From explainable AI to explainable simulation: using machine learning and XAI to understand system robustness. - In: ACM SIGSIM-PADS 2023, (2023), S. 96-106
Evaluating robustness is an important goal in simulation-based analysis. Robustness is achieved when the controllable factors of a system are adjusted in such a way that any possible variance in uncontrollable factors (noise) has minimal impact on the variance of the desired output. The optimization of system robustness using simulation is a dedicated and well-established research direction. However, once a simulation model is available, there is a lot of potential to learn more about the inherent relationships in the system, especially regarding its robustness. Data farming offers the possibility to explore large design spaces using smart experiment design, high performance computing, automated analysis, and interactive visualization. Sophisticated machine learning methods excel at recognizing and modelling the relation between large amounts of simulation input and output data. However, investigating and analyzing this modelled relationship can be very difficult, since most modern machine learning methods like neural networks or random forests are opaque black boxes. Explainable Artificial Intelligence (XAI) can help to peak into this black box, helping us to explore and learn about relations between simulation input and output. In this paper, we introduce a concept for using Data Farming, machine learning and XAI to investigate and understand system robustness of a given simulation model.
https://doi.org/10.1145/3573900.3591114
Dynamic Time Warping und Synthesedaten zur Validierung von Seq2Seq für die Simulation. - In: ASIM Workshop 2023, (2023), S. 133-142
Seq2Seq is a machine learning method that allows to translate sequences into other sequences. This method has been tried in hybrid simulation of machine tools. The method has been used to generate time series of energy consumption of jobs from the corresponding numerical control code that runs on a machine tool. Seq2Seq suffers from various problems. Firstly, the creation of training data is costly. Secondly, standard Seq2Seq metrics only allow for the evaluation of a prediction of one timestamp at a time, not an entire time series. Thirdly, training metrics are failing when vanilla data is used, as two identical numerical control codes can result in deviating time series. This causes confusion for the model in the training loop, as it is not clear which time series should be considered correct. Here we propose a holistic framework to all three problems, that contains synthetic data, additional metrics for time series and dynamic time warping.
https://doi.org/10.11128/arep.21.a2113
Hybridization of the Digital Twin - overcoming implementation challenges. - In: Proceedings of the 56th Annual Hawaii International Conference on System Sciences, (2023), S. 1438-1447
In the context of Industry 4.0 the concept of the Digital Twin has gained significant momentum in industry as well as academia. Researchers have hypothesized a great number of potential benefits of the concept's usage. However, few real-world implementations have been recorded. This paper addresses the most pressing challenges inhibiting the concept's industrial application. It describes the process of the concept's hybridization to achieve a practical implementation strategy: the Hybrid Digital Twin. Subsequently, a prototype is implemented using a presently operational real-world manufacturing system to substantiate the viability of the methodology. Finally, the benefits, remaining issues and future developments of the concept are discussed.
https://hdl.handle.net/10125/102810
On the role of HLA-based simulation in New Space. - In: 2022 Winter Simulation Conference (WSC), (2022), S. 430-440
This paper discusses High Level Architecture (HLA) based simulation in the context of the emergence of the private spaceflight industry called New Space. We postulate that distributed simulation plays a fundamental role in facilitating new opportunities of a cost efficient access to space. HLA defines a simulation system's architecture framework with a focus on reusability and interoperability. The article will therefore discuss the impact of its usage on the potential of affordable new aerospace systems developments. Future possibilities with an increased level of loose component coupling are presented.
https://doi.org/10.1109/WSC57314.2022.10015338
Optimization of the design of modular production systems. - In: 2022 Winter Simulation Conference (WSC), (2022), S. 1783-1793
The desire for more flexibility in manufacturing systems, especially when different products or many product variants are manufactured in one production system is leading to a move away from the manufacturing principle of classic line production to more flexible and workshop-oriented production systems, particularly in the automotive industry. One of the challenges in these so-called modular assembly or production systems is the system design, especially the allocation of activities to the individual production cells. One approach to improve this allocation is offered by simulation-based optimization. In this paper, a concept for simulation-based optimization of the design of modular production systems is presented and demonstrated by means of a small academic case study. Classical genetic algorithms and additionally the NSGA-II algorithm, which also allows multi-objective optimization, are used.
https://doi.org/10.1109/WSC57314.2022.10015350
Explainable AI for data farming output analysis: a use case for knowledge generation through black-box classifiers. - In: 2022 Winter Simulation Conference (WSC), (2022), S. 1152-1163
Data farming combines large-scale simulation experiments with high performance computing and sophisticated big data analysis methods. The portfolio of analysis methods for those large amounts of simulation data still yields potential to further development, and new methods emerge frequently. Among the most interesting are methods of explainable artificial intelligence (XAI). Those methods enable the use of black-box-classifiers for data farming output analysis, which has been shown in a previous paper. In this paper, we apply the concept for XAI-based data farming analysis on a complex, real world case study to investigate the suitability of such concept in a real world application, and we also elaborate on which black-box classifiers are actually the most suitable for large-scale simulation data that accumulates in a data farming project.
