Tutorials

One of the unique features of the PHM conferences is free technical tutorials on various topics in health management taught by industry experts. As educational events tutorials provide a comprehensive introduction to the state-of-the-art in the tutorial’s topic. Proposed tutorials address the interests of a varied audience: beginners, developers, designers, researchers, practitioners, and decision makers who wish to learn a given aspect of prognostic health management. Tutorials will focus both on theoretical aspects as well as industrial applications of prognostics. These tutorials reach a good balance between the topic coverage and its relevance to the community.

Tutorial Topics

    • Introduction to Integrated Vehicle Health Monitoring
      Presented by: Prof. Ian K. Jennions
    • Multivariate Post-processing of Big Data in Prognostics and Health Management: Theory and Use Cases (slides)
      Presented by: Dr. Gueorgui Mihaylov
    • Ensembles of Models for Prognostics and Health Management (slides)
      Presented by:Prof. Piero Baraldi

PHM Conference tutorials have been a popular event in the past and the PHM society is proud to continue this service to the community. Topics of interest of these tutorials span fundamentals of PHM (Diagnostics, Prognostics, Health Management, Uncertainty Management, etc.) as well as specialized topics such as Cost-Benefit analysis, Data-Mining, Electronics PHM, Bayesian Filtering for Prognosis, etc. For a more comprehensive list of past tutorials please look at the following links:

Past PHM Tutorials
[PHM 2009] [PHM 2010] [PHM 2011] [PHM 2012] [PHME 2012] [PHM 2013]

Tutorials Chair:
Giovanni Jacazio giovanni.jacazio@polito.it

Tutorial Details

Tutorial Title: Introduction to Integrated Vehicle Health Monitoring

Prof. Ian K. Jennions.
Abstract:Integrated Vehicle Health Management is a broad ranging set of capabilities aimed at using data from high valued assets to affect decisions as to their maintenance and operation for business benefit. While the idea for IVHM started in space many years ago, it was not until the last 10 years that the subject has received wider attention. A number of books have been written on the subject and an overall taxonomy defined. It is from this taxonomy and the experiences at the IVHM Centre at Cranfield University, UK, that material for this tutorial is drawn. Rather than address the fundamental technologies of diagnostics, prognostics, sensors, etc, which can be found elsewhere, the tutorial will focus on how these technologies are brought into the product. This involves the quite separate subject areas of: business case formulation (cost/benefit for deploying IVHM), IVHM design system (not the asset design system), architecture and standards (how to decide what’s best), and health management development (how does it fit together).
Presenter Bio: Ian’s career spans some 40 years, working mostly for a variety of gas turbine companies. He has a Mechanical Engineering degree and a PhD in CFD both from Imperial College, London. He has worked for Rolls-Royce (twice), General Electric and Alstom in a number of technical roles, gaining experience in aerodynamics, heat transfer, fluid systems, mechanical design, combustion, services and IVHM. He moved to Cranfield in July 2008 as Professor and Director of the newly formed IVHM Centre. The Centre is funded by a number of industrial companies, including Boeing, BAE Systems, Thales, Meggitt, MOD, DRS, Alstom Transport and Novartis. He has led the development and growth of the Centre, in research and education, since its inception.

Ian is on the editorial Board for the International Journal of Condition Monitoring, a Director of the PHM Society, vice-chair of the SAE IVHM Steering Group and contributing member of the HM-1 IVHM committee, a Chartered Engineer and a Fellow of IMechE, RAeS and ASME. He is the editor of five recent SAE books on IVHM and a co-author of the book: ‘No Fault Found – The Search for the Root Cause’.

 

Tutorial Title: Multivariate Post-processing of Big Data in Prognostics and Health Management: Theory and Use Cases (slides)

Dr. Gueorgui Mihaylov
Abstract: The tutorial aims to present relevant applications of multivariate statistical and manifold learning methods to the post-processing of massive records of multiple health and efficiency indicators. Application of these techniques will be illustrated by concrete examples based on in-service data on different levels of system organization (single aircraft, fleet of aircrafts, single production line inside an industrial plant, industrial plant and large multinational systems of industrial plants). In many cases mathematically sophisticated post-processing recovers relevant hidden information.

Emergent phenomena in complex systems are a broad and very important topic in modern science which is strongly related to interesting qualitative issues in big data processing. Special attention will be dedicated do modelling of emergent phenomena by means of modern manifold learning methods. The characteristic behavior of an industrial plant from the viewpoint of its efficiency (its “efficiency fingerprint”) is investigated and mathematically interpreted as an emergent phenomenon.

