical implementation of an ongoing research project, the defined requirements are transformed into features and applied in an online platform. The validation in a system of dynamic value networks serves to confirm the practical suitability of the specified requirements. The innovation of the platform is its clear focus on IDS project practitioners, who are typically comprised of several different domains. They often use a variety of analytics tools and aim for long-term use of deployed solutions. A pilot implementation of the developed platform is available online and additionally serves to validate the platform suitability.

Sustainability is the current global challenge. This is reflected in the demand for healthy food and CO2 neutrality. These challenges can be met with the industrial cultivation of algae: Algae can be used as food supplements, nutraceuticals, pharmaceuticals, fuel, CO2 sinks, and obtain high relative yield density per area. Current limitations in their large-scale use exists, as scaling up from laboratory environments to pilot applications typically requires more than 5 years, because of highly complex interactions in the growth behavior: They are influenced by current and past environmental conditions. These interactions make current pilot applications inefficient due to insufficient control and monitoring techniques. This limitation can be countered: By using modern communication and evaluation technologies, a “smart” bioreactor can be developed, which evaluates algae growth in real-time, performs process adaptations and thus significantly accelerates algae growth and scale-up. Therefore, an algae bioreactor was established at the University of Technology Sydney. The subject of this paper is the study of algae growth using Long Short-Term Memory Neural Networks (LSTMs). In order to learn the behavior of algae in the shortest possible series of experiments, repetitive change intervals were run by systematically varying the environmental parameters. LSTMs were trained to model algae growth. Attention mechanism is used on variable and temporal direction for importance. The LSTM is compared to a Transformer and an ARIMA. Based on the trained models, the behavior of algae growth is interpreted.

Learning from demonstration LfD has the potential to greatly increase the applicability of robotic manipulators in modern industrial applications Recent progress in LfD methods have put more emphasis in learning robustness than in guiding the demonstration itself in order to improve robustness The latter is particularly important to consider when the target system reproducing the motion is structurally different to the demonstration system as some demonstrated motions may not be reproducible In light of this this paper introduces a new guided learning from demonstration paradigm where an interactive graphical user interface GUI guides the user during demonstration preventing them from demonstrating non reproducible motions The key aspect of our approach is determining the space of reproducible motions based on a motion planning framework which finds regions in the task space where trajectories are guaranteed to be of bounded length We evaluate our method on two different setups with a six degree of freedom DOF UR5 as the target system First our method is validated using a seven DOF Sawyer as the demonstration system Then an extensive user study is carried out where several participants are asked to demonstrate with and without guidance a mock weld task using a hand held tool tracked by a VICON system With guidance users were able to always carry out the task successfully in comparison to only 44 of the time without guidance

Forgetting effects occur whenever production is interrupted and workers lose previously acquired routine. Existing forgetting models primarily focused on the impacts of production breaks in single-product production environments. Consequently, forgetting models neglected times during which a different product type is produced despite being an important cause for forgetting in industrial mixed-model production. Therefore, this work extends existing forgetting models for application in mixed-model line scheduling models by considering times of intermittent production. These are modelled as interruptions in addition to production break interruptions. A routine loss factor is introduced that allows to partially consider times elapsed during which different products are processed. The higher the routine loss factor is, the higher the considered portion of the time elapsed due to intermittent production becomes and the higher becomes the amount of forgetting. Through means of simulation studies it is investigated how the routine loss factor affects the height of learning effects present and the line performance in terms of makespan and flow time. While sequence types with a higher degree of grouping are almost insensitive with respect to the height of the routine loss factor, sequences with no grouping are strongly affected. Depending on the height of the routine loss factor different sequence types are preferred. Hence, not modelling forgetting appropriately can lead to non-optimal schedules.

In recent years, there has been an increasing interest in using industrial data science (IDS) in manufacturing companies. Structured IDS projects proceed according to process models such as the cross industry standard process for data mining (CRISP-DM), knowledge discovery in databases (KDD), or the process chain of industrial data science. Because of the process Chain’s transdisciplinary procedure, the participation of different people in the analysis with different tasks and competencies is necessary. Therefore, a concept to define specific roles is required, since it provides unequivocal descriptions of the respective tasks. As no role concept for the process chain of IDS exists, this paper develops and presents a transdisciplinary role concept including the four essential roles: Data engineer, analyst, user, and project manager. These roles are described in terms of their characteristics to enable structured cooperation between them. In addition, this paper presents the AKKORD platform for learning and collaboration, which especially should make an important contribution for small and medium-sized enterprises (SME) to develop knowledge in the usage of the process chain of IDS. The platform provides the opportunity for the users to train the essential roles with their characteristics in the company through targeted competence development.

