مهندسی سیستم و بهره‌وری

مهندسی سیستم و بهره‌وری

طراحی نمودار کنترل مبتنی بر یادگیری ماشین برای پایش پراکندگی جریان داده‌های با ابعاد بالا در فاز 2

مقالات آماده انتشار

نوع مقاله : پژوهشی

نویسندگان
میر میلاد قزوینی 1
علی سلماس نیا 2
1 دانشجوی دکتری، دانشکده مهندسی صنایع، دانشگاه ایوان‌کی، سمنان، ایران
2 نویسنده مسئول: دانشیار، دانشکده فنی و مهندسی، گروه مهندسی صنایع، دانشگاه قم، قم، ایران
10.22034/sep.2025.2069790.1387
چکیده
این یک نمودار کنترل مبتنی بر شبکه عصبی کاملاً متصل برای پایش پراکندگی فرآیندهای چند متغیره با دو ویژگی مهم توسعه داده می‌شود: (1) قابلیت استفاده برای پایش پراکندگی فرآیندها با جریان داده ابعاد بالا، و (2) عدم نیاز به برقراری مفروضات آماری محدودکننده ازجمله نرمال بون مشخصه‌های کیفی تحت مطالعه و مستقل بودن مشاهدات در نمونه‌های اخذشده. چالش مهمی که معمولاً در آموزش شبکه‌های عصبی با آن مواجه هستیم بیش برازش یا شکست تعمیم است. در این مطالعه برای مواجهه با چالش ذکرشده از دو ابزار لایه حذف تصادفی و منظم‌سازی وزن در طراحی شبکه استفاده می‌شود. به‌علاوه جهت آموزش بهتر شبکه عصبی و برخلاف بیشتر نمودارهای کنترل مبتنی بر ابزارهای یادگیری که از الگوی دو کلاسه صفر و یک به‌عنوان مقادیر هدف بهره می‌برند، در این مطالعه مقادیر هدف بر اساس اندازه و تعداد مؤلفه‌های شیفت‌یافته تعیین‌شده‌اند، به‌گونه‌ای که با افزایش اندازه شیفت و تعداد مؤلفه‌های شیفت‌یافته، مقادیر هدف نیز افزایش یابند. در ادامه، به‌منظور افزایش توان نمودار کنترل توسعه داده‌شده در کشف اختلالات رخ‌داده در مؤلفه‌های ماتریس کوواریانس، نسخه بهبودیافته‌ای از آن به کمک دو قانون حساس‌سازی 2 از 3 و 4 از 5 ارائه می‌شود. عملکرد رویکردهای پیشنهادی با استفاده از یک مثال عددی با دو نمودار کنترل ATL و RPLR مقایسه می‌شوند. نتایج نشان می‌دهد که رویکرد مجهز شده به قوانین حساس‌سازی عملکرد بهتری از نمودارهای رقیب برحسب دو شاخص ARL و SDRL دارد.

تازه های تحقیق

  • پایش پراکندگی فرآیندهای ابعاد بالا با شبکه عصبی
  • مطالعه اثر قوانین حساس‌سازی بر ویژگی‌های طول دنباله
  • قابلیت استفاده در شرایط نقض مفروضات آماری

کلیدواژه‌ها
پایش پراکندگی
فرآیند ابعاد بالا
نمودار کنترل مبتنی بر شبکه عصبی
تنظیم‌سازی
قانون حساس‌سازی
لایه حذف تصادفی
موضوعات

عنوان مقاله English

A Machine Learning-based Control Chart for Monitoring the Dispersion of High-dimensional Data Streams in Phase II

نویسندگان English

Mir Milad Ghazvini 1
Ali Salmasnia 2
1 Ph.D. Student, Faculty of Industrial Engineering, University of Eyvanekey, Eyvanekey, Iran
2 Corresponding: Associate Professor, Department of Industrial Engineering, Faculty of Engineering, University of Qom, Qom, Iran
چکیده English

This fully connected neural network-based control chart is developed for monitoring the dispersion of multivariate processes with two important features: (1) the ability to be used for monitoring the dispersion of processes with high-dimensional data streams, and (2) the absence of the need to establish restrictive statistical assumptions, such as normality of the quality characteristics under study and the independence of observations in the samples taken. An important challenge that we usually face in training neural networks is overfitting or generalization failure. In this study, two tools of dropout layer and weight regularization are used in network design to face the mentioned challenge. In addition, in order to better train the neural network and unlike most control charts based on learning tools that use a two-class pattern of zero and one as target values, in this study the target values are determined based on the size and number of shifted components, so that as the shift size and the number of shifted components increase, the target values also increase. Next, in order to increase the power of the developed control chart in detecting disturbances in the covariance matrix elements, an improved version is presented with the help of two sensitizing rules 2 out of 3 and 4 out of 5. The performance of the proposed approaches is compared with two control charts ATL and RPLR using a numerical example. The results show that the approach equipped with sensitizing rules performs better than the competing charts in terms of two indices ARL and SDRL.

