Data-driven prognostics using a combination of constrained K-means clustering, fuzzy modeling and LOF-based score

A data-driven prognostics approach for monitoring sensor data is proposed.It relies on a combination of constrained K-means clustering, fuzzy modeling and LOF-based score.Fully comprehensive yet accurate models are obtained and deployed in a CBM+ platform.The approach is tested on a real data set concerning a marine diesel engine.A very small percentage of real faults are present in data.Obtained precision, sensitivity and specificity are above 93% and Cohens kappa is 0.93. Today, failure modes characterization and early detection is a key issue in complex assets. This is due to the negative impact of corrective operations and the conservative strategies usually put in practice, focused on preventive maintenance. In this paper anomaly detection issue is addressed in new monitoring sensor data by characterizing and modeling operational behaviors. The learning framework is performed on the basis of a machine learning approach that combines constrained K-means clustering for outlier detection and fuzzy modeling of distances to normality. A final score is also calculated over time, considering the membership degree to resulting fuzzy sets and a local outlier factor. Proposed solution is deployed in a CBM+ platform for online monitoring of the assets. In order to show the validity of the approach, experiments have been conducted on real operational faults in an auxiliary marine diesel engine. Experimental results show a fully comprehensive yet accurate prognostics approach, improving detection capabilities and knowledge management. The performance achieved is quite high (precision, sensitivity and specificity above 93% and =0.93), even more so given that a very small percentage of real faults are present in data.

[1]  J. Moubray.  Reliability-Centered Maintenance , 1991 .

[2]  Plamen P. Angelov,et al.  Fully unsupervised fault detection and identification based on recursive density estimation and self-evolving cloud-based classifier , 2015, Neurocomputing.

[3]  Yousef Saad,et al.  Trace optimization and eigenproblems in dimension reduction methods , 2011, Numer. Linear Algebra Appl..

[4]  V. Sugumaran,et al.  Fault diagnosis of automobile hydraulic brake system using statistical features and support vector machines , 2015 .

[5]  J. Fleiss,et al.  Statistical methods for rates and proportions , 1973 .

[6]  Marcello Braglia,et al.  Data classification and MTBF prediction with a multivariate analysis approach , 2012, Reliab. Eng. Syst. Saf..

[7]  Farid Kadri,et al.  Seasonal ARMA-based SPC charts for anomaly detection: Application to emergency department systems , 2016, Neurocomputing.

[8]  A. Pescetto,et al.  The Condition Monitoring System for Optimal Maintenance – Possible Application on Offshore Vessels , 2013 .

[9]  C. James Li,et al.  Acoustic emission analysis for bearing condition monitoring , 1995 .

[10]  Zhenghua Zhou,et al.  A novel approach for fault diagnosis of induction motor with invariant character vectors , 2014, Inf. Sci..

[11]  Darrin E. Barber.  Shipboard condition based maintenance and integrated power system initiatives , 2011 .

[12]  Huijun Gao,et al.  Data-Driven Process Monitoring Based on Modified Orthogonal Projections to Latent Structures , 2016, IEEE Transactions on Control Systems Technology.

[13]  Ian Davidson,et al.  Constrained Clustering: Advances in Algorithms, Theory, and Applications , 2008 .

[14]  Richard Curran,et al.  Knowledge-Based Engineering Review: Conceptual Foundations and Research Issues , 2010, ISPE CE.

[15]  Ju H. Park,et al.  Differential feature based hierarchical PCA fault detection method for dynamic fault , 2016, Neurocomputing.

[16]  William C. Greene.  Evaluation of non-intrusive monitoring for condition based maintenance applications on US Navy propulsion plants , 2005 .

[17]  Claire Cardie,et al.  Proceedings of the Eighteenth International Conference on Machine Learning, 2001, p. 577–584. Constrained K-means Clustering with Background Knowledge , 2022 .

[18]  Wei Li,et al.  Fault diagnosis of rotating machinery with a novel statistical feature extraction and evaluation method , 2015 .

[19]  Peter W. Tse,et al.  Wavelet Analysis and Envelope Detection For Rolling Element Bearing Fault Diagnosis—Their Effectiveness and Flexibilities , 2001 .

[20]  Brian A. Weiss,et al.  A review of diagnostic and prognostic capabilities and best practices for manufacturing , 2019, J. Intell. Manuf..

[21]  Alberto Carrascal,et al.  Unsupervised Methods for Anomalies Detection through Intelligent Monitoring Systems , 2009, HAIS.

[22]  Shashi Shekhar,et al.  Clustering and Information Retrieval , 2011, Network Theory and Applications.

[23]  Marvin Rausand,et al.  Reliability Centred Maintenance , 2008 .

[24]  Anastassios N. Perakis,et al.  Statistical Methods for Planning Diesel Engine Overhauls in the U. S. Coast Guard , 2004 .

[25]  Ayhan Demiriz,et al.  Constrained K-Means Clustering , 2000 .

[26]  Charu C. Aggarwal,et al.  Proximity-Based Outlier Detection , 2013 .

[27]  Ah-Hwee Tan,et al.  On Quantitative Evaluation of Clustering Systems , 2003, Clustering and Information Retrieval.

[28]  Weili Wu,et al.  Clustering and Information Retrieval (Network Theory and Applications) , 2003 .

[29]  Csaba Legány,et al.  Cluster validity measurement techniques , 2006 .

[30]  Hans-Peter Kriegel,et al.  LOF: identifying density-based local outliers , 2000, SIGMOD 2000.

[31]  Jesús Alcalá-Fdez,et al.  jFuzzyLogic: a Java Library to Design Fuzzy Logic Controllers According to the Standard for Fuzzy Control Programming , 2013, Int. J. Comput. Intell. Syst..

[32]  Lotfi A. Zadeh,et al.  Fuzzy Sets , 1996, Inf. Control..

[33]  Davide Anguita,et al.  Machine learning for wear forecasting of naval assets for condition-based maintenance applications , 2015, 2015 International Conference on Electrical Systems for Aircraft, Railway, Ship Propulsion and Road Vehicles (ESARS).

[34]  Dustin Harvey,et al.  Characterization and Prognosis of Multirotor Failures , 2015 .