0
Research Papers: Gas Turbines: Oil and Gas Applications

Resistant Statistical Methodologies for Anomaly Detection in Gas Turbine Dynamic Time Series: Development and Field Validation

[+] Author and Article Information
Giuseppe Fabio Ceschini, Thomas Hubauer, Alin Murarasu

Siemens AG,
Nürnberg 90461, Germany

Nicolò Gatta, Mauro Venturini

Dipartimento di Ingegneria,
Università degli Studi di Ferrara,
Ferrara 44122, Italy

Contributed by the Oil and Gas Applications Committee of ASME for publication in the JOURNAL OF ENGINEERING FOR GAS TURBINES AND POWER. Manuscript received July 14, 2017; final manuscript received August 10, 2017; published online November 21, 2017. Editor: David Wisler.

J. Eng. Gas Turbines Power 140(5), 052401 (Nov 21, 2017) (11 pages) Paper No: GTP-17-1361; doi: 10.1115/1.4038155 History: Received July 14, 2017; Revised August 10, 2017

The reliability of gas turbine (GT) health state monitoring and forecasting depends on the quality of sensor measurements directly taken from the unit. Outlier detection techniques have acquired a major importance, as they are capable of removing anomalous measurements and improve data quality. To this purpose, statistical parametric methodologies are widely employed thanks to the limited knowledge of the specific unit required to perform the analysis. The backward and forward moving window (BFMW) k–σ methodology proved its effectiveness in a previous study performed by the authors, to also manage dynamic time series, i.e., during a transient. However, the estimators used by the k–σ methodology are usually characterized by low statistical robustness and resistance. This paper aims at evaluating the benefits of implementing robust statistical estimators for the BFMW framework. Three different approaches are considered in this paper. The first methodology, k-MAD, replaces mean and standard deviation (SD) of the k–σ methodology with median and mean absolute deviation (MAD), respectively. The second methodology, σ-MAD, is a novel hybrid scheme combining the k–σ and the k-MAD methodologies for the backward and the forward windows, respectively. Finally, the biweight methodology implements biweight mean and biweight SD as location and dispersion estimators. First, the parameters of these methodologies are tuned and the respective performance is compared by means of simulated data. Different scenarios are considered to evaluate statistical efficiency, robustness, and resistance. Subsequently, the performance of these methodologies is further investigated by injecting outliers in field datasets taken on selected Siemens GTs. Results prove that all the investigated methodologies are suitable for outlier identification. Advantages and drawbacks of each methodology allow the identification of different scenarios in which their application can be most effective.

Copyright © 2018 by ASME
Your Session has timed out. Please sign back in to continue.