کلیدواژه‌ها English

Dispersion monitoring
High-dimensional process
Neural network-based control chart
Regularization
Sensitizing rule
Random removal layer

Copyright © Mir Milad Ghazvini, Ali Salmasnia

 

License

This article is released under the Creative Commons Attribution (CC BY 4.0) license. Anyone is free to copy, share, translate, and adapt this article for any purpose, whether commercial or non-commercial, as long as proper citation is given to the authors and original publication.

Abbasi, B., & Niaki, S. T. A. (2007). Monitoring high-yields processes with defects count in nonconforming items by artificial neural network. Applied mathematics and computation, 188(1), 262-270. https://doi.org/10.1016/j.amc.2006.09.114
Abdella, G. M., Al-Khalifa, K. N., Kim, S., Jeong, M. K., Elsayed, E. A., & Hamouda, A. M. (2017). Variable selection-based multivariate cumulative sum control chart. Quality and Reliability Engineering International, 33(3), 565–578.  https://doi.org/10.1002/qre.2041
Abdella, G. M., Kim, J., Kim, S., Al-Khalifa, K. N., Jeong, M. K., Hamouda, A. M., & Elsayed, E. A. (2019). An adaptive thresholding-based process variability monitoring. Journal of Quality Technology, 51(3), 242–256. https://doi.org/10.1080/00224065.2019.1569952
Adegoke, N. A., Dawod, A., Adeoti, O. A., Sanusi, R. A., & Abbasi, S. A. (2022). Monitoring multivariate coefficient of variation for high-dimensional processes. Quality and Reliability Engineering International, 38(5), 2606–2621. https://doi.org/10.1002/qre.3094
Adeoti, O. A., & Malela-Majika, J. C. (2020). Double exponentially weighted moving average control chart with supplementary runs-rules. Quality Technology & Quantitative Management, 17(2), 149–172. https://doi.org/10.1080/16843703.2018.1560603
Ahmadi Karavigh, M. H., & Amiri, A. (2024). MEWMA-based control charts with runs rules for monitoring multivariate simple linear regression profiles in Phase II. Communications in Statistics—Simulation and Computation, 53(3), 1107–1134. https://doi.org/10.1080/03610918.2022.2028833
Amiri, A., Salmasnia, A., Zarifi, M., & Maleki, M. R. (2023). Adaptive Shewhart control charts under fuzzy parameters with tuned particle swarm optimization algorithm. Journal of Industrial Integration and Management, 8(2), 241–276. https://doi.org/10.1142/S2424862221500226
Antzoulakos, D. L., & Rakitzis, A. C. (2008). The modified r out of m control chart. Communications in Statistics—Simulation and Computation, 37(2), 396–408. https://doi.org/10.1080/03610910701501906
Arkat, J., Niaki, S. T. A., & Abbasi, B. (2007). Artificial neural networks in applying MCUSUM residuals charts for AR(1) processes. Applied Mathematics and Computation, 189(2), 1889–1901. https://doi.org/10.1016/j.amc.2006.12.081
Chew, X., Khoo, M. B. C., Khaw, K. W., & Lee, M. H. (2021). An improved Hotelling’s T² chart for monitoring a finite horizon process based on run rules schemes: A Markov-chain approach. Applied Stochastic Models in Business and Industry, 37(3), 577–591. https://doi.org/10.1002/asmb.2596
Dafnis, S. D., Perdikis, T., & Papadopoulos, G. K. (2024). Improved chi-square control charts with weak-run rules. Communications in Statistics—Theory and Methods, 53(14), 5248–5264. https://doi.org/10.1080/03610926.2023.2215357
Ebadi, M., Chenouri, S., & Steiner, S. H. (2023). Phase I analysis of high-dimensional processes in the presence of outliers. Journal of Quality Technology, 55(4), 469–488. https://doi.org/10.1080/00224065.2023.2196034
Fan, J., Shu, L., Yang, A., & Li, Y. (2021). Phase I analysis of high-dimensional covariance matrices based on sparse leading eigenvalues. Journal of Quality Technology, 53(4), 333–346. https://doi.org/10.1080/00224065.2020.1746212
Feng, L., Ren, H., & Zou, C. (2020). A setwise EWMA scheme for monitoring high-dimensional datastreams. Random Matrices: Theory and Applications, 9(2), 2050004. https://doi.org/10.1142/S2010326320500045
Gómez, A. M. E., Li, D., & Paynabar, K. (2022). An adaptive sampling strategy for online monitoring and diagnosis of high-dimensional streaming data. Technometrics, 64(2), 253–269. https://doi.org/10.1080/00401706.2021.1967198
