Suchergebnisse

Über neue Möglichkeiten der Vernetzung und Datenverarbeitung von Anlagen und -komponenten gewinnt die prädiktive Instandhaltung stetig an Bedeutung. Während sie insbesondere in der Luftfahrt sowie bei der Versorgungs- und Energietechnik schon seit Jahren zum Einsatz kommt, sind innerhalb der Produktion derzeit noch wenige Anwendungsfälle zu finden. Der Beitrag stellt die bisher ungenutzten Potentiale dar, indem branchenübergreifende Anwendungsfälle sowie deren Übertragbarkeit auf die Produktion und Voraussetzungen für eine erfolgreiche Einführung vorgestellt werden.
 
Due to new possibilities of connectivity and data processing of assets and components, predictive maintenance has gained a growing importance. While it has been used in aerospace, supply or energy technology for many years, there are still few applications to be found within production. This paper outlines unused potential in presenting a classification of existing predictive maintenance applications and their transferability to production as well as prerequisites for a successful implementation.

Ungeplante Maschinenausfälle führen zu Stillständen in der Produktion, die große Auswirkungen auf die Wertschöpfungskette eines Unternehmens haben. Der Einsatz von Predictive Maintenance (PdM) entlang dieser Kette erhöht die Maschinenverfügbarkeit und sichert einen reibungslosen Produktionsablauf. Im Rahmen verschiedener Forschungsprojekte am iwb werden Anwendungsfälle für PdM in der Fertigung, der Montage und der Produktionssteuerung betrachtet. Dieser Beitrag beleuchtet individuelle Herausforderungen, Lösungsansätze und Grenzen von PdM im produktionstechnischen Umfeld.
 
Unplanned machine downtimes can have a big impact on the value chain of a company. Predictive Maintenance (PdM) shows the potential to increase machine availability and secures smooth production processes. Different research projects at iwb examine applications of PdM within manufacturing and assembly, as well as the integration of such approaches in production planning. This article highlights individual challenges, solutions and limits of PdM within production.

Predictive Maintenance ist eines der bestimmenden Themen im Kontext von Industrie 4.0. Ein Blick in heutige Produktionsstätten zeigt jedoch, dass die Voraussetzungen zur Umsetzung dieser Technologie für den überwiegenden Teil der Industrieanwendungen bisher nicht gegeben sind. Dieser Beitrag beschäftigt sich mit den notwendigen Schritten – vom Geschäftsmodell zur Anwendung in der industriellen Praxis – am Beispiel einer Schutzabdeckung für Werkzeugmaschinen.
 
Predictive maintenance is one of the defining topics in the context of Industrie 4.0, but a look at todays production facilities shows that the conditions for implementing this technology are not yet in place for the majority of industrial applications. This article is concerned with the necessary steps – from business model to application in industrial practice – using the example of a protective cover for machine tools.

Zunehmende Anforderungen an die Anlagenzuverlässigkeit und -verfügbarkeit führen zu einer immer höheren Bedeutung von Predictive Maintenance Tools. Herkömmliche Fertigungsressourcen sind jedoch nicht mit einer ausreichenden Anzahl von Sensoren ausgestattet, um den Maschinenzustand zu überwachen und notwendige Instandhaltungsleistungen vorherzusagen. Dieser Beitrag stellt ein Vorgehen zur Integration zusätzlicher Sensoren in bestehende Fertigungsressourcen für Predictive-Maintenance-Werkzeuge vor.
 
Increasing requirements for machine reliability and availability lead to an increased importance of predictive maintenance tools. How-ever, conventional manufacturing resources are not equipped with a sufficient number of sensors to monitor the machine condition and predict necessary maintenance measures. This paper presents a procedure for integrating additional sensors into existing manufacturing resources for predictive maintenance tools.

