Öznitelik Seçimi ve Makine Öğrenmesi Yöntemleri ile Maske Üretiminde Kalite Performansının İyileştirilmesi ve Bir Uygulama
Abstract
Teknolojinin gelişmesi ile birlikte otomatik veri tutan otomasyon sistemleri sayesinde büyük veri tabanları daha ulaşılabilir hale gelmekte ve birçok alanda büyük very tabanlarının kullanımına imkân vermektedir. Veri tabanlarında tutulan verilerin analizlerinin yapılıp bilgiye dönüştürülerek süreç ile ilgili kararlar alınmasında yapay zekânın alt dalı olan makine öğrenmesi yaklaşımlarından yararlanılmaktadır. Bu makalede cerrahi (tıbbi) maskenin gövde üretim süreci analiz edilmektedir. Bilindiği gibi cerrahi maskeler, COVID-19 pandemisi ile birlikte tüm dünyada yaygınlaşarak hayatımızın bir parçası haline gelmiştir. Cerrahi maske gövde üretim sürecinde üretim faktörlerine ait gerçek veriler kullanılarak öncelikle filtreleme öznitelik seçim yöntemleri ile analizler yapılıp kullanılacak öznitelik seçim yöntemi belirlenmiştir. Belirlenen öznitelik seçimi yöntemi ile ürün kalitesi üzerinde etkili olan faktörler belirlenir. İkinci olarak, hatasız ürünlerin üretimindeki faktörlerin (özniteliklerin) değerlerini ve değer aralıklarını belirlemek için makine öğrenmesi yöntemlerinden yararlanılmıştır. İkinci aşamada kurulan makine öğrenmesi modellerinin performansları öznitelik seçimi analizi ile artırılmıştır. Çalışmada makine öğrenmesi algoritmalarına yapılan parametre optimizasyonları ile birlikte hatalı ürün oranını tahmin etmek için en iyi algoritmanın %92,3 doğruluk, %91,9 F ölçümü ve %93 AUC değeri ile Ibk algoritması olduğu görülmüştür. Son olarak çalışmada ortaya çıkan karar kuralları doğrultusunda, maske gövde üretim sürecinde gövde kısmını oluşturan üst/orta/alt katmanlar için kullanılan kumaş türlerinin, hatalı veya hatasız ürün oranlarını büyük ölçüde etkilediği gözlemlendi. Burun etrafını saran çubuk aparatları k sınıfına ait ise birçok maskenin hatalı olduğu tespit edilmiştir. Uygulama sonuçlarına göre iyileştirme önerileri sunulmuştur.
Keywords
Makine Öğrenmesi Özellik Seçimi Sınıflandırma Hatalı ve Hatasız Ürün Tahmini Filtreleme Methodları
References
- [1] Öztemel E. (2016). Artificial Neural Networks. 4nd ed. Istanbul, Turkey, Papatya Press.
- [2] Doğan A, Birant D. (2021). “Machine learning and data mining in manufacturing”. Expert Systems with Applications, 166(2021), 1-22.
- [3] Öztemel E, Gürsev S. (2018). “Literature review of industry 4.0 and related technologies”. Journal of Intelligent Manufacturing, 31(2020), 127-182.
- [4] Cho E, Jun J, Chang T, Choi Y. (2020). “Quality prediction modeling of plastic extrusion process”. ICIC Express Letters Part B: Applications, 11(5), 447–452.
- [5] Chen K, Hu YH, Hsieh YC. (2015). “Predicting customer churn from valuable B2B customers in the logistics industry: A case study”. Information Systems and E-Business Management, 13(2015), 475–494.
- [6] Brillinger M, Wuwer M, Hadi MU, Haas F. (2021). “Energy prediction for CNC machining with machine learning”. CIRP Journal of Manufacturing Science and Technology, 35(2021), 715-723.
- [7] Miguéis V, Freitas LA, Garcia PJV, Silva A. (2018). “Early segmentation of students according to their academic performance: A predictive modelling approach”. Decision Support Systems, 115(2018), 36–51.
- [8] Go A, Bhayani R, Huang L. (2019). “Twitter sentiment Classification Using Distant Supervision”. Stanford, United States of America, Project Report, CS224N.