Parts can be supplied from warehouses to assembly lines via several production-order-independent and -dependent parts supply strategies. Order-dependent parts supply strategies sequencing, kitting and batch supply share enough similarities that allow joint modelling in picking order planning and execution, whereas line stocking, just-in-time, just-in-sequence and just-in-sequence kit supply require separate modelling. Joint modelling is the precondition for setting up a software system in industrial practice, covering multiple parts supply strategies efficiently. With similar input–output relations and process steps, joint algorithms can be used. This work presents insights from a software system for sequencing, kitting and batch supply implemented in the automotive industry. The main process steps modelled are order-dependent part requests determination, bundling of part requests to picking orders, scheduling and release, and picking and transportation execution. Given the prevalence of assembly lines, sharing knowledge about successful modelling of parts supply strategies is crucial both for practitioners and researchers.

The gap between production scheduling theory and practice still persists as theoretical scheduling models do not cover real-world characteristics such as the multidimensional nature of scheduling decisions appropriately. Moreover, the set of objectives used in classical scheduling does not fully capture multidimensional scheduling decisions. This work therefore reviews which further objectives besides the classical ones have been established so far, which objectives are currently discussed in the literature, and identifies gaps for additional objectives for unpaced flow shops with discrete products. The aim is to provide a clustering of scheduling objectives that covers the multidimensional nature of scheduling decisions holistically. The scheduling objectives are clustered with respect to the respective stakeholders of the supply chain: customer, manufacturer and supplier. Within the objectives of interest for the manufacturer, objectives are further clustered into productivity-, flow- and human-related objectives. The classification serves as basis for the choice of a meaningful subset of objectives in industrial practice.

AbstractOver the past decades, the world has seen a continuous increase of globalisation and interconnectedness – in part supported by advances in digital communication and production technologies. In the case of industrial production, this trend has led to global, integrated supply chains in order to provide the most competitive and innovative products utilising the most competitive market conditions. In Australia, due to its remote geographic location and socioeconomic conditions, such as high labour costs and negative economics of scale, this has resulted in a loss of domestic manufacturing capabilities. With recent changes in the geopolitical environment (trade wars, actual wars, Covid-19, climate crisis etc.) calls to produce local are becoming louder again. In this article, we therefore explore the potential of digital technologies to overcome Australia’s challenges in reshoring its manufacturing capabilities. Findings indicate that a highly skilled digital workforce is needed to leverage the country’s potential in world-leading niche manufacturing. The Associate Degree of Advanced Manufacturing, developed and delivered by the Centre for Advanced Manufacturing at the University of Technology Sydney (UTS), is presented as an example of how to upskill the manufacturing workforce.

Advance automation digitization and interconnectivity are driving rapid evolution in the manufacturing industry As a result smart manufacturing has become increasingly popular driven by the integration of deep reinforcement learning DRL and digital twins in agent training DRL has proven to be particularly effective in smart manufacturing allowing agents to respond dynamically to an ever changing environment with a wide range of manufacturing components which leads to a slight variation in the objectives This study proposes an efficient DRL approach to smart manufacturing tasks by breaking them into three simpler tasks reach push and pick and place The Truncated Quantile Critics TQC algorithm is employed as the learning algorithm for the agent which is trained using Hindsight Experience Replay HER to formulate virtual goals in a sparserewarded environment Our experiments on these environments demonstrate that the proposed method of TQC HER achieves the best results on all three of the smart manufacturing tasks

Although industrial robots are predetermined to overtake repetitive and physically demanding assembly tasks of human workforce while delivering constant product quality, their precision comes at high hardware costs. Efficient control strategies make lightweight robotic solutions more attractive to companies with limited resources. This is especially important for assembly tasks requiring high adaptability, sensitivity and precision such as through-hole electronics assembly (THT). In this paper, a cost-efficient system for lightweight robot THT assembly and methods for approach, search and insertion via visual and force control are presented and evaluated. Using this system we can deal with large initial deviations of electronic components in the robot gripper and successfully insert THT components with two pins into circuit board holes under tight sub-millimeter tolerances. The proposed control methods do not require training and the system can be easily reconfigured for different hardware or assembly requirements. We evaluate our system on six different electronic components with two pins such as capacitors and inductors.