References

Roumeliotis, I. , Aretakis, N. , and Alexiou, A. , 2016, “ Industrial Gas Turbine Health and Performance Assessment With Field Data,” ASME Paper No. GT2016-57722.
Tsoutsanis, E. , Meskin, N. , Benammar, M. , and Khashayar, K. , 2015, “ Transient Gas Turbine Performance Diagnostics Through Nonlinear Adaptation of Compressor and Turbine Maps,” ASME J. Eng. Gas Turbines Power, 137(9), p. 091201. [CrossRef]
Jiang, X. , and Foster, C. , 2014, “ Plant Performance Monitoring and Diagnostics: Remote, Real-Time and Automation,” ASME Paper No. GT2014-27314.
Venturini, M. , and Therkorn, D. , 2013, “ Application of a Statistical Methodology for Gas Turbine Degradation Prognostics to Alstom Field Data,” ASME J. Eng. Gas Turbines Power, 135(9), p. 091603. [CrossRef]
Cavarzere, A. , and Venturini, M. , 2011, “ Application of Forecasting Methodologies to Predict Gas Turbine Behavior Over Time,” ASME J. Eng. Gas Turbines Power, 134(1), p. 012401. [CrossRef]
Lanzante, J. , 1996, “ Resistant, Robust and Non-Parametric Techniques for the Analysis of Climate Data: Theory and Examples, Including Applications to Historical Radiosonde Station Data,” Int. J. Climatol., 16(11), pp. 1197–1226. [CrossRef]
Van Paridon, A. , Bacic, M. , and Ireland, P. T. , 2016, “ Kalman Filter Development for Real Time Proper Orthogonal Decomposition Disc Temperature Model,” ASME Paper No. GT2016-56330.
Hurst, A. M. , Carter, S. , Firth, D. , Szary, A. , and vanDeWeert, J. , 2015, “ Real-Time, Advanced Electrical Filtering for Pressure Transducer Frequency Response Correction,” ASME Paper No. GT2015-42895.
Gutierrez, L. A. , Pezzini, P. , Tucker, D. , and Banta, L. , 2014, “ Smoothing Techniques for Real-Time Turbine Speed Sensors,” ASME Paper No. GT2014-25407.
Dewallef, P. , and Borguet, S. , 2013, “ A Methodology to Improve the Robustness of Gas Turbine Engine Performance Monitoring Against Sensor Faults,” ASME J. Eng. Gas Turbines Power, 135(5), p. 051601. [CrossRef]
Gomez, J. , 2011, Kalman Filtering, Nova Science Publishers, Hauppauge, NY. [PubMed] [PubMed]
Pinelli, M. , Venturini, M. , and Burgio, M. , 2003, “ Statistical Methodologies for Reliability Assessment of Gas Turbine Measurements,” ASME Paper No. GT2003-38407.
Martin, R. , and Thomson, D. , 1982, “ Robust-Resistant Spectrum Estimation,” Proc. IEEE, 70(9), pp. 1097–1115. [CrossRef]
Hampel, F. , 1974, “ The Influence Curve and Its Role in Robust Estimation,” J. Am. Stat. Assoc., 69(346), pp. 383–393. [CrossRef]
Bhattacharya, G. , Ghosh, K. , and Chowdhury, A. , 2015, “ Outlier Detection Using Neighborhood Rank Difference,” Pattern Recognit. Lett., 60–61, pp. 24–31. [CrossRef]
Takahashi, T. , Tomioka, R. , and Yamanishi, K. , 2014, “ Discovering Emerging Topics in Social Streams Via Link-Anomaly Detection,” IEEE Trans. Knowl. Data Eng., 26(1), pp. 120–130. [CrossRef]
Takeuchi, J. , and Yamanishi, K. , 2006, “ A Unifying Framework for Detecting Outliers and Change Points From Time Series,” IEEE Trans. Knowl. Data Eng., 18(4), pp. 482–492. [CrossRef]
Xu, S. , Baldea, M. , Edgar, T. , Wojsznis, W. , Blevins, T. , and Nixon, M. , 2014, “ An Improved Methodology for Outlier Detection in Dynamic Datasets,” AIChE J., 61(2), pp. 419–433. [CrossRef]
Ceschini, G. , Gatta, N. , Venturini, M. , Hubauer, T. , and Murarasu, A. , 2017, “ Optimization of Statistical Methodologies for Anomaly Detection in Gas Turbine Dynamic Time Series,” ASME Paper No. GT2017-63409.
Ceschini, G. , Gatta, N. , Venturini, M. , Hubauer, T. , and Murarasu, A. , 2017, “ A Comprehensive Approach for Detection, Classification and Integrated Diagnostics of Gas Turbine Sensors (DCIDS),” ASME Paper No. GT2017-63411.
Huber, P. , 1981, Robust Statistics, Wiley, New York. [CrossRef]
Rousseeuw, P. , and Croux, C. , 1993, “ Alternatives to the Median Absolute Deviation,” J. Am. Stat. Assoc., 88(424), pp. 1273–1283.
Kafadar, K. , 1983, “ The Efficiency of the Biweight as a Robust Estimator of Location,” J. Res. Natl. Bur. Stand., 88(2), pp. 105–116.
Leys, C. , Ley, C. , Klein, O. , Bernard, P. , and Licata, L. , 2013, “ Detecting Outliers: Do Not Use Standard Deviation Around the Mean, Use Absolute Deviation Around the Median,” J. Exp. Soc. Psychol., 49(4), pp. 764–766. [CrossRef]
Yahaya, S. , Othman, A. , and Keselman, H. , 2004, “ Testing the Equality of Location Parameters for Skewed Distributions Using S1 With High Breakdown Robust Scale Estimators,” Statistics for Industry and Technology, Birkhäuser Verlag, Basel, Switzerland pp. 319–328.
Hoaglin, D. , Mosteller, F. , and Tukey, J. , 1983, Understanding Robust and Exploratory Data Analysis, Wiley, New York.
Young, G. A. , and Smith, R. L. , 2005, Essential of Statistical Inference, Cambridge University Press, Cambridge, UK. [CrossRef]
Sharma, A. , Golubchik, L. , and Govindan, R. , 2010, “ Sensor Faults: Detection Methods and Prevalence in Real-World Datasets,” ACM Trans. Sens. Networks, 6(3), p. 23.