Hric, P., & Sabahno, H. (2024). Developing machine learning-based control charts for monitoring different GLM-type profiles with different link functions. Applied Artificial Intelligence, 38(1), 2362511. https://doi.org/10.1080/08839514.2024.2362511
Imran, M., Dai, H. L., Zaidi, F. S., Hu, X., Tran, K. P., & Sun, J. (2024). Analyzing out-of-control signals of T² control chart for compositional data using artificial neural networks. Expert Systems with Applications, 238, 122165. https://doi.org/10.1016/j.eswa.2023.122165
Jafari, M., Maleki, M. R., & Salmasnia, A. (2023). A high-dimensional control chart for monitoring process variability under gauge imprecision effect. Production Engineering, 17(3), 547-564. https://doi.org/10.1007/s11740-022-01166-3
Jiang, W., Wang, K., & Tsung, F. (2012). A variable-selection-based multivariate EWMA chart for process monitoring and diagnosis. Journal of Quality Technology, 44(3), 209–230. https://doi.org/10.1080/00224065.2012.11917896
Khilare, S. K., & Shirke, D. T. (2023). The fraction nonconforming m-of-m control chart with warning limits. Thailand Statistician, 21(2), 435-449.
Khoo, M. B. C. (2003). Design of runs rules schemes. Quality Engineering, 16(1), 27–43. https://doi.org/10.1081/QEN-120020769
Kim, J. M., & Ha, I. D. (2022). Deep learning-based residual control chart for count data. Quality Engineering, 34(3), 370–381. https://doi.org/10.1080/08982112.2022.2044049
Kim, J., Abdella, G. M., Kim, S., Al-Khalifa, K. N., & Hamouda, A. M. (2019). Control charts for variability monitoring in high-dimensional processes. Computers & Industrial Engineering, 130, 309–316. https://doi.org/10.1016/j.cie.2019.02.012
Klein, M. (2000). Two alternatives to the Shewhart X̄ control chart. Journal of Quality Technology, 32(4), 427–431. https://doi.org/10.1080/00224065.2000.11980028
Koutras, M. V., Bersimis, S., & Maravelakis, P. (2007). Statistical process control using Shewhart control charts with supplementary runs rules. Methodology and Computing in Applied Probability, 9(2), 207–224. https://doi.org/10.1007/s11009-007-9016-8
Maboudou-Tchao, E. M. (2021). High-dimensional data monitoring using support machines. Communications in Statistics—Simulation and Computation, 50(7), 1927–1942. https://doi.org/10.1080/03610918.2019.1588312
Mallikarjun, N. S., Tayyapa, J. B., Moshayedi, A. J., & Kisanrao, S. S. (2024). Performance analysis of ANN control chart for monitoring the individual measurements of manufacturing process. In Proceedings of the 2024 3rd International Symposium on Robotics, Artificial Intelligence and Information Engineering (pp. 130–135). https://doi.org/10.1145/3689299.3689323
Mehmood, R., Riaz, M., Ali, I., & Lee, M. H. (2021). Generalized Hotelling T² control chart based on bivariate ranked set techniques with runs rules. Transactions of the Institute of Measurement and Control, 43(10), 2180–2195. https://doi.org/10.1177/0142331221992670
Okhrin, Y., Schmid, W., & Semeniuk, I. (2025). A control chart for monitoring image processes based on convolutional neural networks. Statistica Neerlandica, 79(1), e12366. https://doi.org/10.1111/stan.12366
Quinino, V. B., Ho, L. L., Cruz, F. R. B., & Quinino, R. C. (2023). A simple procedure to improve the performance of Klein’s 2-of-2 run rule in the Shewhart X̄ control chart. Quality and Reliability Engineering International, 39(7), 3014–3029. https://doi.org/10.1002/qre.3412
Riaz, M., Abbas, N., & Does, R. J. M. M. (2011). Improving the performance of CUSUM charts. Quality and Reliability Engineering International, 27(4), 415–424. https://doi.org/10.1002/qre.1124
Ruelas-Santoyo, E. A., Figueroa-Fernández, V., Tapia-Esquivias, M., Pantoja-Pacheco, Y. V., Bravo-Santibáñez, E., & Cruz-Salgado, J. (2024). Monitoring and interpretation of process variability generated from the integration of the multivariate cumulative sum control chart and artificial intelligence. Applied Sciences, 14(21), 9705. https://doi.org/10.3390/app14219705
