A rotary dryer reduces the humidity in particulate matter through direct contact with combustion gases from fossil or renewable fuels (biomass). The design and operation of a rotary dryer present significant challenges, mainly because it must be flexible enough to adapt to a multitude of operating modes. Under this context, we want to prove the benefits for the manufacturer of the industrial product of the development of a digital twin for the rotary dryer. The digital twin will support the design, production and operation phases of the rotary dryer. It will contribute to the faster and more reliable design, better integration of the equipment in the customers’ general process and provide an optimal configuration of the process.

This DT uses the detailed simulation of a process to determine the optimal control of an installation. This methodology, applied to continuous processes, we believe is innovative and applicable to multiple manufacturing operations. From the economic point of view, as indicated above, the proposed tool is positioned in markets that will experience considerable growth in the coming years. In addition, this digital twin has multiple types of users: from rotary dryer manufacturing companies to companies that use these dryers in their processes. This will facilitate the exploitation of the tool. Finally, this DT is built under the concept of a sustainable and digital transition since the original application of the digital twin will focus on a rotary dryer to produce biomass pellets, one of the expected impacts being the improvement of the energy efficiency in the process.

<p><a href="https://youtu.be/LGz12-6UDb8" target="_blank">https://youtu.be/LGz12-6UDb8</a><a href="https://youtu.be/LGz12-6UDb8" target="_blank"></a>&nbsp;</p><p><a href="https://digitbrain.eu/2nd-wave-of-digitbrain-experiments/dt4dryer/" target="_blank">https://digitbrain.eu/2nd-wave-of-digitbrain-experiments/dt4dryer/</a>&nbsp;</p><p><a href="https://www.itainnova.es" target="_blank">https://www.itainnova.es</a>&nbsp;<br></p>

Inymon,'s core business is the injection and decoration of small to medium-sized plastic parts. For every new part to put in production by INYMON, process set-up and optimization is a key phase to keep the company competitiveness in an extremely time and cost demanding market. Today, this adjustment is performed based on experience and through trial and error loops. The use of simulation tools to design the injection process of thermoplastic parts is common practice in the plastics processing industry. Basically, these tools available in commercial packages such as Moldflow, Moldex3D etc. are used in the mould design stage, with the main objective of anticipating manufacturing problems and finding an “a priori” solution analysing alternatives of part and mould design, thereby reducing development costs and development time. However, the use of these tools for process set-up and parameter optimization is much less common and based on a trial-and-error approach. When the mould is already built and the piece goes into production, it is no longer feasible to perform such simulations to make decisions about the adjustments to be made on the line in the event of any variation produced, due to the excessive time it would take to make them. This constitutes a significant loss of opportunity since the power provided by simulation tools reproducing the physics of processes stops being useful because of the impossibility of having results immediately, which is the necessary requirement in decision-making to control the process in the injection line.

The main impact for the injection moulding industry is that the process setup is guided by the knowledge provided by the simulation tools, which account for the physics of the process. That is, the knowledge that the physics-based simulation tools provide extends from just being used in the mould design stage, to the process set-up and the part production stages.

This approach complements other commonly used approaches purely based on data captured in the injection line and applying Machine Learning techniques. The main contribution of the present approach is that it provides insight to the process engineer from the very beginning when the mould is set up and there is still no real data available in the monitoring system. In addition, it also provides a layer of interpretability of what happens inside the mould, complementing usually black-box approaches based on neuronal networks.

To achieve this point, the developer previously has to complete a sequence of activities. First to establish the proper knowledge-based simulation model that represents the injection process steps (filling-packing-cooling-part release) and which outputs are estimations of the part quality indicators. These simulations will be exploited through a simulation DOE that will allow exploring the processing window. Reduced-order techniques will allow encapsulating these results in simple models that run on the fly and that will constitute a Digital Twin of the process, being operable through the user interface. Based on data collected directly in the injection line, the developer can also update/correct the Digital Twin for achieving higher accuracy.