Companies with Make to Order MTO manufacturing have always faced the conflict of meeting a large volume of individual customer orders on time while remaining as flexible as possible Unlike Make to Stock MTS manufacturing planning the production requirements for MTO is a challenging task since incoming orders may vary in time and quantity while also being subject to a number of variables In some cases the delivery time allocated by the customer can be less than the required Order Lead Time to fulfil the order Manufacturers can respond to this with either with approached from production management or from data science This paper presents a framework to leverage the benefit of both domains We conduct a literature review and present the results of an expert workshop We propose criteria for a suitable data enrichment and the application of Machine Learning ML methods in sales and demand forecasting In conclusion the proposed framework helps to equip manufacturing companies with a structured strategy for data management and to utilize the benefit of ML for sales and demand forecasting

The electrification of the powertrain is leading to profound changes in the automotive industry. However, manufacturing must continue to maintain high quality standards in the production of innovative product systems. Despite the widespread use of advanced methods and tools for preventive quality management, nonconformities-deviations from the defined target state or objective state-occur. The industry standard for managing nonconformities is problem solving with the 8D method. In the traditional, manual approach, the problem description in particular requires a great deal of effort, which means that analyses are often too perfunctory and take too long. This paper therefore presents a novel knowledge graph-based and interactive IT application for problem solving in 8D. For the technical implementation, we use OWL ontologies and SKOS taxonomies compliant with W3C standards. The proposed approach was evaluated at a Tier-l automotive supplier using a practical example in the area of electrified drive and mobility solutions.

Since bolted joints are ubiquitous in manufacturing their effective and reliable quality assurance is particularly important Most tightening processes rely on statistical methods to detect faulty screw connections already during assembly In this paper we address the detection of faulty tightening processes using a clustering based approach from the field of Unsupervised Machine Learning In particular we deploy the k Means algorithm on a real world dataset from the automotive industry The model uses Dynamic Time Warping to determine the similarity between the normal and abnormal tightening processes treating each one as an independent temporal sequence This approach offers three advantages compared to existing supervised methods 1 time series with different lengths can be utilized without extensive preprocessing steps 2 errors never seen before can be detected using the unsupervised approach and 3 extensive manual efforts to generate labels are no longer necessary To evaluate the approach it is applied in a scenario where actual class labels are available This allows evaluating the clustering results using traditional classification scores The approach manages to achieve an accuracy of up to 88 89 and a macro average F1 score of up to 63 65

In addition to economic design, the ergonomic optimization of work systems is gaining in importance. However, high personnel costs for the creation of movement-economic analyses are obstacles for industrial use. Markerless motion capturing offers potential for the effort-reduced creation of these analyses based on the process building block system MTM-Human Work Design. The paper describes how machine learning can be used for motion analysis using abstracted video data.

Product life cycles change, market developments and quantities are increasingly difficult to predict, as is the case in the production of charging stations. For these reasons, scalable assembly concepts with an adaptable degree of automation are becoming increasingly important. Currently, charging stations are still manufactured manually. With increasing quantities, however, manual production is no longer economical. New technologies such as lightweight robotics offer a great potential for making production more flexible in terms of quantity. At the same time, new challenges arise because these requirements must be taken into account from the very beginning of product development and process planning. Currently, there are no planning approaches and recommendations for action that take this into consideration. Therefore, the research project “Simultaneous product and process development of a charging station outlet module suitable for automation” (SUPPLy) develops an integrated, digital and simultaneous product and process development of a modular charging station suitable for automation. The aim of the project is to develop an assembly process which enables an economic production of charging stations in case of fluctuating sales figures. The focus is not only on changes in the production process but also on a product design that is suitable for automation. The paper presents the ideas on a conceptual level.

Industrial Engineering, through its role as design, planning and organizational body of the industrial production, has been crucial for the success of manufacturing companies for decades. The potential, expected over the course of Industry 4.0 and through the application of Data Analytic tools and methods, requires a coupling to established methods. This creates the necessity to extend the traditional job description of Industrial Engineering by new tools from the field of Data Analytics, namely Industrial Data Science. Originating from the historic pioneers of Industrial Engineering, it is evident that the basic principles will remain valuable. However, further development in view of the data analytic possibilities is already taking place. This paper reviews the origins of Industrial Engineering with reference to four pioneers, draws a connection to current day usage, and considers possibilities for future applications of Industrial Data Science.