Sabahno, H., & Amiri, A. (2023). New statistical and machine learning-based control charts with variable parameters for monitoring generalized linear model profiles. Computers & Industrial Engineering, 184, 109562. https://doi.org/10.1016/j.cie.2023.109562
Sabahno, H., & Khodadad, D. (2025). A spatiotemporal scheme for process control using image analysis of speckle patterns. Quality and Reliability Engineering International, 41(5), 1773–1788. https://doi.org/10.1002/qre.3758
Saemian, M., Salmasnia, A., & Maleki, M. R. (2022). A generalized multiple dependent state sampling chart based on ridge penalized likelihood ratio for high-dimensional covariance matrix monitoring. Scientia Iranica. https://doi.org/10.24200/SCI.2022.60169.6640
Safikhani, E., Salmasnia, A., & Maleki, M. R. (2023). A ridge penalized likelihood ratio chart for Phase II monitoring of high-dimensional process dispersion under measurement system inaccuracy. International Journal of Industrial Engineering, 34(2), 1-17. https://doi.org/10.22068/ijiepr.34.2.12
Salmasnia, A., Asghari, A., & Maleki, M. R. (2025a). Effect of parameter estimation on Phase II performance of ridge penalized likelihood ratio control chart for monitoring high-dimensional process variability. Journal of Data, Information and Management, 7(3), 265-286. https://doi.org/10.1007/s42488-025-00152-5
Salmasnia, A., Maleki, M. R., & Mirzaei, M. (2025b). Double sampling adaptive thresholding LASSO variability chart for Phase II monitoring of high-dimensional data streams. Journal of Industrial Integration and Management, 10(02), 277-299. https://doi.org/10.1142/S242486222350001X
Sun, J., Zhou, S., & Veeramani, D. (2023). A neural network-based control chart for monitoring and interpreting autocorrelated multivariate processes using layer-wise relevance propagation. Quality Engineering, 35(1), 33–47. https://doi.org/10.1080/08982112.2022.2087041
Tang, A., Mukherjee, A., & Wang, X. (2023). Distribution-free Phase-II monitoring of high-dimensional industrial processes via origin and modified interpoint distance-based algorithms. Computers & Industrial Engineering, 179, 109161. https://doi.org/10.1016/j.cie.2023.109161
Tran, K. P. (2017). Run rules median control charts for monitoring process mean in manufacturing. Quality and Reliability Engineering International, 33(8), 2437–2450. https://doi.org/10.1002/qre.2201
Woodall, W. H., Saleh, N. A., & Mahmoud, M. A. (2023). Equivalences between multiple dependent state sampling, chain sampling, and control chart runs rules. Quality Engineering, 35(1), 142–151. https://doi.org/10.1080/08982112.2022.2102428
Yeganeh, A., Johannssen, A., Chukhrova, N., Abbasi, S. A., & Pourpanah, F. (2023a). Employing machine learning techniques in monitoring autocorrelated profiles. Neural Computing and Applications, 35(22), 16321–16340. https://doi.org/10.1007/s00521-023-08483-3
Yeganeh, A., Shadman, A., Shongwe, S. C., & Abbasi, S. A. (2023b). Employing evolutionary artificial neural network in risk-adjusted monitoring of surgical performance. Neural Computing and Applications, 35(14), 10677–10693. https://doi.org/10.1007/s00521-023-08257-x
Yeganeh, A., Pourpanah, F., & Shadman, A. (2021a). An ANN-based ensemble model for change point estimation in control charts. Applied Soft Computing, 110, 107604. https://doi.org/10.1016/j.asoc.2021.107604
Yeganeh, A., & Shadman, A. (2021b). Monitoring linear profiles using artificial neural networks with run rules. Expert Systems with Applications, 168, 114237. https://doi.org/10.1016/j.eswa.2020.114237
Yeganeh, A., Shadman, A., & Abbasi, S. A. (2022). Enhancing the detection ability of control charts in profile monitoring by adding RBF ensemble model. Neural Computing and Applications, 34(12), 9733–9757. https://doi.org/10.1007/s00521-022-07109-4
مهندسی سیستم و بهره‌وری

مقالات آماده انتشار، پذیرفته شده
انتشار آنلاین از 12 شهریور 1404

  • تاریخ دریافت 27 تیر 1404
  • تاریخ بازنگری 26 مرداد 1404
  • تاریخ پذیرش 12 شهریور 1404
  • تاریخ اولین انتشار 12 شهریور 1404
  • تاریخ انتشار 12 شهریور 1404