https://doi.org/10.1109/WSC57314.2022.10015304
Hochaufgelöste Energieprofile durch hybride Simulation. - In: ASIM SST 2022 Proceedings Langbeiträge, (2022), S. 243-251
The price of a commodity, as electricity, is determined on a commodity market. A market is efficient when the supply and demand in the market are at an equilibrium. Efficient markets run on information. Information can cause a spontaneous and instantaneous change within the supply and demand in a market. The market communicates this new equilibrium through the change of the price of a commodity. In the electricity market the supplier and consumer communicate through electrical load profiles. A load profile signals when and how much energy should be consumed within a certain time frame without causing a change in the price of electricity. Creating such load profiles is commonly done by the supplier of energy by means of standard load profiles. Here we propose a data-driven simulation-based method that allows for the consumer to create its own specific load profile, which potentially will bring down the cost of energy consumed.
https://dx.doi.org/10.11128/arep.20.a2004
Simulationsbasiertes Deep Reinforcement Learning für Modulare Produktionssysteme. - In: ASIM SST 2022 Proceedings Langbeiträge, (2022), S. 65-72
Modulare Produktionssysteme sollen die traditionelle Fließbandproduktion in der Automobilindustrie ablösen. Die Idee dabei ist, dass sich hochgradig individualisierte Produkte dynamisch und autonom durch ein System flexibler Arbeitsstationen bewegen können. Dieser Ansatz stellt hohe Anforderungen an die Planung und Organisation solcher Systeme. Da jedes Produkt seinen Weg durch das System frei und individuell bestimmen kann, kann die Implementierung von Regeln und Heuristiken, die die Flexibilität des Systems zur Leistungssteigerung ausnutzen, in diesem dynamischen Umfeld schwierig sein. Transportaufgaben werden in der Regel von fahrerlosen Transportsystemen (FTS) ausgeführt. Daher bietet die Integration von KI-basierten Steuerungslogiken eine vielversprechende Alternative zu manuell implementierten Entscheidungsregeln für den Betrieb der FTS. In diesem Beitrag wird ein Ansatz für den Einsatz von Reinforcement Learning (RL) in Kombination mit Simulation vorgestellt, um FTS in modularen Produktionssystemen zu steuern. Darüber hinaus werden Untersuchungen zu dessen Flexibilität und Skalierbarkeit durchgeführt.
https://dx.doi.org/10.11128/arep.20.a2007
Optimierung des Designs modularer Montagesysteme. - In: ASIM SST 2022 Proceedings Langbeiträge, (2022), S. 15-22
Der Wunsch nach mehr Flexibilität in Fertigungssystemen, insbesondere, wenn verschiedene Produkte bzw. viele Produktvarianten in einem Produktionssystem gefertigt werden, führt, besonders in der Automobilindustrie, zur Abkehr vom Fertigungsprinzip der klassischen Linienfertigungen hin zu eher flexiblen und werkstattorientierten Produktionssystemen. Eine der Herausforderungen in diesen so genannten modularen Montage- bzw. Produktionssystemen ist das Systemdesign, insbesondere die Zuordnung der Tätigkeiten auf die einzelnen Fertigungsinseln. Ein Ansatz, diese Zuordnung zu verbessern bietet die simulationsbasierte Optimierung. In diesem Beitrag wird ein Konzept zur simulationsbasierten Optimierung des Designs modularer Montagesysteme vorgestellt und anhand einer Fallstudie demonstriert. Zum Einsatz kommen hierbei genetische Algorithmen, speziell der NSGA-II-Algorithmus, welcher auch mehrkriterielle Optimierung ermöglicht.
https://dx.doi.org/10.11128/arep.20.a2006
Data Analytics Production Line Optimization Model (DAPLOM) - a systematic framework for process optimizations. - In: Production at the leading edge of technology, (2022), S. 412-420
In this paper, we present a new framework for process optimizations, the Data Analytics Production Line Optimization Model (DAPLOM). Due to increasing efforts in the digitalization of production systems, an extensive amount of production data is available for analytics. This data can be used for the optimization of production lines and the prediction of their performance (e.g. drift of parameters or component quality) in order to achieve economic and technical improvements. The demand for systematical usage of data-driven methods involving technologies like Data Analytics and Machine Learning and the combination of engineering approaches is growing continuously.DAPLOM guides the implementation process of IT supported problem-solving solutions in production environments. It combines classical process- with data-driven approaches. Specific focus lies on achieving a holistic perspective with a macro- as well as a microscopic view on the given conditions. Here the macroscopic view covers the general material flow, whereas microscopic view considers process details. Additionally, DAPLOM provides useful methods in a step-by-step procedure structured in seven phases. The framework is validated in an industrial use case of an automated wire bending process. Thus, the effectiveness of the framework is demonstrated and further development potentials are identified.