- [9] Yucalar, F., Özçift, A., Borandağ, E., & Kılınç, D. (2020). Yazılım kalitesi mühendisliğinde çoklu sınıflandırıcılar: Yazılım hata tahmin yeteneğini geliştirmek için tahmin edicileri birleştirmek. Engineering Science and Technology, International Journal, 23 (4), 938-950.
- [10] Kececi A, Yıldırak A, Özyazıcı K, Ayluctarhan G, Ağbulut O, Zincir I. (2020). “Implementation of machine learning algorithms for gait recognition”. Engineering Science and Technology, an International Journal, 23(4), 931–937.
- [11] Ali M, Kumar BP, Ahmad KM, Francis B, Julian MWQ, Moni MA. (2021). “Heart disease prediction using supervised machine learning algorithms: performance analysis and comparison”. Computers in Biology and Medicine, 136 (2021), 1-10.
- [12] Droomer M, Bekker J. (2020). “Usıng machine learning to predict the next purchase date for an individual retail customer”. South African Journal of Industrial Engineering, 31(3), 69-82.
- [13] Yan W, Shao H. (10-14 June 2002) “Application of support vector machine nonlinear classifier to fault diagnoses”. Proceedings of the 4rd World Congress on Intelligent Control and Automation, Shanghai, China.
- [14] Kayaalp K. (2007). Fault Detection in Induction Motors Using Data Mining. MSc Thesis, Suleyman Demirel University, Isparta, Turkey.
- [15] Şanlıtürk E. (2018). Prediction of Defective Product with Machine Learning Algorithms. MSc Thesis, Istanbul Technical University, Istanbul, Turkey.
- [16] Fourie CJ, Plessis JA. (2020). “Implementatıon of machine learning techniques for prognostics for railway wheel flange wear”. South African Journal of Industrial Engineering, 31(1), 78-92.
- [17] Karadağ G. (2018). Prediction of Production Wastage Via Data Mining. MSc Thesis, Yasar University, Izmir, Turkey.
- [18] Zhang X, Kano M, Tani M, Mori J, Ise S, Harada K. (2020). “Prediction and causal analysis of defects in steel products: Handling nonnegative and highly overdispersed count data”. Control Engineering Practice, 95(2020), 1-8.
- [19] Tobias G, Falco B, Robert M, Alexander V, Martyna B, Alexander D. (2020). “Evaluation of machine learning for sensorless detection and classification of faults in electromechanical drive systems”. Procedia Computer Science, 176(2020), 1586–1595.
Improvement of Quality Performance in Mask Production by Feature Selection and Machine Learning Methods and An Application
Abstract
With the development of technology, large databases become more accessible thanks to automation systems that automatically keep data and allow the use of large databases in many areas. Machine learning approaches, a sub-branch of artificial intelligence, are used in making decisions about the process by analyzing the data stored in databases and converting them into information. In this paper, the body production process of the surgical (medical) mask is analyzed. As it is known, surgical masks have become a part of our lives by becoming widespread all over the world with the COVID-19 pandemic. In the surgical mask body production process, using the real data of the production factors, first of all, filtering feature selection methods and analyzes were made and the feature selection method to be used was determined. With the specified feature selection method, the factors affecting the product quality are determined. Secondly, machine learning methods were used to determine the values and value ranges of factors (features) in the production of defect-free products. The performances of the machine learning models established in the second stage were increased by feature selection analysis. In the study, together with the parameter optimizations made to machine learning algorithms, it was seen that the best algorithm to estimate the defective product rate was the Ibk algorithm with 92.3% accuracy, 91.9% F measurement and 93% AUC value. Finally, in line with the decision rules revealed in the study, it was observed that the fabric types used for the upper/middle/lower layers that make up the body part in the mask body production process greatly affect the rates of defective or defect-free products. If the rod apparatus around the nose belongs to class k, it has been determined that many masks are defective. Improvement suggestions were presented according to the application results.
Keywords
Machine Learning Feature Selection Classification Defective and Defect-Free Product Prediction Filters Methods
References
- [1] Öztemel E. (2016). Artificial Neural Networks. 4nd ed. Istanbul, Turkey, Papatya Press.