<p><a href="https://youtu.be/gqpN4FkkxGg" target="_blank">https://youtu.be/gqpN4FkkxGg</a><a href="https://youtu.be/gqpN4FkkxGg" target="_blank"></a>&nbsp;</p><p><a href="https://digitbrain.eu/1st-wave-of-digitbrain-experiments/digital-brain-for-injection-moulding/" target="_blank">https://digitbrain.eu/1st-wave-of-digitbrain-experiments/digital-brain-for-injection-moulding/</a><a href="https://digitbrain.eu/1st-wave-of-digitbrain-experiments/digital-brain-for-injection-moulding/" target="_blank"></a>&nbsp;<a href="https://digitbrain.eu/1st-wave-of-digitbrain-experiments/digital-brain-for-injection-moulding/" target="_blank"></a></p><p><a href="https://www.sciencedirect.com/science/article/pii/S2590005622000352" target="_blank">https://www.sciencedirect.com/science/article/pii/S2590005622000352</a>&nbsp;<br></p><p><a href="https://github.com/caeliaITAINNOVA/" target="_blank">https://github.com/caeliaITAINNOVA/</a><a href="https://github.com/caeliaITAINNOVA/" target="_blank"></a>&nbsp;</p><p>https://www.itainnova.es&nbsp;<br></p>

Recent advances in sensor and edge data processing systems have created new opportunities for the creation of accurate digital representations (or digital twins) of physical processes (e.g., manufacturing processes). At the same time, the induced modeling tasks have grown in scale and computational needs, e.g., going from small scale control system technologies applied on a single machine or a single process step to modeling and supervising complete industrial process lines. Naturally, and due to the recent successes of deep-learning and neural networks, data-based modeling techniques and predictive data analytics have drawn significant attention as an alternative to physics-based modeling techniques. However, the implementation, integration and acceptance of data-based methods brings up several new challenges, namely a) integrating systems of different nature and modeling granularity, b) integrating expert/domain knowledge, c) handling unforeseen or new scenarios and operating conditions and d) the need for human-machine interaction. In this presentation, we will review some of the most prominent challenges in the implementation of data-based modeling techniques for predictive data analytics and optimization in process industry and discuss current practices for tackling them. The presentation of these challenges will be organized in three thematic areas, namely a) virtual sensors, b) industrial reactive optimization, and c) hierarchical predictive modeling and optimization.

The implementation of data-based predictive analytics and digital twins faces challenges with respect to a) domain generalization (i.e., the ability of the model to perform equally well in different scenarios and operating conditions), b) complexity in decision support and reaction speed, c) human-machine interaction and explainability. In this presentation, we will discuss these challenges in the context of real-world industrial use-cases and discuss some of the current practices for tackling them. The presentation of these challenges will be organized in three thematic areas, namely a) virtual sensors, b) industrial reactive optimization, and c) hierarchical predictive modeling and optimization.

Virtual Sensors

In recent years, sensor and edge processing technologies allow for almost real-time monitoring of complex industrial processes. However, there might still be phenomena and process details that cannot accurately be monitored and regulated without including a form of prediction or simulation of the process evolution. An example is demonstrated in Figure 1 of the metal bending process. A set of sensors can monitor the process, however the operator requires a simulation of the upcoming evolution of the process in order to proactively regulate the applied force and achieve the desired final angle. The problem becomes even more challenging when we consider the case of different types of material and different scenarios (e.g., machine tools and configurations). We will present a framework for combining physical-based modeling techniques with machine-learning techniques in order to improve the prediction accuracy of the process evolution. Furthermore, we will present a methodology for domain generalization into multiple scenarios.

Figure 1. Metal bending process and comparison between physics-based modeling and hybrid modeling.

Industrial Reactive Optimization

Industrial processes could be highly complex due to the increased variability in the input and configuration parameters as well as the influence of unpredictable exogenous factors. In certain cases operators need to continuously monitor the sensor data and react within a few seconds by re-configuring the process parameters. Although operators might be quite experienced in regulating the process, often support is required from sensors, predictive models and digital twins to allow for more informative and quick decisions. In this part of the presentation, we discuss challenges and best practices in establishing a data processing pipeline for providing near real-time suggestions to the operator.

Figure 2. Big Data Management and Monitoring of a Lime Calcination Process.

Hierarchical Predictive Modeling and Optimization

The traditional paradigm of a digital twin usually considers specific parts of the process or process steps. When there is a need for simulating, monitoring or predicting complex processes of possibly heterogeneous systems or multiple process steps, several new challenges emerge. Specifically, monitoring of the overall process by the operators becomes a great challenge given the potentially large amount of sensor data from different parts of the process, and the lack of better and more human-readable interpretation of the process operating conditions. In this part of the presentation, we present some techniques for providing a more human interpretable and understandable monitoring of the process through causal inference modeling techniques.

