0
Research Papers

Manufacturing Process Monitoring With Nonparametric Change-Point Detection in Automotive Industry

[+] Author and Article Information
Shenghan Guo

Department of Industrial and Systems Engineering,
Rutgers, The State University of New Jersey,
Piscataway, NJ 08854
e-mail: sg888@scarletmail.rutgers.edu

Weihong (Grace) Guo

Mem. ASME
Department of Industrial and Systems Engineering,
Rutgers, The State University of New Jersey,
Piscataway, NJ 08854
e-mail: wg152@rutgers.edu

Amir Abolhassani

Global Data Insight and Analytics,
Ford Motor Company,
American Road, Dearborn, MI 48126
e-mail: aabolhas@ford.com

Rajeev Kalamdani

Global Data Insight and Analytics,
Ford Motor Company,
American Road, Dearborn, MI 48126
e-mail: rkalamda@ford.com

Saumuy Puchala

Powertrain Manufacturing Engineering,
Ford Motor Company,
American Road, Dearborn, MI 48126
e-mail: spuchala@ford.com

Annette Januszczak

Powertrain Manufacturing Engineering,
Ford Motor Company,
American Road, Dearborn, MI 48126
e-mail: ajanuszc@ford.com

Chandra Jalluri

Powertrain Manufacturing Engineering,
Ford Motor Company,
American Road, Dearborn, MI 48126
e-mail: cjalluri@ford.com

1Corresponding author.

Manuscript received July 3, 2018; final manuscript received April 12, 2019; published online May 28, 2019. Assoc. Editor: Satish Bukkapatnam.

J. Manuf. Sci. Eng 141(7), 071013 (May 28, 2019) (23 pages) Paper No: MANU-18-1506; doi: 10.1115/1.4043732 History: Received July 03, 2018; Accepted April 15, 2019

Automatic sensing devices and computer systems have been widely adopted by the automotive manufacturing industry, which are capable to record machine status and process parameters nonstop. While a manufacturing process always has natural variations, it is crucial to detect significant changes to the process for quality control, as such changes may be the early signs of machine faults. This motivates our study on change-point detection methods for automotive manufacturing. We aim at developing a systematic approach for detecting process changes retrospectively in complex, nonstationary data. The proposed approach consists of nonparametric change-point detection, alarm generation based on change-point estimations, and performance evaluation against historical maintenance records. For change-point detection, three nonparametric methods are suggested—least absolute shrinkage and selection operator (LASSO), thresholded LASSO, and wild binary segmentation (WBS). Multiple decision rules are proposed to determine how to generate alarms from change-point estimations. Numerical studies are conducted to demonstrate the performance of the proposed systematic approach. The different change-point detection methods and different decision rules are evaluated and compared, with scenarios for choosing one set of change-point detection method and decision rule over another combination identified. It is shown that LASSO and thresholded-LASSO outperform WBS when the shift size is small, but WBS produces a smaller false alarm rate and handles the clustering of changes better than LASSO or thresholded LASSO. Data from an automotive manufacturing plant are used in the case study to demonstrate the proposed approach. Guidelines for implementation are also provided.

Copyright © 2019 by ASME
Your Session has timed out. Please sign back in to continue.

