Title: "Soft Computing in Prognostics and Health Management (PHM) Applications: a Case Study in Anomaly Detection."

Abstract
Soft Computing (SC) is a term that has evolved ,since its inception in 1991, to represent a methodology and a set of techniques covering the aspects of data-driven models design, domain knowledge integration, model generation, and model tuning. We distinguish between offline Meta-heuristics (MH’s), used for model design and tuning, and online MH’s, used for models selection or aggregation. This view suggests the use of hybrid SC at each MH’s level as well as at the object level. We manage model complexity by finding the best model architecture to support problem decomposition, generate local models with high-performance in focused applicability regions, provide smooth interpolations among local models, and increase robustness to imperfect data by aggregating diverse models. We illustrate this concept with a case study in anomaly detection for a fleet of physical assets (such as an aircraft engines or a gas turbines.) Anomaly detection typically uses unsupervised learning techniques to extract the underlying structural information from the data, define normal structures and regions, and identify departures from such regions. We focus on one of the most common causes for anomalies: the inadequate accuracy of the anomaly detection models, which are prone to create false alarms. To address this issue, we propose a hybrid approach based on a fuzzy supervisory system and an ensemble of locally trained auto associative neural networks (AANN’s.) The design and tuning of this hierarchical model is performed using evolutionary algorithms. In our approach we interpolate among the outputs of the local models (AANN’s) to assure smoothness in operating regime transition and provide continuous condition monitoring to the system. Experiments on simulated data from a high bypass, turbofan aircraft engine model demonstrated promising results.



Piero Bonissone A Chief Scientist at GE Global Research, Dr. Bonissone has been a pioneer in the field of fuzzy logic, AI, soft computing, and approximate reasoning systems applications since 1979. He is a Coolidge Fellow at GE Global Research (1993). He is also a Fellow of the Association for the Advancement of Artificial Intelligence (1996), the IEEE (2004), and the International Fuzzy Systems Association (2005).  He served as Editor in Chief of the International Journal of Approximate Reasoning for 13 years (1993-2005).  He co-edited six books and co-authored 150 publications. He received 54 patents issued from the USPTO (plus 50 pending). He has co-chaired 12 scientific conferences and symposia focused on Multi-Criteria Decision-Making, Fuzzy sets, Diagnostics, Prognostics, and Uncertainty Management in AI. In 2002, he was President of the IEEE Neural Networks Society (now Computational Intelligence Society). He has been an Executive Committee member of NNC/NNS/CIS society since the past 16 years and an IEEE CIS Distinguished Lecturer since 2004.  In 2008 he received the II Cajastur International Prize for Soft Computing, from CajAstur and the Foundation for the Advancement of Soft Computing.  He is the Chair of the Scientific Committee of the European Centre of Soft Computing.