Application of Meta-Heuristic Hybrid Artificial Intelligence Techniques for Modeling of Bonding Strength of Plywood Panels

Öz

Plywood, which is one of the most important wood based panels, has many usage areas changing from traffic signs to building constructions in many countries. It is known that the high quality plywood panel manufacturing has been achieved with a good bonding under the optimum pressure conditions depending on adhesive type. This is a study of determining the using possibilities of modern meta-heuristic hybrid artificial intelligence techniques such as IKE and AANN methods for prediction of bonding strength of plywood panels. This study has composed of two main parts as experimental and analytical. Scots pine, maritime pine and European black pine logs were used as wood species. The pine veneers peeled at 32°C and 50°C were dried at 110°C, 140°C and 160°C temperatures. Phenol formaldehyde and melamine urea formaldehyde resins were used as adhesive types. EN 314-1 standard was used to determine the bonding shear strength values of plywood panels in experimental part of this study. Then the intuitive k-nearest neighbor estimator (IKE) and adaptive artificial neural network (AANN) were used to estimate bonding strength of plywood panels. The best estimation performance was obtained from MA metric for k-value=10. The most effective factor on bonding strength was determined as adhesive type. Error rates were determined less than 5% for both of the IKE and AANN. It may be recommended that proposed methods could be used in applying to estimation of bonding strength values of plywood panels.

Anahtar Kelimeler

• Adaptive artificial neural network (AANN); Bonding Strength; Intuitive k-nearest neighbor estimator (IKE); Plywood

Kontrplaklarda Yapışma Direnci Modellenmesinde Meta-Buluşsal Yapay Sinir Ağları Tekniklerinin Kullanılması

Öz

En önemli ahşap kökenli levha ürünlerinden biri olan kontrplak trafik levhalarından inşaata kadar pek çok kullanım yerine sahiptir. Yüksek kalitede kontrplak üretimi için tutkal türüne bağlı olarak optimum pres koşulları altında iyi bir yapışmanın sağlanması gerektiği bilinen bir gerçektir. Bu çalışmada kontrplağın yapışma direncinin tahmin edilmesi için modern meta buluşsal tekniklerden IKE ve AANN metotlarının kullanım imkanları araştırılmıştır. Çalışma deneysel ve analitik olarak iki kısımdan oluşmaktadır. Çalışmada ağaç türü olarak sarıçam, sahil çamı ve karaçam kullanılmıştır. Kaplamalar 2 farklı sıcaklıkta (32°C ve 50°C) soyulmuş ve 3 farklı sıcaklıkta (110°C, 140°C ve 160°C) kurutulmuştur. Kontrplak üretimi için fenol formaldehit ve melamin üre formaldehit tutkalları olmak üzere iki farklı tutkal türü kullanılmıştır. Deneysel olarak kontrplakların yapışma direnci değerleri EN 314-1 standardına göre yapılmıştır. IKE ve AANN teknikleri analitik olarak yapışma direnci tahmininde kullanılmışlardır. En iyi tahmin performansı k değeri 10 için elde edilmiştir. Yapışma direnci üzerine en etkili faktör olarak tutkal türü belirlenmiştir. IKE ve AANN için belirlenen hata oranları %5’in altında bulunmuştur. Çalışma neticesinde uygulanan tekniklerin kontrpalklarda yapışma direnci tahmininde kullanılabilir oldukları tespit edilmiştir.

Anahtar Kelimeler

• Adaptif yapay sinir ağı (AANN), Yapışma Direnci, Sezgisel k en yakın komşu tahmincisi (IKE), Kontrplak

Kaynakça

1. Aksoy A, Iskender E, Kahraman HT (2012) Application of the Intuitive k-NN Estimator for Prediction of the Marshall Test (AstmD1559) Results For Asphalt Mixtures. Construction & Building Materials 34: 561-569.
2. Aydin I, Colakoglu G (2005) Formaldehyde Emission, Surface Roughness, and Some Properties of Plywood as Function of Veneer Drying Temperature. Drying Technology 23: 1107-117
3. Aydın I, Colakoglu G, Hiziroglu S (2006) Surface Characteristics of Spruce Veneers and Shear Strength of Plywood as a Function of Log Temperature in Peeling Process. International Journal of Solids and Structures 43: 6140-6147.
4. Babu GS, Suresh S (2013) Parkinson’s Disease Prediction Using Gene Expression – A Projection Based Learning Meta-Cognitive Neural Classifier Approach. Expert Systems with Applications 40: 1519–1529.
5. Bayindir R, Colak I, Sagiroglu S, Kahraman HT (2012) Application of Adaptive Artificial Neural Network Method to Model the Excitation Currents of Synchronous Motors. The 11th IEEE International Conference on Machine Learning Applications (ICMLA 2012), 12-15 Dec. 2012, Florida, USA, 2, 498-502.
6. Chow S, Chunsi KS (1979) Adhesion strength and wood failure relationship in wood-glue bonds. Mokuzai Gakkaishi 25(2): 125–31.
7. Christiansen AW (1990) How Overdrying Wood Reduces Its bonding to Phenol Formaldehyde Adhesives: A Critical Review of The Literature. Part I. Physical Responses. Wood and Fiber Science 22(4): 441-459.
8. Demirkır C (2012) Using Possibilities of Pine Species in Turkey for Structural Plywood Manufacturing, PhD Thesis, Karadeniz Technical University, The Graduate School of Natural and Applied Sciences, Forest Industrial Engineering Department.

9. EN 314-1 (1998) Plywood; Bonding Quality, Part 1: Test Methods, CEN, Brüssel.
10. Esteban LG, Fernandez FG, De Palacios P (2011) Prediction Of Plywood Bonding Quality Using An Artificial Neural Network. Holzforschung 65(2): 209-214.
11. Fernandez FG, De Palacios P, Esteban LG, Iruela AG, Rodrigo BG, Menasalvas E (2012) Prediction of MOR and MOE of Structural Plywood Board Using An Artificial Neural Network And Comparison With A Multivariate Regression Model. Composites: Part B. 43: 3528–3533. 1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97 101 105 109 113 117 121 125 129 133 137 141 145 149 153 157 161 165 169 173 177 181 185 189 193 197 Sample Number