Figure 3. Generalized Hidden Markov Models for supervisory process analysis.

<p style="margin:0in;line-height:85%"> </p> <p style="margin:0in;line-height:85%"><b><span style="font-size:10.0pt; line-height:85%;font-family:Roboto;color:black;mso-themecolor:text1;background: white">Virtual Sensors</span></b></p> <p style="margin:0in;line-height:85%"><span style="font-size:10.0pt;line-height: 85%;font-family:Roboto;mso-fareast-font-family:&quot;Times New Roman&quot;;mso-fareast-theme-font: minor-fareast;mso-bidi-font-family:Arial;color:black;mso-themecolor:text1; letter-spacing:-.1pt;mso-font-kerning:12.0pt">&nbsp;</span></p> <p style="margin-top:0in;margin-right:0in;margin-bottom:0in;margin-left:.5in; text-indent:-.25in;line-height:85%;mso-list:l0 level1 lfo1"><span style="font-size:10.0pt;line-height:85%;font-family:Symbol;mso-fareast-font-family: Symbol;mso-bidi-font-family:Symbol;color:black;mso-themecolor:text1;letter-spacing: -.1pt;mso-font-kerning:12.0pt"><span style="mso-list:Ignore">·<span style="font:7.0pt &quot;Times New Roman&quot;">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; </span></span></span><span style="font-size:10.0pt;line-height:85%; font-family:Roboto;color:black;mso-themecolor:text1;mso-ligatures:none">T. Grubinger, G. Chasparis, and T. Natschläger, “</span><span style="font-size: 10.0pt;line-height:85%;font-family:Roboto;color:#2E74B5;mso-themecolor:accent5; mso-themeshade:191;mso-ligatures:none"><a href="https://www.digital-twin-atlas.com/knowledge-hub/article/8/10.1016/j.enbuild.2016.12.074"><span style="color:#2E74B5;mso-themecolor:accent5;mso-themeshade:191">Generalized online transfer learning for climate control in residential buildings</span></a></span><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;color:black; mso-themecolor:text1;mso-ligatures:none">,” <i>Energy and Buildings</i>, vol. 139, pp. 63–71, Mar. 2017.</span><span style="font-size:10.0pt;line-height: 85%;font-family:Roboto;mso-fareast-font-family:&quot;Times New Roman&quot;;mso-fareast-theme-font: minor-fareast;mso-bidi-font-family:Arial;color:black;mso-themecolor:text1; letter-spacing:-.1pt;mso-font-kerning:12.0pt"></span></p> <p style="margin-top:0in;margin-right:0in;margin-bottom:0in;margin-left:.5in; text-indent:-.25in;line-height:85%;mso-list:l0 level1 lfo1"><span style="font-size:10.0pt;line-height:85%;font-family:Symbol;mso-fareast-font-family: Symbol;mso-bidi-font-family:Symbol;color:black;mso-themecolor:text1;letter-spacing: -.1pt;mso-font-kerning:12.0pt"><span style="mso-list:Ignore">·<span style="font:7.0pt &quot;Times New Roman&quot;">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; </span></span></span><span style="font-size:10.0pt;line-height:85%; font-family:Roboto;mso-bidi-font-family:Arial;color:black;mso-themecolor:text1; letter-spacing:-.1pt">W. Zellinger, T. Grubinger, M. Zwick, "</span><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;mso-bidi-font-family: Arial;color:#2E74B5;mso-themecolor:accent5;mso-themeshade:191;letter-spacing: -.1pt"><a href="https://doi.org/10.1007/s10845-019-01499-4"><span style="color:#2E74B5;mso-themecolor:accent5;mso-themeshade:191">Multi-source transfer learning of time series in cyclical manufacturing</span></a></span><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;mso-bidi-font-family: Arial;color:black;mso-themecolor:text1;letter-spacing:-.1pt">," <i>J Intell Manuf </i>31, pp. 777-787, 2020.</span><span style="font-size:10.0pt; line-height:85%;font-family:Roboto;mso-fareast-font-family:&quot;Times New Roman&quot;; mso-fareast-theme-font:minor-fareast;mso-bidi-font-family:Arial;color:black; mso-themecolor:text1;letter-spacing:-.1pt;mso-font-kerning:12.0pt"></span></p> <p style="margin-top:0in;margin-right:0in;margin-bottom:0in;margin-left:.5in; text-indent:-.25in;line-height:85%;mso-list:l0 level1 lfo1"><span style="font-size:10.0pt;line-height:85%;font-family:Symbol;mso-fareast-font-family: Symbol;mso-bidi-font-family:Symbol;color:black;mso-themecolor:text1;letter-spacing: -.1pt;mso-font-kerning:12.0pt"><span style="mso-list:Ignore">·<span style="font:7.0pt &quot;Times New Roman&quot;">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; </span></span></span><span