ISIL İŞLEM GÖRMÜŞ SEDİR ODUNU DARALMA VE GENİŞLEME DEĞERLERİNİN YAPAY SİNİR AĞLARI VE RASTGELE ORMAN ALGORİTMASI İLE TAHMİNİ

Abstract

Ahşap malzeme, hem yapısal hem de yapısal olmayan uygulamalarda kullanılabilen doğal, sürdürülebilir, yenilenebilir ve çevre dostu bir malzemedir. Ancak ahşap malzemenin en önemli olumsuz özelliklerinden biri higroskopik bir malzeme olmasıdır. Isıl işlem uygulaması ahşap malzemenin boyutsal stabilitesini arttırmakta ve daha hidrofobik hale getirmektedir. Bu çalışmada öncelikle, Sedir odununun temas açısı değerleri, ahşap malzeme yüzeyine damlatma ile teğet ve radyal yönde belirlenmiştir. Daha sonra aynı numunelerin genişleme ve daralma miktarları belirlenmiştir. Genişleme ve daralma miktarlarının belirlenmesinde TS 4084 standardı kullanılmıştır. Deneysel çalışma sonucunda, yapay sinir ağı (ANN) ve rastgele orman algoritması kullanılarak örneklerin daralma ve genişleme miktarları tahmin edilmiştir. Rastgele orman ve ANN yöntemleri ile yapılan tahminlerde temas açısı değerleri girdi olarak kullanılmıştır. Rastgele Orman Algoritması ile yapılan tahminlerin en doğru sonuçları verdiği tespit edilmiştir (teğet yön, R2= 0.91, radyal yön, R2= 0.97). Sonuç olarak, temas açısı değerleri bilinen bir ahşap malzemenin Rastgele Orman Algoritması ile daralma ve genişleme değerlerinin daha iyi tahmin edilebileceği belirlenmiştir.

Keywords

• Isıl işlem
• Sedir
• Tahmin
• Yapay Sinir Ağları
• Rastgele Orman Algoritması

PREDICTION OF HEAT-TREATED CEDAR WOOD SWELLING AND SHRINKAGE VALUES WITH ARTIFICIAL NEURAL NETWORKS AND RANDOM FOREST ALGORITHM

Abstract

Wood material is a natural, sustainable, renewable and environmentally friendly material that can be used in both structural and non-structural applications. However, one of the most important negative features of wood material is that it is a hygroscopic material. Heat treatment application increase dimensional stability of the wood material and becomes more hydrophobic. In this study, firstly, the contact angle values of Cedar wood have been determined in the tangential and radial direction by dropping them on the surface of the wood material. Then the swelling and shrinkage amounts of the same samples were determined. TS 4084 standard was used to determine the swelling and shrinkage amounts. As a result, shrinkage and swelling amounts of the samples were estimated by using artificial neural network (ANN) and Random Forest (RF) algorithm. In the estimation made by RF and ANN methods, contact angle values were used as input. It has been determined that the predictions made with RF Algorithm give the most accurate results (tangential direction, R2= 0.91, radial direction, R2= 0.97). As a result, it has been determined by RF Algorithm that shrinkage and swelling values of a wood material whose con-tact angle values are known can be better predicted.

Keywords

• Heat Treatment
• Cedar
• Prediction
• Artificial Neural Networks
• Random Forest Algorithm

References

1. Belgiu, M., Drăguţ, L., 2016. Random Forest in Remote Sensing: A Review of Applications and Future Directions. ISPRS Journal of Photogrammetry and Remote Sensing, 114: 24-31.
2. Brunetti M., Cremonini C., Crivellaro A., Feci E., Palanti S., Pizzo B., Santoni I., Zanuttini R., 2007. Thermal Treatment of Hardwood Species from Italian Plantations: Preliminary Studies on Some Effects on Technological Properties of Wood. alin: International Scientific Conference on Hardwood Processing, Que´bec City, 24–26 Sep. Available at http://www.ischp.ca/FR/pdf/ISCHP_proceedings.pdf. Accessed 26 Feb 2009.
3. Cengiz, O., 2010. Design Of Contact Angle Meter. University of İstanbul Technical, Mechanical Engineering Department, Master Thesis: 83s.
4. Esteves, B., Marques A.V., Domingos I. and Pereira, H., 2007. Influence of Steam Heating on the Properties of Pine (Pinus pinaster) and Eucalypt (Eucalyptus globulus) Wood. Wood Sci Technol 41(3): 193–207.
5. Garcia, R.A., Riedl, B. and Cloutier, A., 2008. Chemical Modification and Wetting of Medium Density Fibreboard Produced from Heat-Treated Fibres. J Mater Sci 43: 5037–5044.
6. Ghosh-Dastidar, S. and Adeli, H., 2009. A New Supervised Learning Algorithm for Multiple Spiking Neural Networks with Application in Epilepsy and Seizure Detection. Neural networks, 22(10): 1419-1431.
7. Kamdem, D.P., Pizzi, A. and Jermannaud, A., 2002. Durability of Heat-Treated Wood. Holz Roh Werkst 60: 1–6.
8. Kilincarslan, S. and Simşek Türker, Y., 2019. ‘’The Effect of Strengthening With Fiber Reinforced Polymers on Strength Properties of Wood Beams’’, 2nd International Turkish World Engineering and Science Congress, November 7-10, Turkey.