References

Lowery, N. L., Vahdati, M. M., Potthast, R. W., and Holderbaum, W., 2013, “Classification and Fault Detection Methods for Fuel Cell Monitoring and Quality Control,” ASME J. Fuel Cell Sci. Technol., 10(2), p. 021002. [CrossRef]
Magarian, J. N., White, R. D., and Matson, D. M., 2016, “Real-Time Acoustic and Pressure Characterization of Two-Phase Flow for Quality Control of Expanded Polystyrene Injection Molding Processes,” ASME J. Manuf. Sci. Eng., 138(5), p. 051002. [CrossRef]
Abell, J. A., Chakraborty, D., Escobar, C. A., Im, K. H., Wegner, D. M., and Wincek, M. A., 2017, “Big Data-Driven Manufacturing—Process-Monitoring-for-Quality Philosophy,” ASME J. Manuf. Sci. Eng., 139(10), p. 101009. [CrossRef]
Wang, P., Fan, Z., Kazmer, D. O., and Gao, R. X., 2017, “Orthogonal Analysis of Multisensor Data Fusion for Improved Quality Control,” ASME J. Manuf. Sci. Eng., 139(10), p. 101008. [CrossRef]
Lu, G., Zhou, Y., Lu, C., and Li, X., 2017, “A Novel Framework of Change-Point Detection for Machine Monitoring,” Mech. Syst. Signal Proc., 83, pp. 533–548. [CrossRef]
Aminikhanghahi, S., and Cook, D. J., 2017, “A Survey of Methods for Time Series Change Point Detection,” Knowl. Inf. Syst., 51(2), pp. 339–367. [CrossRef] [PubMed]
Ho, L. L., and Aparisi, F., 2016, “ATTRIVAR: Optimized Control Charts to Monitor Process Mean With Lower Operational Cost,” Int. J. Prod. Econ., 182, pp. 472–483. [CrossRef]
Gunay, E., and Kula, U., 2016, “Integration of Production Quantity and Control Chart Design in Automotive Manufacturing,” Comput. Ind. Eng., 102, pp. 374–382. [CrossRef]
Čampulová, M., Veselik, P., and Michalek, J., 2017, “Control Chart and Six Sigma Based Algorithms for Identification of Outliers in Experimental Data, W an Application to Particulate Matter PM10,” Atmos. Pollut. Res., 8(4), pp. 700–708. [CrossRef]
Lin, Y.-C., Chou, C.-Y., and Chen, C.-H., 2017, “Robustness of the EWMA Median Control Chart to Non-Normality,” Int. J. Ind. Syst. Eng., 25(1), pp. 35–58.
Hawkins, D. M., Qiu, P., and Kang, C. W., 2003, “The Changepoint Model for Statistical Process Control,” J. Qual. Technol., 35(4), pp. 355–366. [CrossRef]
Chang, S. T., and Lu, K. P., 2016, “Change-Point Detection for Shifts in Control Charts Using EM Change-Point Algorithms,” Qual. Reliab. Eng. Int., 32(3), pp. 889–900. [CrossRef]
Sullivan, J. H., 2002, “Detection of Multiple Change Points From Clustering Individual Observations,” J. Qual. Technol., 34(4), pp. 371–383. [CrossRef]
Perry, M. B., Pignatiello, J. J., and Simpson, J. R., 2006, “Estimating the Change Point of a Poisson Rate Parameter With a Linear Trend Disturbance,” Qual. Reliab. Eng. Int., 22(4), pp. 371–384. [CrossRef]
Perry, M. B., Pignatiello, J. J., and Simpson, J. R., 2007, “Estimating the Change Point of the Process Fraction Non-Conforming With a Monotonic Change Disturbance in spc,” Qual. Reliab. Eng. Int., 23(3), pp. 327–339. [CrossRef]
