Formlu Keçelerin Akustik Performansı için Makine Öğrenimi Tabanlı Tahmin Modeli

Öz

Otomotiv endüstrisinde artan konfor beklentileriyle birlikte akustik performans kritik bir odak noktası haline gelmiştir. Bu doğrultuda akustik beklentilere uygun malzemeler sunan, hızlı ve doğru bir şekilde ölçüm yapabilen tedarikçiler sektörde ön plana çıkmaktadır. Bu çalışmada, ses yalıtımı amacıyla kullanılan keçe malzemelerinin ses yutum katsayısını tahmin etmek için üç farklı makine öğrenmesi algoritmasının (doğrusal regresyon, rastgele orman, k-en yakın komşu) performansı karşılaştırılmıştır. Çalışma kapsamında hazır bir veri seti kullanılmamış olup, tüm veriler epoksi keçe malzeme ile hazırlanan numuneler üzerinden gerçekleştirilen akustik performans ölçümlerinden elde edilmiştir. Model performansları ise eğitim ve test veri setlerinden elde edilen çeşitli performans metrikleri ile değerlendirilmiştir. Sonuçlar, rastgele orman algoritmasının diğer algoritmalara kıyasla daha iyi bir performans sergilediğini göstermektedir. Sonuç olarak, bu çalışma, makine öğrenmesi tekniklerinin akustik performansı hızlı ve doğru bir şekilde tahmin etme potansiyelini vurgulamaktadır. Empedans tüpü testi, alfa kabin testi gibi ampirik yöntemler ve analitik yöntemlerle karşılaştırıldığında, makine öğrenmesi temelli tahmin modellerinin zaman, iş gücü ve maliyet açısından sağladığı avantajlar, bu teknolojinin endüstriyel süreçlerdeki önemini daha belirgin hale getirmektedir.

Anahtar Kelimeler

• Makine öğrenmesi
• ses yutum katsayısı
• lineer regresyon
• k-en yakın komşu
• rastgele orman

Machine Learning Based Prediction Model for Acoustic Performance of Formed Felts

Öz

The automotive industry has an increasing focus on comfort expectations, with acoustic performance becoming a critical focus point. In this regard, suppliers offering materials that meet acoustic expectations and can facilitate rapid and precise measurements are becoming increasingly prominent in the sector. The present study aims to contribute to this emerging field by comparing the performance of three distinct machine learning algorithms – linear regression, random forest, k-nearest neighbour– in predicting the sound absorption coefficient of felt materials employed for sound insulation. It is important to note that the study did not utilise any pre-existing data sets; all data were obtained from acoustic performance measurements performed on samples prepared with epoxy felt material. The performance of each model was evaluated using a range of performance metrics derived from both the training and test data sets. The findings of this study demonstrate that the random forest algorithm exhibits superior performance compared to the other algorithms. The study thus demonstrates the potential of machine learning techniques to predict acoustic performance quickly and accurately. Compared to conventional methods such as impedance tube testing, alpha cabinet testing, and analytical methods, the machine learning-based prediction models demonstrate significant advantages in terms of reduced time, labour, and cost, thereby underscoring the significance of this technology in industrial processes.

Anahtar Kelimeler

• Machine learning
• sound absorption coefficient
• linear regression
• k-nearest neighbour
• random forest

Kaynakça

1. [1] Yıldız, B. İ. ve Karabağ, K., Prediction of beef production using linear regression, random forest and k-nearest neighbors algorithms, Kahramanmaraş Sütçü İmam Üniversitesi Tarım ve Doğa Dergisi, 28(1), 247-255, (2025).
2. [2] Ahmed, M. U. ve Hussain, I., Prediction of wheat production using machine learning algorithms in northern areas of Pakistan, Telecommunications Policy, 46(6), 102370, (2022).
3. [3] Yuan, B., You, T., Jiang, H., Qian, H., Cao, L. ve Zhu, R., Prediction of the sound absorption performance for micro-perforated panel based on machine learning, Journal of Engineering and Applied Science, 72(1), 27, (2025).
4. [4] Wang, R., Yao, D., Zhang, J., Xiao, X. ve Xu, Z., Identification of key factors ınfluencing sound ınsulation performance of high-speed train composite floor based on machine learning, Acoustics, Vol. 6, No. 1, pp. 1-17, (2023).
5. [5] Sasaki, S., Suganuma, S. and Aramaki, T., Development of a low-noise wind tunnel and prediction of broaSadband noise from a flat plate based on machine learning, Nagasaki University Graduate School of Engineering Research Reports, 53 (101), 1-7, (2023).
6. [6] Xu, J., Mao, W., Yu, X., and Liu, B., Prediction and optimization of acoustic absorption performance of quasi-Helmholtz acoustic metamaterials based on LightGBM algorithm, AIP Advances, 15(1), (2025).
7. [7] Zhang, Z., Ye, R., Zhu, G., Huang, X., Qian, Y. And Yurchenko, D., Research on sound absorption performance prediction of microporous structures based on integrated learning, Materials Today Communications, 46, 112866, (2025).
8. [8] Das, D., Pradhan, A. K., Chattopadhyay, R. ve Singh, S. N. Composite nonwovens, Textile Progress, 44(1), 1-84, (2012).