The work presented in this thesis deals with the predictive maintenance of one or two-units systems that undergo a stochastic deterioration process. The value of the state of the system is supposed to be known periodically in time and can take an infinite number of values. We first study the optimization of parametric maintenance policies. The goal is to calculate the optimal value of one or more parameters of the policy that minimize the average cost of the maintenance. In two particular cases, for systems with one or two units under a negative exponential deterioration process, it is possible to find analytical expressions of the average expected cost. The comparison of the optimal cost of the parametric maintenance policy with the one of a "classical" preventive policy shows the advantages of the proposed method. However, this approach does not allow in the general case to end up to an analytical expression of the average cost. Moreover, there is no guarantee that the pre-defined parametric policy is optimal among all the possible maintenance policies. We propose then a general model of maintenance. This model, much more compréhensive than ones found in the literautre, is based on the use of dynamic programming. It allows to compute the optimal maintenance policy for systems with one or more units when different levels of maintenance are possible. Under some conditions, it is proved that the optimal maintenance policy is to perform a maintenance when the state of deterioration of the system becomes greater than a limit. This property allows an easy computation of the optimal maintenance cost ; Le travail présenté dans cette thèse concerne la maintenance prédictive de systèmes constitués de une ou deux unités subissant un processus stochastique de détérioration. L'état de ces systèmes est connu de manière discrète dans le temps et peut prendre une infinité de valeurs. Nous nous intéressons tout d'abord à l'optimisation de politiques de maintenance paramétriques. Il s'agit de déterminer la valeur du ou des ...

The work presented in this thesis deals with the predictive maintenance of one or two-units systems that undergo a stochastic deterioration process. The value of the state of the system is supposed to be known periodically in time and can take an infinite number of values. We first study the optimization of parametric maintenance policies. The goal is to calculate the optimal value of one or more parameters of the policy that minimize the average cost of the maintenance. In two particular cases, for systems with one or two units under a negative exponential deterioration process, it is possible to find analytical expressions of the average expected cost. The comparison of the optimal cost of the parametric maintenance policy with the one of a "classical" preventive policy shows the advantages of the proposed method. However, this approach does not allow in the general case to end up to an analytical expression of the average cost. Moreover, there is no guarantee that the pre-defined parametric policy is optimal among all the possible maintenance policies. We propose then a general model of maintenance. This model, much more compréhensive than ones found in the literautre, is based on the use of dynamic programming. It allows to compute the optimal maintenance policy for systems with one or more units when different levels of maintenance are possible. Under some conditions, it is proved that the optimal maintenance policy is to perform a maintenance when the state of deterioration of the system becomes greater than a limit. This property allows an easy computation of the optimal maintenance cost ; Le travail présenté dans cette thèse concerne la maintenance prédictive de systèmes constitués de une ou deux unités subissant un processus stochastique de détérioration. L'état de ces systèmes est connu de manière discrète dans le temps et peut prendre une infinité de valeurs. Nous nous intéressons tout d'abord à l'optimisation de politiques de maintenance paramétriques. Il s'agit de déterminer la valeur du ou des ...

International audience ; We present an optimization problem for the maintenance of a multi-componentsystem subject to random deteriorations or failures of its components, resulting in the evolutionof the overall system's state. The system can be required to perform missions with deterministicoccurrences and durations. Our long term objective is to set up an optimized maintenance policyto ensure the smooth running of missions while minimizing maintenance costs. The main ideaof this work is to propose a mathematical model for the evolution of the system by using theformalism of a Markov Decision Processes (MDP). Through Monte Carlo simulations, we comparethe performances of several reference policies. ; Nous présentons un problème d'optimisation pour la maintenance d'un système multi-composants sujet à des détériorations ou défaillances aléatoires de ses composants, en-traînant l'évolution de l'état général du système. Le système peut être requis pour effectuer des missions de fréquences et durées déterministes. Notre objectif à long terme est de mettre en place une politique de maintenance optimisée afin de garantir le bon déroulement des missions tout en minimisant les coûts de maintenance. L'idée principale de ce travail est de proposer un modèle mathématique pour l'évolution du système en utilisant le formalisme d'un Processus Marko-vien Décisionnel (MDP). Par simulations de Monte Carlo, nous comparons les performances de plusieurs politiques de référence.