Figures

Grahic Jump Location
Fig. 1

Simulated time series for efficiency evaluation (1% noise; outlier magnitude equal to 7%)

Grahic Jump Location
Fig. 2

True-positive rate results for efficiency evaluation

Grahic Jump Location
Fig. 3

False-negative rate results for efficiency evaluation

Grahic Jump Location
Fig. 4

Simulated time series for robustness evaluation (2% noise; outlier magnitude equal to 7%)

Grahic Jump Location
Fig. 5

True-positive rate results for robustness evaluation

Grahic Jump Location
Fig. 6

False-negative rate results for robustness evaluation

Grahic Jump Location
Fig. 7

Simulated time series for resistance evaluation (1% noise; outlier magnitude equal to 5%)

Grahic Jump Location
Fig. 8

True-positive rate results for resistance evaluation

Grahic Jump Location
Fig. 9

False-negative rate results for resistance evaluation

Grahic Jump Location
Fig. 10

Nondimensional temperature T1 dataset with 7% magnitude injected outliers

Grahic Jump Location
Fig. 11

Nondimensional temperature T2 dataset with 7% magnitude injected outliers

Grahic Jump Location
Fig. 12

Nondimensional vibration V1 dataset with 7% magnitude injected outliers

Grahic Jump Location
Fig. 13

Nondimensional pressure P1 dataset with 7% magnitude injected outliers

Grahic Jump Location
Fig. 14

Nondimensional temperature T3 dataset with 5% magnitude injected outliers

Grahic Jump Location
Fig. 15

Nondimensional rotational speed S1 dataset with 5% magnitude injected outliers

Grahic Jump Location
Fig. 16

Nondimensional temperature T4 dataset with 5% magnitude injected outliers

Grahic Jump Location
Fig. 17

Nondimensional pressure P1 dataset with 5% magnitude injected outliers

Grahic Jump Location
Fig. 18

True-positive rate and FNR values as a function of outlier magnitude for the temperature T1 dataset

Grahic Jump Location
Fig. 19

True-positive rate and FNR values as a function of outlier magnitude for the temperature T2 dataset

Grahic Jump Location
Fig. 20

True-positive rate and FNR values as a function of outlier magnitude for the vibration V1 dataset

Grahic Jump Location
Fig. 21

True-positive rate and FNR values as a function of outlier magnitude for the pressure P1 dataset

Grahic Jump Location
Fig. 22

True-positive rate and FNR values as a function of outlier contamination rate for the temperature T3 dataset

Grahic Jump Location
Fig. 23

True-positive rate and FNR values as a function of the presence of outlier contamination rate for the rotational speed S1 dataset

Grahic Jump Location
Fig. 24

True-positive rate and FNR values as a function of the presence of outlier contamination rate for the temperature T4 dataset

Grahic Jump Location
Fig. 25

True-positive rate and FNR values as a function of the presence of outlier contamination rate for the pressure P1 dataset

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In