9. [9] Xue, B., Xie, L., Bao, Y. ve Zhang, J., Multilayered epoxy/glass fiber felt composites with excellently acoustical and thermal insulation properties, Journal of Applied Polymer Science, 136(3), 46935, (2019).
10. [10] Polack, J. D., Revisiting reverberation formulae, In INTER-NOISE and NOISE-CON Congress and Conference Proceedings, (Vol. 270, No. 8, pp. 3497-3504), Institute of Noise Control Engineering, (2024).
11. [11] Peng, L. Sound absorption and insulation functional composites. In Advanced high strength natural fibre composites in construction, (pp. 333-373), Woodhead Publishing, (2017).
12. [12] Türkmenoglu, C. ve Tantug, A. C., Sentiment analysis in Turkish media, In International Conference on Machine Learning, Beijing, (2014)
13. [13] McCulloch, W. S. ve Pitts, W., A logical calculus of the ideas immanent in nervous activity. The Bulletin of Mathematical Biophysics, 5, 115-133, (1943).
14. [14] Korkmaz, G. ve Eroğlu, E., Model karmaşıklığının kontrolü, İktisadi ve İdari Yaklaşımlar Dergisi, 2(2), 146-162, (2020).
15. [15] Kılıç, S., Doğrusal regresyon analizi, Journal of Mood Disorders, 3(2), 90-92, (2013).
16. [16] Tonta Y., Regresyon analizi, Ders Notu, Hacettepe Üniversitesi, Ankara, (2008).
17. [17] Breiman L., Random forests, Machine Learning, 45, 5-32, (2001).
18. [18] What is random forest? https://www.ibm.com/think/topics/random-forest, (01.04.2025).
19. [19] Altman, N. ve Krzywinski, M., Ensemble methods: bagging and random forests, Nature Methods, 14(10), 933-935, (2017).
20. [20] Taunk, K., De, S., Verma, S. ve Swetapadma, A. A brief review of nearest neighbor algorithm for learning and classification. International Conference on Intelligent Computing and Control Systems, p. 1255-1260, Madurai, (2019).
21. [21] Taşcı, E. ve Onan, A., K-en yakın komşu algoritması parametrelerinin sınıflandırma performansı üzerine etkisinin incelenmesi, Akademik Bilişim, 1(1), 4-18. (2016).
22. [22] Kızılkaya, Y. M. ve Oğuzlar, A., Bazı denetimli öğrenme algoritmalarının R programlama dili ile kıyaslanması, Karadeniz Uluslararası Bilimsel Dergi, 37(37), 90-98. (2018).
23. [23] Figueiredo Filho, D. B., Júnior, J. A. S. ve Rocha, E. C., What is R2 all about?, Leviathan (São Paulo), (3), 60-68. (2011).
24. [24] Varoquaux, G. ve Colliot, O., Evaluating machine learning models and their diagnostic value, Machine Learning for Brain Disorders, 601-630. (2023).
25. [25] Chicco, D., Warrens, M. J., ve Jurman, G., The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation, Peerj Computer Science, 7, e623. (2021).
26. [26] Bui, D. T., Khosravi, K., Tiefenbacher, J., Nguyen, H. ve Kazakis, N., Improving prediction of water quality indices using novel hybrid machine-learning algorithms, Science of the Total Environment, 721, 137612, (2020).
27. [27] Prayudani, S., Hizriadi, A., Lase, Y. Y. ve Fatmi, Y., Analysis accuracy of forecasting measurement technique on random K-nearest neighbor (RKNN) using MAPE and MSE, In Journal of Physics: Conference Series, Vol. 1361, No. 1, p. 012089, (2019).
28. [28] Flom, P., The disadvantages of linear regression (2018) https://www.sciencing.com/disadvantages-linear-regression-8562780/, (2.04.2025).
29. [29] Yuan, Z. X., Acoustic properties of multilayered structures, Acoustics Australia, 48(3), 395-405, (2020).
30. [30] Qui, H., ve Enhui, Y., Effect of thickness, density and cavity depth on the sound absorption properties of wool boards, Autex Research Journal, 18(2), 203-208, (2018).