International audience ; We present an optimization problem for the maintenance of a multi-componentsystem subject to random deteriorations or failures of its components, resulting in the evolutionof the overall system's state. The system can be required to perform missions with deterministicoccurrences and durations. Our long term objective is to set up an optimized maintenance policyto ensure the smooth running of missions while minimizing maintenance costs. The main ideaof this work is to propose a mathematical model for the evolution of the system by using theformalism of a Markov Decision Processes (MDP). Through Monte Carlo simulations, we comparethe performances of several reference policies. ; Nous présentons un problème d'optimisation pour la maintenance d'un système multi-composants sujet à des détériorations ou défaillances aléatoires de ses composants, en-traînant l'évolution de l'état général du système. Le système peut être requis pour effectuer des missions de fréquences et durées déterministes. Notre objectif à long terme est de mettre en place une politique de maintenance optimisée afin de garantir le bon déroulement des missions tout en minimisant les coûts de maintenance. L'idée principale de ce travail est de proposer un modèle mathématique pour l'évolution du système en utilisant le formalisme d'un Processus Marko-vien Décisionnel (MDP). Par simulations de Monte Carlo, nous comparons les performances de plusieurs politiques de référence.

Predictive maintenance (PdM) within asset management improves savings in operational cost, productivity, and safety management capabilities. While PdM can be administered using various methods, growing interest in Arti cial Intelligence (AI) has lead to current state of the art PdM relying on machine learning (ML) technology. Like other tools used in PdM for asset management, standards for applying ML technology for PdM are required. This work introduces a standard of practice in regards to usage of asset data to develop ML analytic tools for PdM. It provides a standard method for ensuring asset data is in a form conducive to ML algorithms, and ensuring retention of asset information necessary for optimum PdM during the data transform. In the ML domain, it has been proven through research initiatives that the data structure used to train and test ML algorithms has a great impact on their performance and accuracy. Using poorly trained models for estimation due to improper data usage, can leave some AI-based PdM tools vulnerable to high rates of inaccurate estimations. Thus, leading to value loss during an asset's life cycle. ; This research was supported by the Australian Government Research Training Program Scholarship, American Australian Association Education Fund Fellowship, National GEM Consortium, Sustain-Owner (Sustainable Design and Management of Industrial Assets through Total Value and Cost of Ownership), a project sponsored by the EU Framework Programme Horizon 2020, MSCA-RISE-2014: Marie Sk lodowska-Curie Actions Research and Innovation Staff Exchange (Rise) (grant agreement number 645733 Sustain-owner H2020-MSCA-RISE-2014)

Industry 4.0, allied with the growth and democratization of Artificial Intelligence (AI) and the advent of IoT, is paving the way for the complete digitization and automation of industrial processes. Maintenance is one of these processes, where the introduction of a predictive approach, as opposed to the traditional techniques, is expected to considerably improve the industry maintenance strategies with gains such as reduced downtime, improved equipment effectiveness, lower maintenance costs, increased return on assets, risk mitigation, and, ultimately, profitable growth. With predictive maintenance, dedicated sensors monitor the critical points of assets. The sensor data then feed into machine learning algorithms that can infer the asset health status and inform operators and decision-makers. With this in mind, in this paper, we present TIP4.0, a platform for predictive maintenance based on a modular software solution for edge computing gateways. TIP4.0 is built around Yocto, which makes it readily available and compliant with Commercial Off-the-Shelf (COTS) or proprietary hardware. TIP4.0 was conceived with an industry mindset with communication interfaces that allow it to serve sensor networks in the shop floor and modular software architecture that allows it to be easily adjusted to new deployment scenarios. To showcase its potential, the TIP4.0 platform was validated over COTS hardware, and we considered a public data-set for the simulation of predictive maintenance scenarios. We used a Convolution Neural Network (CNN) architecture, which provided competitive performance over the state-of-the-art approaches, while being approximately four-times and two-times faster than the uncompressed model inference on the Central Processing Unit (CPU) and Graphical Processing Unit, respectively. These results highlight the capabilities of distributed large-scale edge computing over industrial scenarios.