Two-stage classifications for improving time-to-failure estimates: a case study in prognostic of train wheels

In order to meet the need for higher equipment availability and lower maintenance cost, much attention is being paid to the development of prognostic systems. Such systems support a proactive maintenance strategy by continuously monitoring the components of interest and predicting their failures sufficiently in advance to avoid disruptions during operation. Recent research demonstrated the potential of a comprehensive data mining methodology for building prognostic models from readily available operational and maintenance data. This approach builds a binary classifier that can determine the likelihood of a failure within a broad target window but cannot provide precise time to failure (TTF) estimations. This paper introduces a two-stage classification approach that helps improve the precision of TTF estimations. The new approach uses the initial methodology to learn a variety of base classifiers and then relies on meta-learning to integrate them. The paper details the model building process and demonstrates the usefulness of the proposed approach through a real-world prognostic application.

[1]  Chunsheng Yang,et al.  A DOMAIN INDEPENDENT DATA MINING METHODOLOGY FOR PROGNOSTICS , 2010 .

[2]  Chris Hunt,et al.  Monitoring and Managing Wheel Condition and Loading , 1999 .

[3]  G DietterichThomas.  An Experimental Comparison of Three Methods for Constructing Ensembles of Decision Trees , 2000 .

[4]  Nagi Gebraeel,et al.  Residual life predictions from vibration-based degradation signals: a neural network approach , 2004, IEEE Transactions on Industrial Electronics.

[5]  Mark A. Hall,et al.  Correlation-based Feature Selection for Discrete and Numeric Class Machine Learning , 1999, ICML.

[6]  Bertrand Poumellec,et al.  Precise life-time prediction using demarcation energy approximation for distributed activation energy reaction , 2006 .

[7]  Robert C. Holte,et al.  Explicitly representing expected cost: an alternative to ROC representation , 2000, KDD '00.

[8]  Joshua R. Smith,et al.  Multi-stage classi cation of images from features and related text , 1997 .

[9]  Danuta Rutkowska,et al.  A Multi-stage Classification Method in Application to Diagnosis of Larynx Cancer , 2004, ICAISC.

[10]  Chunsheng Yang,et al.  Learning to predict train wheel failures , 2005, KDD '05.

[11]  Adam Krzyżak,et al.  Methods of combining multiple classifiers and their applications to handwriting recognition , 1992, IEEE Trans. Syst. Man Cybern..

[12]  S. Matwin,et al.  Data Mining For Prediction of Aircraft Component Replacement , 1999 .

[13]  Tom Fawcett,et al.  Analysis and Visualization of Classifier Performance: Comparison under Imprecise Class and Cost Distributions , 1997, KDD.

[14]  Bernard Zenko,et al.  Is Combining Classifiers with Stacking Better than Selecting the Best One? , 2004, Machine Learning.

[15]  Stan Matwin,et al.  Data mining to predict aircraft component replacement , 1999, IEEE Intell. Syst..

[16]  Kishor S. Trivedi,et al.  On a method for mending time to failure distributions , 2005, 2005 International Conference on Dependable Systems and Networks (DSN'05).

[17]  Younès Bennani,et al.  Decision Fusion and Reliability Control in Handwritten Digit Recognition System , 2002 .

[18]  P. Wartenhorst,et al.  Time to failure, time to repair and availability of a two-unit standby system with Markovian degrading units , 1990 .

[19]  Mark Schwabacher,et al.  A Survey of Data -Driven Prognostics , 2005 .

[20]  Bernard Zenko,et al.  Stacking with Multi-response Model Trees , 2002, Multiple Classifier Systems.

[21]  D. Sornette,et al.  Optimal prediction of time-to-failure from information revealed by damage , 2006 .

[22]  Luigi P. Cordella,et al.  Network Intrusion Detection by a Multi-stage Classification System , 2004, Multiple Classifier Systems.

[23]  David H. Wolpert,et al.  Stacked generalization , 1992, Neural Networks.

[24]  Grigorios Tsoumakas,et al.  Effective Voting of Heterogeneous Classifiers , 2004, ECML.

[25]  Ted E. Senator,et al.  Multi-stage classification , 2005, Fifth IEEE International Conference on Data Mining (ICDM'05).

[26]  Thomas G. Dietterich.  An Experimental Comparison of Three Methods for Constructing Ensembles of Decision Trees: Bagging, Boosting, and Randomization , 2000, Machine Learning.