style="font-size:10.0pt;line-height:85%; font-family:Roboto;color:black;mso-themecolor:text1;mso-ligatures:none">S. Luftensteiner and M. Zwick, “</span><span style="font-size:10.0pt;line-height: 85%;font-family:Roboto;color:#2E74B5;mso-themecolor:accent5;mso-themeshade: 191;mso-ligatures:none"><a href="http://www.thinkmind.org/index.php?view=article&amp;articleid=dbkda_2021_1_60_50033"><span style="color:#2E74B5;mso-themecolor:accent5;mso-themeshade:191">A Framework for Improving Offline Learning Models with Online Data</span></a></span><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;color:black; mso-themecolor:text1;mso-ligatures:none">,” presented at the <i>DBKDA 2021, The Thirteenth International Conference on Advances in Databases, Knowledge, and Data Applications</i>, May 2021, pp. 32–37. </span><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;mso-fareast-font-family: &quot;Times New Roman&quot;;mso-fareast-theme-font:minor-fareast;mso-bidi-font-family: Arial;color:black;mso-themecolor:text1;letter-spacing:-.1pt;mso-font-kerning: 12.0pt"></span></p> <span style="font-size:10.0pt;line-height:85%; font-family:Roboto;color:black;mso-themecolor:text1;mso-ligatures:none"></span><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;mso-fareast-font-family: &quot;Times New Roman&quot;;mso-fareast-theme-font:minor-fareast;mso-bidi-font-family: Arial;color:black;mso-themecolor:text1;letter-spacing:-.1pt;mso-font-kerning: 12.0pt"></span> <p style="margin:0in;line-height:85%"><span style="font-size:10.0pt;line-height: 85%;font-family:Roboto;mso-bidi-font-family:Arial;color:black;mso-themecolor: text1"><br></span></p> <p style="margin:0in;line-height:85%"><b><span style="font-size:10.0pt; line-height:85%;font-family:Roboto;mso-bidi-font-family:Arial;color:black; mso-themecolor:text1">Industrial Reactive Optimization</span></b></p> <p style="margin:0in;line-height:85%"><span style="font-size:10.0pt;line-height: 85%;font-family:Roboto;mso-bidi-font-family:Arial;color:black;mso-themecolor: text1">&nbsp;</span></p> <p style="margin-top:0in;margin-right:0in;margin-bottom:0in;margin-left:.5in; text-indent:-.25in;line-height:85%;mso-list:l0 level1 lfo1"><span style="font-size:10.0pt;line-height:85%;font-family:Symbol;mso-fareast-font-family: Symbol;mso-bidi-font-family:Symbol;color:black;mso-themecolor:text1;letter-spacing: -.1pt;mso-font-kerning:12.0pt"><span style="mso-list:Ignore">·<span style="font:7.0pt &quot;Times New Roman&quot;">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; </span></span></span><span style="font-size:10.0pt;line-height:85%; font-family:Roboto;color:black;mso-themecolor:text1;background:white">S. Luftensteiner, M. Mayr, <strong><span style="font-family:Roboto;font-weight: normal">G. Chasparis</span></strong><span style="font-variant-ligatures: normal; font-variant-caps: normal;orphans: 2;text-align:start;widows: 2;-webkit-text-stroke-width: 0px; text-decoration-thickness: initial;text-decoration-style: initial;text-decoration-color: initial; float:none;word-spacing:0px">, M. Pichler, “</span></span><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;color:#0070C0"><a href="https://doi.org/10.3389/fceng.2021.665545" style="box-sizing: border-box; transition: color 0.1s ease-in-out 0s, background-color 0.1s ease-in-out 0s; font-variant-ligatures: normal;font-variant-caps: normal;orphans: 2;text-align: start;widows: 2;-webkit-text-stroke-width: 0px;word-spacing:0px"><span style="color:#0070C0;background:white">A Versatile Usable Big Data Infrastructure for Process Industry and Its Monitoring Applications</span></a></span><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;color:black; mso-themecolor:text1;background:white"><span style="font-variant-ligatures: normal; font-variant-caps: normal;orphans: 2;text-align:start;widows: 2;-webkit-text-stroke-width: 0px; text-decoration-thickness: initial;text-decoration-style: initial;text-decoration-color: initial; float:none;word-spacing:0px">,” <i>Frontiers of Chemical Engineering: Computational Methods for Chemical Engineering</i>, vol. 3, 2021.