Presenter Bio: : Gueorgui Mihaylov got a Master’s degree in theoretical phyics in 2004 and then a PhD in mathematics in 2008 from the University of Milan with thesis in differential geometry. Since then he has been involved in several theoretical projects, but mostly in a series of applied research projects in different areas as prognostics and health management, spectroscopic analysis of chemical and biological samples, applications to chemiometrics, power absorbometry modelling, engineering and business data mining and data analysis etc. These projects have been developed in collaboration between the Polytechnic University of Turin and important companies such as AgustaWestland, New Tera Technology, Dylog Italia, Teklook and others.

His research interests regard applications of the theory of geometrical structures, emergent phenomena in complex adaptive systems, manifold learning and multivariate data processing.
Currently he holds a position of senior researcher within a research project developed jointly by the Polytechnic University of Turin and Tetra Pak Packaging Solutions.

 

Tutorial Title: Ensembles of Models for Prognostics and Health Management (slides)

Prof. Piero Baraldi
Abstract: The intent of the tutorial is to illustrate methods of Prognostics and Health Management (PHM) based on the use of ensembles of models. Under the framework of PHM, we consider the three tasks of fault detection, diagnostics and prognostics.
In general, PHM may rely on quite different information and data on the past, present and future behavior of the equipment: there may be situations in which sufficient statistical equipment failure data are available, others in which the equipment behavior is known in a way to allow building a model of the degradation process, and others with scarce failure data, but with available sensor data related to the equipment degradation and failure processes (this is typically the case of highly valued equipment which is rarely allowed to run to failure). Correspondingly, a variety of approaches have been developed, based on different sources of information and data, modeling and computational schemes, and data processing algorithms.
In this context, increasing interest is arising towards the use of ensembles of diagnostic and prognostic models for PHM. These ensembles build their outcome from a combination of the outcomes of a set of individual models. The individual models perform well and make errors in different regions of the parameters space; the errors are balanced out in the ensemble combination and as a result the performance of the ensemble is superior to that of the single best model of the ensemble. Furthermore, by exploiting the nature of the ensemble itself, it is possible to provide measures of confidence in the ensemble outcomes.
This tutorial illustrates the development and application of ensemble of models for fault detection, diagnosis and prognosis in industrial equipment. In the first part of the tutorial, we consider the full scale implementation of models for anomaly detection. Since in practice the number of signals to be monitored is often too large to be handled effectively by a single detection model, the problem will be tackled by resorting to an ensemble of models, each one handling an individual group of signals. Applications of the proposed ensemble approach to real case studies concerning the monitoring of equipments and sensors in electricity production plants will be shown.
The second part is devoted to fault diagnosis and we will show the capabilities of ensemble of models for the incremental learning of new information on the basis of datasets that consecutively become available: this allows diagnosing new faults in dynamic applications and estimating the degree of confidence in the fault diagnosis.
Finally, the last part of the tutorial is devoted to prognostics and we present ensembles of models for the prediction of the Remaining Useful Life (RUL) of degrading equipment. The performance of the ensemble approach, from the points of view of prediction accuracy and robustness, is shown in case studies regarding components of nuclear power plants and off-shore oil platforms.
Presenter Bio: Dr. Piero Baraldi is associate professor of Nuclear Engineering at the department of Energy at the Politecnico di Milano (Italy). He received his BS in nuclear engng., Politecnico di Milano, 2002 and PhD in nuclear engng., Politecnico di Milano, 2006. He is the current chairman of the European Safety and Reliability Association, ESRA, Technical Committee on Fault Diagnosis. He is functioning as Technical Committee Co-chair of the European Safety and Reliability Conference, ESREL 2014, and he has been the Technical Programme Chair of the 2013 Prognostics and System Health Management Conference (PHM-2013). He is associate editor of the international scientific journal: “Journal of Risk and Reliability” and he serves as referee of several international journals. His main research efforts are currently devoted to the development of methods and techniques (neural networks, deep learning techniques, ensemble system, kernel regression methods, clustering techniques, genetic algorithms) for system health monitoring, fault diagnosis, prognosis and maintenance optimization. He is also interested in methodologies for rationally handling the uncertainty and ambiguity in the information. He is co-author of more than 100 peer-reviewed papers on international journals and proceedings of international conferences and of 2 books.