The importance of cross-process multivariate data analysis for improving products and processes is continuously increasing. Artificial intelligence and machine learning offer new possibilities to represent complex cause-effect relationships in models and to use them for optimisation. For consistent and scalable usage, unified data structures and representations of products, processes and resources are required in order to be able to use larger data populations as well as deploy these models in different application contexts. The paper presents an approach of shared prediction models for recipe optimisation based on unified digital twins in the beverage industry. For this purpose, a central generic data model was created, which is the basis for unified digital twins and thus the integration of physical and digital entities, as well as the foundation for cross-process data analysis.

Manually defined control limits remain a common strategy for quality control in manufacturing due to their ease of deployment on the shop floor compared to more advanced data analysis approaches. Despite their continued importance, there is no systematic method of defining these control limits. However, sub-optimal control limits can lead to undetected faults or cause unnecessary interruption to production. This manuscript presents an algorithm that systematizes this manual process into an efficient search task. We conceptualized the search task as a sequence of sub-problems that are based on the conventional steps taken by process experts when defining control limits. This algorithm can be integrated into an expert tool for shop floor personnel to automate the definition of control limits in annotated time series data. We demonstrate the efficacy of the control limits found by our algorithm by comparing them to those manually defined by process experts in real-world process data from the automotive industry. Furthermore, we show that our algorithm generalizes to traditional time series classification problems and achieves state-of-the-art performance on selected benchmark datasets. Our work is the first effort in automating the otherwise manual definition of control limits for fault detection.

Purpose: Computer-aided production engineering simulation is a common approach in the search for improvements to real systems. They are used in various industrial sectors and are a basis for process improvement. Such production simulations have found limited use in the wool industry. This study aims to compare the performance of different woolshed layouts (curved vs linear). Design/methodology/approach: A discrete event simulation is constructed for both considered layouts in Siemens Tecnomatix Plant Simulation software. Data from an in-field observational visit to a working woolshed and industry gray literature is used to validate the simulation model. The two layouts are compared in their base configuration and with equipment and worker changes to evaluate the impacts on throughput. Findings: In the base configurations, the curved layout reduces total worker travel time which increases production by 11 fleeces per day over the linear layout. The addition of an extra skirting table in the curved layout further increases throughout by 30 fleeces per day. The addition of more wool handlers does not have as large of an impact indicating that processing limits occur due to equipment capacity and shearer speed. Research limitations/implications: The sample size of the collected field data was small; some data have been collected from literature and not directly measured. Processing time is assumed to be distributed uniformly as a conservative distribution form. The study’s purpose is to evaluate relative differences in two different layouts using consistent worker parameters. Practical implications: This verifies the proposed curved shed layout improves production and gives farmers the ability to compute the long-term economic impact. The results also highlight that other processing stages in the shed need adjustment for more system gains. Originality/value: As the first application of discrete event simulation to evaluate woolsheds operations this work shows throughput

In the context of Industry 4.0, manufacturing companies have been increasingly adopting digital technologies such as Internet of Things, data analytics and cyber-physical systems to seize opportunities for productivity improvements. At the same time, established manufacturing philosophies such as Group Technology have assisted companies in managing the complexity of production processes for decades. To support manufacturing management with more informed decision-making tools, the literature has been proposing new approaches that exploit the potential of digital technologies to enhance the effectiveness of traditional manufacturing techniques. This study focuses on Production Flow Analysis (PFA) as an established approach for Group Technology. Although the existing literature has been focusing on Artificial Intelligence (AI) based approaches to plan the change to Group Technology for decades, few studies rely on production data directly extracted from the factory floor. This is partly due to the fact that technologies such as sensors and data analytics have been increasingly adopted in recent years, and this has led to an increasing amount of data that can be exploited to develop models that can support decisions. In particular, in the context of Industry 4.0, process mining has gained increasing interest, as it provides a data-driven methodology to capture real production processes. The goal of this study is to explore how PFA has evolved in the last decade thanks to the adoption of digital technologies, and to investigate potential synergies between PFA and process mining. This study uses a structured literature review to map advances in industrial applications of PFA in relation to digital technologies, as well as process mining applications in manufacturing to present a future research agenda. This provides manufacturing managers with a structured overview of existing industrial applications and the digital technologies adopted to enhance decision-making tools.