- [2] Doğan A, Birant D. (2021). “Machine learning and data mining in manufacturing”. Expert Systems with Applications, 166(2021), 1-22.
- [3] Öztemel E, Gürsev S. (2018). “Literature review of industry 4.0 and related technologies”. Journal of Intelligent Manufacturing, 31(2020), 127-182.
- [4] Cho E, Jun J, Chang T, Choi Y. (2020). “Quality prediction modeling of plastic extrusion process”. ICIC Express Letters Part B: Applications, 11(5), 447–452.
- [5] Chen K, Hu YH, Hsieh YC. (2015). “Predicting customer churn from valuable B2B customers in the logistics industry: A case study”. Information Systems and E-Business Management, 13(2015), 475–494.
- [6] Brillinger M, Wuwer M, Hadi MU, Haas F. (2021). “Energy prediction for CNC machining with machine learning”. CIRP Journal of Manufacturing Science and Technology, 35(2021), 715-723.
- [7] Miguéis V, Freitas LA, Garcia PJV, Silva A. (2018). “Early segmentation of students according to their academic performance: A predictive modelling approach”. Decision Support Systems, 115(2018), 36–51.
- [8] Go A, Bhayani R, Huang L. (2019). “Twitter sentiment Classification Using Distant Supervision”. Stanford, United States of America, Project Report, CS224N.
- [9] Yucalar, F., Özçift, A., Borandağ, E., & Kılınç, D. (2020). Yazılım kalitesi mühendisliğinde çoklu sınıflandırıcılar: Yazılım hata tahmin yeteneğini geliştirmek için tahmin edicileri birleştirmek. Engineering Science and Technology, International Journal, 23 (4), 938-950.
- [10] Kececi A, Yıldırak A, Özyazıcı K, Ayluctarhan G, Ağbulut O, Zincir I. (2020). “Implementation of machine learning algorithms for gait recognition”. Engineering Science and Technology, an International Journal, 23(4), 931–937.
- [11] Ali M, Kumar BP, Ahmad KM, Francis B, Julian MWQ, Moni MA. (2021). “Heart disease prediction using supervised machine learning algorithms: performance analysis and comparison”. Computers in Biology and Medicine, 136 (2021), 1-10.
- [12] Droomer M, Bekker J. (2020). “Usıng machine learning to predict the next purchase date for an individual retail customer”. South African Journal of Industrial Engineering, 31(3), 69-82.
- [13] Yan W, Shao H. (10-14 June 2002) “Application of support vector machine nonlinear classifier to fault diagnoses”. Proceedings of the 4rd World Congress on Intelligent Control and Automation, Shanghai, China.
- [14] Kayaalp K. (2007). Fault Detection in Induction Motors Using Data Mining. MSc Thesis, Suleyman Demirel University, Isparta, Turkey.
- [15] Şanlıtürk E. (2018). Prediction of Defective Product with Machine Learning Algorithms. MSc Thesis, Istanbul Technical University, Istanbul, Turkey.
- [16] Fourie CJ, Plessis JA. (2020). “Implementatıon of machine learning techniques for prognostics for railway wheel flange wear”. South African Journal of Industrial Engineering, 31(1), 78-92.
- [17] Karadağ G. (2018). Prediction of Production Wastage Via Data Mining. MSc Thesis, Yasar University, Izmir, Turkey.
- [18] Zhang X, Kano M, Tani M, Mori J, Ise S, Harada K. (2020). “Prediction and causal analysis of defects in steel products: Handling nonnegative and highly overdispersed count data”. Control Engineering Practice, 95(2020), 1-8.
- [19] Tobias G, Falco B, Robert M, Alexander V, Martyna B, Alexander D. (2020). “Evaluation of machine learning for sensorless detection and classification of faults in electromechanical drive systems”. Procedia Computer Science, 176(2020), 1586–1595.
Details
| Primary Language | English |
|---|---|
| Subjects | Engineering |
| Journal Section | Articles |
| Authors | |
| Publication Date | May 7, 2024 |
| Submission Date | May 17, 2023 |
| Published in Issue | Year 2024 Volume: 20 Issue: 1 |