</span><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;mso-fareast-font-family: &quot;Times New Roman&quot;;mso-fareast-theme-font:minor-fareast;mso-bidi-font-family: Arial;color:black;mso-themecolor:text1;letter-spacing:-.1pt;mso-font-kerning: 12.0pt"></span></span></p> <p style="margin-top:0in;margin-right:0in;margin-bottom:0in;margin-left:.5in; text-indent:-.25in;line-height:85%;mso-list:l0 level1 lfo1"><span style="font-size:10.0pt;line-height:85%;font-family:Symbol;mso-fareast-font-family: Symbol;mso-bidi-font-family:Symbol;color:black;mso-themecolor:text1;letter-spacing: -.1pt;mso-font-kerning:12.0pt"><span style="mso-list:Ignore">·<span style="font:7.0pt &quot;Times New Roman&quot;">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; </span></span></span><span style="font-size:10.0pt;line-height:85%; font-family:Roboto;color:black;mso-themecolor:text1;background:white">A. Kychkin and G. Chasparis, “</span><span style="font-size:10.0pt; line-height:85%;font-family:Roboto;color:#0070C0"><a href="https://ieeexplore.ieee.org/document/10185790" style="box-sizing: border-box; transition: color 0.1s ease-in-out 0s, background-color 0.1s ease-in-out 0s; font-variant-ligatures: normal;font-variant-caps: normal;orphans: 2;text-align: start;widows: 2;-webkit-text-stroke-width: 0px;word-spacing:0px"><span style="color:#0070C0;background:white;text-decoration:none;text-underline:none">Automated Cross Channel Temperature Predictions for the PFR Lime Kiln Operating Support</span></a></span><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;color:black; mso-themecolor:text1;background:white"><span style="font-variant-ligatures: normal; font-variant-caps: normal;orphans: 2;text-align:start;widows: 2;-webkit-text-stroke-width: 0px; text-decoration-thickness: initial;text-decoration-style: initial;text-decoration-color: initial; float:none;word-spacing:0px">,” <i>31st Mediterranean Conference on Control and Automation</i>, 2023.</span><span style="font-size:10.0pt;line-height:85%; font-family:Roboto;mso-fareast-font-family:&quot;Times New Roman&quot;;mso-fareast-theme-font: minor-fareast;mso-bidi-font-family:Arial;color:black;mso-themecolor:text1; letter-spacing:-.1pt;mso-font-kerning:12.0pt"></span></span></p> <p style="margin:0in;line-height:85%"><span style="font-size:10.0pt;line-height: 85%;font-family:Roboto;mso-fareast-font-family:&quot;Times New Roman&quot;;mso-fareast-theme-font: minor-fareast;mso-bidi-font-family:Arial;color:black;mso-themecolor:text1; letter-spacing:-.1pt;mso-font-kerning:12.0pt">&nbsp;</span></p> <p style="margin:0in;line-height:85%"><b><span style="font-size:10.0pt; line-height:85%;font-family:Roboto;mso-fareast-font-family:&quot;Times New Roman&quot;; mso-fareast-theme-font:minor-fareast;mso-bidi-font-family:Arial;color:black; mso-themecolor:text1;letter-spacing:-.1pt;mso-font-kerning:12.0pt">Hierarchical Predictive Modeling and Optimization</span></b></p> <p style="margin:0in;line-height:85%"><span style="font-size:10.0pt;line-height: 85%;font-family:Roboto;mso-fareast-font-family:&quot;Times New Roman&quot;;mso-fareast-theme-font: minor-fareast;mso-bidi-font-family:Arial;color:black;mso-themecolor:text1; letter-spacing:-.1pt;mso-font-kerning:12.0pt">&nbsp;</span></p> <p style="margin-top:0in;margin-right:0in;margin-bottom:0in;margin-left:.5in; text-indent:-.25in;line-height:85%;mso-list:l1 level1 lfo2"><span style="font-size:10.0pt;line-height:85%;font-family:Symbol;mso-fareast-font-family: Symbol;mso-bidi-font-family:Symbol;color:black;mso-themecolor:text1;letter-spacing: -.1pt;mso-font-kerning:12.0pt"><span style="mso-list:Ignore">·<span style="font:7.0pt &quot;Times New Roman&quot;">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; </span></span></span><span style="font-size:10.0pt;line-height:85%; font-family:Roboto;color:black;mso-themecolor:text1;background:white">M. Mayr, S. Luftensteiner, and G. Chasparis, “</span><u><span style="font-size: 10.0pt;line-height:85%;font-family:Roboto;color:#0070C0"><a href="https://www.sciencedirect.com/science/article/pii/S1877050922003544" target="_blank" style="box-sizing: border-box;transition: color 0.1s ease-in-out 0s, background-color 0.1s ease-in-out 0s; font-variant-ligatures: normal;font-variant-caps: normal;orphans: 2;text-align: start;widows: 2;-webkit-text-stroke-width: 0px;word-spacing:0px"><span style="color:#0070C0;background:white">Abstracting Process Mining Event Logs from Process-State Data to Monitor Control-Flow of Industrial Processes</span></a></span></u><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;color:black; mso-themecolor:text1;background:white"><span style="font-variant-ligatures: normal; font-variant-caps: normal;orphans: 2;text-align:start;widows: 2;-webkit-text-stroke-width: 0px; text-decoration-thickness: initial;text-decoration-style: initial;text-decoration-color: initial; float:none;word-spacing:0px">,” <i>Procedia Computer Science</i>, vol 200, 2022, pp. 1442-1450.