Noorossana, R., and Shadman, A., 2009, “Estimating the Change Point of a Normal Process Mean With a Monotonic Change,” Qual. Reliab. Eng. Int., 25(1), pp. 79–90. [CrossRef]
Ayoubi, M., Kazemzadeh, R. B., and Noorossana, R., 2016, “Change Point Estimation in the Mean of Multivariate Linear Profiles With No Change Type Assumption via Dynamic Linear Model,” Qual. Reliab. Eng. Int., 32(2), pp. 403–433. [CrossRef]
Tercero-Gómez, V. G., Carmen Temblador-Pérez, M., Beruvides, M., and Hernández-Luna, A., 2013, “Nonparametric Estimator for the Time of a Step Change in the Trend of Random Walk Models With Drift,” Qual. Reliab. Eng. Int., 29(1), pp. 43–51. [CrossRef]
Ning, W., Yeh, A. B., Wu, X., and Wang, B., 2015, “A Nonparametric Phase I Control Chart for Individual Observations Based on Empirical Likelihood Ratio,” Qual. Reliab. Eng. Int., 31(1), pp. 37–55. [CrossRef]
Harchaoui, Z., and Lévy-Leduc, C., 2010, “Multiple Change-Point Estimation With a Total Variation Penalty,” J. Am. Stat. Assoc., 105(492), pp. 1480–1493. [CrossRef]
Choe, Y., Guo, W., Byon, E., Jin, J. J., and Li, J., 2016, “Change-Point Detection on Solar Panel Performance Using Thresholded LASSO,” Qual. Reliab. Eng. Int., 32(8), pp. 2653–2665. [CrossRef]
Levy-leduc, C., and Harchaoui, Z., 2008, “Catching Change-Points With Lasso,” Advances in Neural Information Processing Systems, Vol. 20, the Neural Information Processing Systems Conference (NIPS 2007), Vancouver, BC, Canada, pp. 617–624.
Zhou, S., 2009, “Thresholding Procedures for High Dimensional Variable Selection and Statistical Estimation,” Advances in Neural Information Processing Systems, Vol. 22, the Neural Information Processing Systems Conference (NIPS 2009), Vancouver, BC, Canada.
Fryzlewicz, P., 2014, “Wild Binary Segmentation for Multiple Change-Point Detection,” Ann. Stat., 42(6), pp. 2243–2281. [CrossRef]
Zhou, M., Zhou, Q., and Geng, W., 2016, “A New Nonparametric Control Chart for Monitoring Variability,” Qual. Reliab. Eng. Int., 32(7), pp. 2471–2479. [CrossRef]
Wang, S., Ji, B., Zhao, J., Liu, W., and Xu, T., 2017, Predicting Ship Fuel Consumption Based on LASSO Regression, 65(December 2018), pp. 817–824.
Zhang, Y., Minchin Jr, R. E., and Agdas, D., 2017, “Forecasting Completed Cost of Highway Construction Projects Using LASSO Regularized Regression,” J. Constr. Eng. Manage., 143(10), p. 04017071. [CrossRef]
Ni, P., Mangalathu, S., and Liu, K., 2018, “Enhanced Fragility Analysis of Buried Pipelines Through Lasso Regression,” Acta Geotechnica. (in press).
Choe, Y., Guo, W., Byon, E., Jin, J., and Li, J., 2016, “Change-Point Detection on Solar Panel Performance Using Thresholded LASSO,” Qual. Reliab. Eng. Int., 32(8), pp. 2653–2665. [CrossRef]
Trigano, T., and Cohen, J., 2017, “Intensity Estimation of Spectroscopic Signals With an Improved Sparse Reconstruction Algorithm,” IEEE Sig. Proc. Lett., 24(5), pp. 530–534. [CrossRef]
Korkas, K. K., and Fryzlewicz, P., 2017, “Multiple Change-Point Detection for Non-Stationary Time Series Using Wild Binary Segmentation,” Stat. Sin., 27(1), pp. 287–311.