9. Kilincarslan, S. and Simsek Türker, Y., 2019. Determination of Contact Angle Values of Heat-treated Spruce (Picea abies) Wood with Image Analysis Program. Biomedical Journal of Scientific & Technical Research, 18(4), 13750-13751 (2019).
10. Kilincarslan, S. and Simşek Türker, Y., 2020a. Physical-Mechanical Properties Variation with Strengthening Polymers, Acta Physica Polonica A, 137(4): 566-568.
11. Kilincarslan, S. and Simşek Türker, Y., 2020b. The Effect Of Heat Treatment Application on Wettability Properties of Wood Materials, Journal of Engineering Sciences and Design, 8(2): 460 – 466.
12. Kilincarslan, S. and Simşek Türker, Y., İnce, M. 2020. Prediction Using Different Classification Methods of Tree Species Depending on Contact Angle Values, Journal of Bartin Faculty of Forestry, 22 (3): 861-870.
13. Kocaefe, Poncsak, Dor´e, Younsi., 2008. Effect of Heat Treatment on The Wettability Of white Ash and Softmaple By Water. Holz Roh Werkst, 66: 355–361.
14. Lahouar, A. and Slama, J. B. H., 2015. Day-Ahead Load Forecast Using Random Forest and Expert Input Selection. Energy Conversion and Management, 103: 1040-1051.
15. Li, C., Sanchez, R. V., Zurita, G., Cerrada, M., Cabrera, D. and Vásquez, R. E., 2016. Gearbox Fault Diagnosis Based on Deep Random Forest Fusion of Acoustic and Vibratory Signals. Mechanical Systems and Signal Processing, 76: 283-293.
16. Neumann, A.W. and Spelt, J.K., (eds) 1996. Applied Surface Thermodynamics (Surfactant series v. 63). Marcel Dekker Inc, New York
17. Nevitt, J. and Hancock, G. R., 2000. Improving The Root Mean Square Error of Approximation for Nonnormal Conditions in Structural Equation Modeling. The Journal of Experimental Education, 68(3): 251-268.
18. Recchia, A., 2010. R-Squared Measures for Two-Level Hierarchical Linear Models Using SAS. Journal of Statistical Software, 32(2): 1-9.
19. Rodriguez-Galiano, V., Sanchez-Castillo, M., Chica-Olmo, M. and Chica-Rivas, M. J. O. G. R., 2015. Machine Learning Predictive Models for Mineral Prospectivity: An Evaluation of Neural Networks, Random Forest, Regression Trees and Support Vector Machines. Ore Geology Reviews, 71: 804-818.
20. Sahin, H.T., Arslan, M.B., Korkut, S., Sahin C., 2011. Colour Changes of Heat‐Treated Woods of Red‐Bud Maple, European Hophornbeam And Oak. Color Research & Application, 36(6):462-466.
21. Sahin, C. K., Onay, B., 2020. Alternatıve Wood Species For Playgrounds Wood From Fruit Trees. Wood Research, 65(1), 149-160.
22. Sahin, C., Topay, M., Var, A.A., 2020. A Study on Some Wood Species For Landscape Applications: Surface Color, Hardness And Roughness Changes at Outdoor Conditions. Wood Research, 65(3): 395-404.
23. Sanger, T. D., 1989. Optimal Unsupervised Learning in a Single-Layer Linear Feedforward Neural Network. Neural networks, 2(6): 459-473.
24. Sasakawa, T., Hu, J. and Hirasawa, K. 2008. A Brainlike Learning System with Supervised, Unsupervised, and Reinforcement Learning. Electrical Engineering in Japan, 162(1): 32-39.
25. Shi, Q., Gardner, D.J. and Wang, J.Z., 1997. Surface Properties of Polymeric Automobile Fluff Particles Characterized by Inverse Gas Chromatography and Contact Angle Analysis. In: Int. Conf. of Woodfiber-Plast. Compos. 4th Forest Product Society, Madison, USA, pp: 245–256.
26. Shrestha, D. L. and Solomatine, D. P., 2006. Machine Learning Approaches for Estimation of Prediction Interval for the Model Output. Neural Networks, 19(2): 225-235.
27. Singh, S., Jaakkola, T., Littman, M. L. and Szepesvári, C., 2000. Convergence Results for Single-Step on-Policy Reinforcement-Learning Algorithms. Machine learning, 38(3): 287-308.
28. Suat, A. Y. A. N. and Ciritcioğlu, H. H., 2012. Determination of Heat Treatment Effect on Some Mechanical Properties and Screw Withdrawal Strength of Laminated Wood Panels, Journal of Advanced Technology Sciences, 1(1): 35-46.
29. Unsal, O. and Ayrilmis, N., 2005. Variations in Compression Strength and Surface Roughness of Heat-Treated Turkish River Red Gum (Eucalyptus camaldulensis) Wood. J Wood Sci 51: 405–409.
30. Van Gerven, M. and Bohte, S., 2017. Artificial Neural Networks as Models of Neural Information Processing. Frontiers in Computational Neuroscience, 11: 114.
31. Vitorino, D., Coelho, S. T., Santos, P., Sheets, S., Jurkovac, B. and Amado, C., 2014. A Random Forest Algorithm Applied to Condition-Based Wastewater Deterioration Modeling and Forecasting. Procedia Engineering, 89: 401-410.
32. Walinder, M.E.P. and Johansson, I., 2001. Measurement of Wood Wettability by the Wilhelmy Method. Holzforschung 1(55): 21–32.
33. Walinder, M.E.P. and Strom, G., 2001. Measurement of Wood Wettability by the Wilhelmy Method. Holzforschung 2(55): 33–41.
34. Willmott, C. J. and Matsuura, K., 2005. Advantages of The Mean Absolute Error (MAE) Over The Root Mean Square Error (RMSE) in Assessing Average Model Performance. Climate Research, 30(1): 79-82.
35. Zhang, G., Patuwo, B. E. and Hu, M. Y., 1998. Forecasting with Artificial Neural Networks: The State of the art. International Journal of Forecasting, 14(1): 35-62.