</span></span><span style="font-size:10.0pt;line-height: 85%;font-family:Roboto;mso-fareast-font-family:&quot;Times New Roman&quot;;mso-fareast-theme-font: minor-fareast;mso-bidi-font-family:Arial;color:black;mso-themecolor:text1; letter-spacing:-.1pt;mso-font-kerning:12.0pt"></span></p> <p style="margin-top:0in;margin-right:0in;margin-bottom:0in;margin-left:.5in; text-indent:-.25in;line-height:85%;mso-list:l1 level1 lfo2"><span style="font-size:10.0pt;line-height:85%;font-family:Symbol;mso-fareast-font-family: Symbol;mso-bidi-font-family:Symbol;color:black;mso-themecolor:text1;letter-spacing: -.1pt;mso-font-kerning:12.0pt"><span style="mso-list:Ignore">·<span style="font:7.0pt &quot;Times New Roman&quot;">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; </span></span></span><span style="font-size:10.0pt;line-height:85%; font-family:Roboto;color:black;mso-themecolor:text1;background:white">G. Chasparis, S. Luftensteiner, and M. Mayr, “</span><u><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;color:#0070C0"><a href="https://www.sciencedirect.com/science/article/pii/S1877050922003507" target="_blank" style="box-sizing: border-box;transition: color 0.1s ease-in-out 0s, background-color 0.1s ease-in-out 0s; font-variant-ligatures: normal;font-variant-caps: normal;orphans: 2;text-align: start;widows: 2;-webkit-text-stroke-width: 0px;word-spacing:0px"><span style="color:#0070C0;background:white">Generalized Input-Output Hidden-Markov-Models for Supervising Industrial Processes</span></a></span></u><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;color:black; mso-themecolor:text1;background:white"><span style="font-variant-ligatures: normal; font-variant-caps: normal;orphans: 2;text-align:start;widows: 2;-webkit-text-stroke-width: 0px; text-decoration-thickness: initial;text-decoration-style: initial;text-decoration-color: initial; float:none;word-spacing:0px">,” <i>Procedia Computer Science</i>, vol 200, 2022, pp. 1402-1411.</span></span><span style="font-size:10.0pt;line-height: 85%;font-family:Roboto;mso-fareast-font-family:&quot;Times New Roman&quot;;mso-fareast-theme-font: minor-fareast;mso-bidi-font-family:Arial;color:black;mso-themecolor:text1; letter-spacing:-.1pt;mso-font-kerning:12.0pt"></span></p> <p style="margin-top:0in;margin-right:0in;margin-bottom:0in;margin-left:.5in; text-indent:-.25in;line-height:85%;mso-list:l1 level1 lfo2"><span style="font-size:10.0pt;line-height:85%;font-family:Symbol;mso-fareast-font-family: Symbol;mso-bidi-font-family:Symbol;color:black;mso-themecolor:text1;letter-spacing: -.1pt;mso-font-kerning:12.0pt"><span style="mso-list:Ignore">·<span style="font:7.0pt &quot;Times New Roman&quot;">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; </span></span></span><span style="font-size:10.0pt;line-height:85%; font-family:Roboto;color:black;mso-themecolor:text1;background:white">A. Abbas, <strong><span style="font-family:Roboto;font-weight:normal">G. Chasparis</span></strong><span style="font-variant-ligatures: normal;font-variant-caps: normal;orphans: 2; text-align:start;widows: 2;-webkit-text-stroke-width: 0px;text-decoration-thickness: initial; text-decoration-style: initial;text-decoration-color: initial;float:none; word-spacing:0px">, J. Kelleher, “</span></span><u><span style="font-size:10.0pt; line-height:85%;font-family:Roboto;color:#0070C0"><a href="https://link.springer.com/chapter/10.1007/978-3-031-12670-3_12" style="box-sizing: border-box;transition: color 0.1s ease-in-out 0s, background-color 0.1s ease-in-out 0s; font-variant-ligatures: normal;font-variant-caps: normal;orphans: 2;text-align: start;widows: 2;-webkit-text-stroke-width: 0px;word-spacing:0px"><span