Recycling of refrigerating appliances plays a major role in protecting the Earth's atmosphere from ozone depletion and emissions of greenhouse gases. The performance of refrigerator recycling plants in terms of material retention is the subject of strict environmental certifications, and is reviewed periodically through specialized audits. The continuous collection of refrigerator data required for the input-output analysis is still mostly manual, error-prone and not digitized. In this paper, we propose an automated data collection system for recycling plants in order to deduce expected material contents in individual end-of-life refrigerating appliances. The system utilizes laser-scanner measurements and optical data to extract attributes of individual refrigerators by applying transfer learning with pre-trained vision models. Based on recognized features, the system collects data required for the estimation of target values of contained material masses, especially foaming and cooling agents. The presented data collection system paves the way for continuous performance monitoring and efficient control of refrigerator recycling plants.

Complex product and production systems often result in high variability in the production flow, prohibiting the sustainable implementation of lean practices. In this paper the authors introduce a PDCA cycle to analyse and reduce variability in value streams. The value stream is divided into zones, which are then qualified as stable or unstable. Lean practices can be applied in stable zones, unstable zones remain expert-driven. Measures are introduced to reduce variability in unstable zones with the ultimate target, to turn them into stable zones, extending sustainable lean activities in the value stream, step by step. An IT system is developed to acquire, process and visualise the vast amount of data to provide structured information for experts and management for the reduction of variability in production.

Many real-world big data applications in domains like finance, telecommunication and manufacturing rely on the detection of exceedingly rare patterns in large time series data sets. In principle, machine learning models can be trained to detect and classify such patterns. However, these models often lack the necessary robustness for practical applications and do not generalize well in production. Additionally, their intransparent decision-making hampers systematic debugging and improvement. Time series shapelets are a popular data mining primitive that can be used to extract a shape-based feature representation from the data. However, existing algorithms do not adequately consider the robustness and redundancy of these features. While these drawbacks can be compensated if sufficient labeled data is available, this is not possible for highly imbalanced data. We propose alterations to the current state-of-the-art shapelet algorithm that consider the margin of separation and the multivariate dependencies between the extracted features. This results in more robust and diverse features, which in turn translates to higher classification accuracy. We compare our algorithm to the current state-of-the-art using a public benchmark data set. Additionally, we showcase its applicability to highly imbalanced data using a suitable data set from the manufacturing domain.

Accompanied with globalisation, companies are confronted with new competitors and price competitions. In order to meet these challenges manufactures strive to increase their efficiency throughout the value stream. Especially special machine manufactures assembling in fixed position assembly islands show a high non-value adding share, since their assembly and logistical processes are characterised by a low degree of standardisation and maturity. This leads to a high spread of throughput times and increases the risk of delivery delays. To improve this weakness, this paper presents an approach to increase the degree of standardisation and to lower and stabilise the throughput time. The new approach bases on heuristics for packing problems and applies them to the material supply. By that, the allocation of the supplied components is optimised, as a result of which search times and path lengths for walking and transporting are shortened and the value adding time share is increased.

Improving the recall of information retrieval systems for similarity search in time series databases is of great practical importance. In the manufacturing domain, these systems are used to query large databases of manufacturing process data that contain terabytes of time series data from millions of parts. This allows domain experts to identify parts that exhibit specific process faults. In practice, the search often amounts to an iterative query–response cycle in which users define new queries (time series patterns) based on results of previous queries. This is a well-documented phenomenon in information retrieval and not unique to the manufacturing domain. Indexing manufacturing databases to speed up the exploratory search is often not feasible as it may result in an unacceptable reduction in recall. In this paper, we present a novel adaptive search algorithm that refines the query based on relevance feedback provided by the user. Additionally, we propose a mechanism that allows the algorithm to self-adapt to new patterns without requiring any user input. As the search progresses, the algorithm constructs a library of time series patterns that are used to accurately find objects of the target class. Experimental validation of the algorithm on real-world manufacturing data shows, that the recall for the retrieval of fault patterns is considerably higher than that of other state-of-the-art adaptive search algorithms. Additionally, its application to publicly available benchmark data sets shows, that these results are transferable to other domains.