Figures

Grahic Jump Location
Fig. 1

A segment of the raw data for cycle time; dashed boxes represent maintenance

Grahic Jump Location
Fig. 2

Visualization of surrogate data for S = 20 and L = 50 with varying δ: (a) δ = 0, (b) δ = 0.1, (c) δ = 0.5, (d) δ = 1, and (e) δ = 1.5

Grahic Jump Location
Fig. 3

Visualization of surrogate data for δ = 0.5 with varying S and L: (a) S = 10, L = 3; (b) S = 20, L = 10; (c) S = 20, L = 50; (d) S = 20, L = 100; (e) S = 5, L = 500

Grahic Jump Location
Fig. 4

FP/FN and AD/DL results for LASSO-based change-point detection without pruning, Dev = 0.7, δ = 1: (a)–(c) performance against S, given L = 10, 50, and 100, (d)–(f) performance against L, given S = 1, 5, and 10

Grahic Jump Location
Fig. 5

FP/FN and AD/DL results for LASSO-based change-point detection pruned by Decision Rule 2 with C = 2: (a)–(c) performance against S and (d)–(f) performance against L

Grahic Jump Location
Fig. 6

FP/FN and AD/DL results for LASSO-based change-point detection pruned by Decision Rule 3 with H/ = 0.5: (a)–(c) performance against S and (d)–(f) performance against L

Grahic Jump Location
Fig. 7

FP/FN and AD/DL results for TLASSO-based change-point detection without pruning: (a)–(c) performance against S and (d)–(f) performance against L

Grahic Jump Location
Fig. 8

FP/FN and AD/DL results for TLASSO-based change-point detection pruned by Decision Rule 2 with C = 2: (a)–(c) performance against S and (d)–(f) performance against L

Grahic Jump Location
Fig. 9

FP/FN and AD/DL results for TLASSO-based change-point detection pruned by Decision Rule 3 with H/ = 0.5: (a)–(c) performance against S and (d)–(f) performance against L

Grahic Jump Location
Fig. 10

FP/FN and AD/DL results for WBS-based change-point detection without pruning: (a)–(c) performance against S and (d)–(f) performance against L

Grahic Jump Location
Fig. 11

FP/FN and AD/DL results for WBS-based change-point detection pruned by Decision Rule 2 with C = 2: (a)–(c) performance against S and (d)–(f) performance against L

Grahic Jump Location
Fig. 12

FP/FN and AD/DL results for WBS-based change-point detection pruned by Decision Rule 3 with H/ = 0.5: (a)–(c) performance against S and (d)–(f) performance against L

Grahic Jump Location
Fig. 13

A segment of the Total Test Time data for accepted parts with one attempt

Grahic Jump Location
Fig. 14

Change-point detection by LASSO: (a) 52.5% data deviance explained by 184 change-points and (b) 62.5% data deviance explained by 356 change-points

Grahic Jump Location
Fig. 15

Change-point detection by TLASSO: (a) σ = 0.09 and (b) σ = 0.11

Grahic Jump Location
Fig. 16

Change-point detection by WBS: (a) M = 5000 and (b) M = 11,000

Grahic Jump Location
Fig. 17

Number of true detections and false positives on change-points from LASSO-based detection: (a) Decision Rule 2 with varying threshold values C and (b) Decision Rule 3 with varying threshold values expressed as H/

Grahic Jump Location
Fig. 18

Number of true detections and false positives on change-points from TLASSO-based detection: (a) Decision Rule 2 with varying threshold values C and (b) Decision Rule 3 with varying threshold values expressed as H/

Grahic Jump Location
Fig. 19

Number of true detections and false positives on change-points from WBS-based detection: (a) Decision Rule 2 with varying threshold values C and (b) Decision Rule 3 with varying threshold values expressed as H/

Grahic Jump Location
Fig. 20

3D visualization of numerical results for LASSO-based change-point detection, with L on the x-axis, S on the y-axis, and δ on the z-axis: (a) no pruning, (b) prune by Decision Rule 2, and (c) prune by with Decision Rule 3

Grahic Jump Location
Fig. 21

3D visualization of numerical results for TLASSO-based change-point detection, with L on the x-axis, S on the y-axis, and δ on the z-axis: (a) no pruning, (b) prune by Decision Rule 2, and (c) prune by Decision Rule 3

Grahic Jump Location
Fig. 22

3D visualization of numerical results for WBS-based change-point detection, with L on the x-axis, S on the y-axis, and δ on the z-axis: (a) no pruning, (b) prune by Decision Rule 2, and (c) prune by Decision Rule 3

Tables

Errata

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