style="color:#0070C0;background:white">Interpretable Input-Output Hidden Markov Model-Based Deep Reinforcement Learning for&nbsp;the&nbsp;Predictive Maintenance of&nbsp;Turbofan Engines</span></a></span></u><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;color:black; mso-themecolor:text1;background:white"><span style="font-variant-ligatures: normal; font-variant-caps: normal;orphans: 2;text-align:start;widows: 2;-webkit-text-stroke-width: 0px; text-decoration-thickness: initial;text-decoration-style: initial;text-decoration-color: initial; float:none;word-spacing:0px">,”&nbsp;</span><em style="box-sizing: border-box; font-variant-ligatures: normal;font-variant-caps: normal;orphans: 2;text-align: start;widows: 2;-webkit-text-stroke-width: 0px;text-decoration-thickness: initial; text-decoration-style: initial;text-decoration-color: initial;word-spacing: 0px"><span style="font-family:Roboto;border:none windowtext 1.0pt;mso-border-alt: none windowtext 0in;padding:0in">Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)</span></em><span style="font-variant-ligatures: normal; font-variant-caps: normal;orphans: 2;text-align:start;widows: 2;-webkit-text-stroke-width: 0px; text-decoration-thickness: initial;text-decoration-style: initial;text-decoration-color: initial; float:none;word-spacing:0px">, 2022, 13428 LNCS, pp. 133–148.</span></span><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;mso-fareast-font-family: &quot;Times New Roman&quot;;mso-fareast-theme-font:minor-fareast;mso-bidi-font-family: Arial;color:black;mso-themecolor:text1;letter-spacing:-.1pt;mso-font-kerning: 12.0pt"></span></p> <p style="margin-top:0in;margin-right:0in;margin-bottom:0in;margin-left:.5in; text-indent:-.25in;line-height:85%;mso-list:l1 level1 lfo2"><span style="font-size:10.0pt;line-height:85%;font-family:Symbol;mso-fareast-font-family: Symbol;mso-bidi-font-family:Symbol;color:black;mso-themecolor:text1;letter-spacing: -.1pt;mso-font-kerning:12.0pt"><span style="mso-list:Ignore">·<span style="font:7.0pt &quot;Times New Roman&quot;">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; </span></span></span><span style="font-size:10.0pt;line-height:85%; font-family:Roboto;color:black;mso-themecolor:text1;background:white">A. Abbas, <strong><span style="font-family:Roboto;font-weight:normal">G. Chasparis</span></strong><span style="font-variant-ligatures: normal;font-variant-caps: normal;orphans: 2; text-align:start;widows: 2;-webkit-text-stroke-width: 0px;text-decoration-thickness: initial; text-decoration-style: initial;text-decoration-color: initial;float:none; word-spacing:0px">, J. Kelleher, “</span><span style="font-size:10.0pt; line-height:85%;font-family:Roboto;color:#0070C0;background:white"><a href="https://doi.org/10.1016/j.datak.2023.102240"><span style="color:#0070C0">Hierarchical framework for interpretable and specialized deep reinforcement learning-based predictive maintenance</span></a></span><span style="font-size:10.0pt; line-height:85%;font-family:Roboto;color:black;mso-themecolor:text1;background: white">,” <i>Data &amp; Knowledge Engineering</i>, vol 149, 2024.</span><span style="font-size:10.0pt;line-height:85%;font-family:Roboto;mso-fareast-font-family: &quot;Times New Roman&quot;;mso-fareast-theme-font:minor-fareast;mso-bidi-font-family: Arial;color:black;mso-themecolor:text1;letter-spacing:-.1pt;mso-font-kerning: 12.0pt"></span></span></p> <p style="margin-top:0in;margin-right:0in;margin-bottom:0in;margin-left:.5in; text-indent:-.25in;line-height:85%;mso-list:l1 level1 lfo2"><span style="font-size:10.0pt;line-height:85%; font-family:Roboto;color:#444444;background:white"><span style="font-size:10.0pt; line-height:85%;font-family:Roboto;color:#444444;background:white"><br></span></span></p><p style="margin-top:0in;margin-right:0in;margin-bottom:0in;margin-left:.5in; text-indent:-.25in;line-height:85%;mso-list:l1 level1 lfo2"><span style="font-size:10.0pt;line-height:85%; font-family:Roboto;color:#444444;background:white"><span style="font-size:10.0pt; line-height:85%;font-family:Roboto;color:#444444;background:white"><br></span></span></p><p style="margin-top:0in;margin-right:0in;margin-bottom:0in;margin-left:.5in; text-indent:-.25in;line-height:85%;mso-list:l1 level1 lfo2"> </p><p></p>