Capacity-limiting bottlenecks in manufacturing systems form the ideal starting point for measures of improvement. However, the inherent variability of modern systems leads to dynamic bottleneck behavior, causing them to shift between stations. Numerous methods for the detection of shifting bottlenecks exist in literature. In this paper, we present and compare three methods: Bottleneck Walk (BNW), Active Period Method (APM), and an adaptation of Interdeparture Time Variances (ITV). The comparative study deploys the methods in a serial production line with seven stations and eight buffers. We vary the individual locations of the bottlenecks by adding more process time. To compare the methods, we determine the overall average ratio of agreement between the three detection methods. APM and ITV have the highest agreement at an average of 80.10%. Pairings with BNW achieve significantly lower rates of agreement, with 56.33% for ITV, and 62.03%% when compared to the APM.

In manufacturing systems, early quality prediction enables the execution of corrective measures as early as possible in the production chain, avoiding thus costly rework and waste of resources. The increasing development of Smart Factory Sensors and Industrial Internet of Things has offered wide opportunities for applying data-driven approaches for early quality prediction in real-time using Machine Learning (ML). With the multiplication of applications, further requirements on the quality of predictive ML models covering multiple aspects such as accuracy, robustness and explainability have to be fulfilled to build trustworthy ML-based solutions. In this context, we investigate the task of early quality classification using a Convolutional Neural Network (CNN) on time series sensor data of an automotive real-world case study. To do so, a gradient-based heat-mapping explanation method for CNNs is computed to determine the most discriminative time series patterns and localize them in time. These patterns are subsequently used to achieve quality prediction in real-time as early as possible.

The linkage of machines in the context of Industry 4.0 through information and communication technology (ICT) to cyber-physical systems with the aim of monitoring, controlling, and optimizing complex production systems, enables real-time capable approaches for data acquisition, analysis, and process knowledge generation. In this context, surface mount technology (SMT) in electronics manufacturing is increasingly enhanced by digitalizing the process. This allows the collection and analysis of sensor data to predict the process quality in real-time. Process control interventions can then be derived in a timely manner based on quality predictions. To further support decision-making for process control by domain experts, explanations for the model-based quality predictions should be supplemented in addition. More specifically, we employ a 1D-convolutional neural network for quality prediction of well-defined Fields Of Views (FOVs) of Printed Circuit Boards (PCBs). Explanations for the model's predictions are provided under various perspectives using a heat-mapping-based technique to highlight the contribution of both local and global PCBs' characterizing features to the quality predictions. This helps to reveal the most decisive features for a given quality assignment and understand which process parts are the most responsible for such decision. Finally, the deployment of the model-based predictive and parts of the prescriptive analytics supported by the provided explanations, are achieved using Edge Cloud Computing technology.

Numerous methods for bottleneck detection, along novel approaches for bottleneck prediction, are available in literature. To facilitate the development and application of such methods, this paper proposes a holistic methodology for Bottleneck Analysis in dynamic value streams. Analogous to established data analytics levels, namely descriptive, diagnostic, predictive, and prescriptive analytics, the methodology specifies objectives for data-driven Bottleneck Analysis. Based on state-of-the-art bottleneck detection methods, the methodology provides measures for the diagnosis of bottleneck severity and frequency. Additionally, it considers prediction methods to anticipate emerging bottlenecks, depending on available databases. Finally, the methodology provides a context for the yet unexplored field of bottleneck prescription, which aims to mitigate bottleneck effects by data-driven control recommendations. Further practical application of the methodology has to confirm its suitability as a holistic framework for analyzing bottlenecks in dynamic value streams.

Since its early days as a rapid prototyping technology, additive manufacturing has significantly evolved and become an important enabling technology for advanced manufacturing. Despite the benefits, its application in industry is not trivial as, for example, products need to be re-designed and processes changed, and it is not always the optimal manufacturing technology. Design for Additive Manufacturing (DfAM) is a key approach to support the successful use of additive manufacturing in industry and to bridge the gap between research and practice. Aside from design process and technology related methods and tools, DfAM also considers organisational and procedural aspects. To support the success of DfAM and as a result additive manufacturing, it is important to understand how industry needs are addressed by current DfAM methods/tools and related research activities. In this respect, a comprehensive analysis is missing. Therefore, this paper systematically analyses current research topics, fields and trends as well as industry needs and DfAM requirements from an engineering management perspective. Mapping them allows for a systematic discussion between academia and industry to identify the most pressing research needs.