Digital Twin of the Lighthouse Factory Industry Laser as a Service demonstrator for training and marketing purposes

In this digital twin realized with stage, the digital twin software of the company Side Effects based in Zurich, Switzerland, we developed a digital twin for marketing and training purposes. This production demonstrator in use at the Swiss Smart Factory weighs around 2 tons and, if moved, requires a long setup time. In order to move this installation to present it a digital twin of it was realized, while navigating trough the vStage software you are instructed and guided over the several components and their specification in order to discover the system. It is composed of a Trotec SP2000 CO2 Laser, a Fanuc CR35 collaborative robot, and several IoT devices.

<p><a href="https://www.youtube.com/watch?v=F7COvVZgMOI">(1452) Digital Twin of our Laser as a Service Demonstrator - YouTube</a></p><p><a href="https://www.sipbb.ch/en/forschung/swiss-smart-factory/leuchtturm-projekt-industrie-4-0/">Leuchtturm Projekt Industrie 4.0 – Switzerland Innovation Park Biel/Bienne (sipbb.ch)</a><br><br></p>

The first scenario is shown below, which visualizes if a connected flexible module is docked or not on each of the tree docking station. As visible in the illustration below it is visible that the digital twin is able to update itself to the status of the pilot. This helps the user to check the status of the pilot line configuration at any given time.

This digital twin that allows a stakeholder to check if the process is feasible or not. It also verifies the following if the pilot line is setup correctly for the SchaltAG employees and if the upcoming product is feasible on this setup or not. This (virtual) validation process is performed in the virtual world while the pilot line is producing another product at SchaltAG, thus, it support in ramping up the production capabilities of the company. In other words, it helps to reduce the down time of the pilot line (downtime drops significantly) because most of the preparation work will be done in virtual environment thanks to the Digital Twin. This enables the employees to be involved in other tasks while the setup takes place in the background and once the setup instructions are completed the user can go to the production line and work with the real machine.

Overall, during the setup creation, the gathering of data is quite fast, but not in real-time. In other words, the Digital Twin environment has around 0.5s delay with respect to movement of the real physical twin. Still, this process helps the end users to have a powerful and useful monitoring of the data from the module in a fast and reliable manner, but not in absolute real-time, which is fine for our requirements in these processes.

<p><a href="https://dimofac.eu/">Home - DIMOFAC</a><br></p>