Yapay Zekâ Uygulamasıyla Seramik Nesnelerin Şekilsel Deformasyonun İncelenmesi: ANFIS

 Seramik endüstrisinde yeni ürünlerin geliştirilmesi ve tasarlanmasında uygun bir formu elde etmek için birçok ilk örnek hazırlamak gereklidir. Çok sayıda deneme üretim sürecinde deformasyona bağlı olarak artan maliyet ve işgücü kaybına yol açmaktadır. Bu çalışmada, seramik endüstrisindeki bu kayıpları azaltmak için yazılım kullanarak bir yapay zekâ modeli geliştirilmesi amaçlanmıştır. Seramik silindirik nesnelerin deformasyonunu araştırmak için, farklı kimyasal kompozisyonlarda, sinterleme sıcaklıklarında ve sinterleme sürelerinde seramik nesneler üretilmiştir. İlk olarak, silindirik seramik objeler içi boş bir forma dökülmüştür. Bu numunelerin suyunun buharlaştırılmasından sonra, numuneler klasik yöntem kullanılarak geometrik ölçümler (taban, yan ve ağız bölgelerinde) yapıldı. Daha sonra, bu numuneler farklı sinterleme sürelerinde ve seramik fırındaki sıcaklıklarda pişirildi. Silindirik seramik örneklerin taban, yan ve ağız bölgelerindeki deformasyonlar daha sonra klasik yöntem kullanılarak yeniden ölçüldü. Bu deneysel sonuçlardan elde edilen verilerden, MatLab Toolbox kullanılarak ANFIS (Adaptive Neuro Fuzzy Inference System) modeli geliştirildi. Geliştirilen ANFIS modelinde sıcaklık, sinterleme süresi ve seramik örneklerin kimyasal bileşimi giriş parametreleri olarak belirlenirken, deformasyon miktarı çıktı parametreleri olarak belirlendi. Elli sekiz deneyin sonuçları gelişmekte olan modelin eğitimi için kullanılırken, yirmi iki deneyin sonuçları geliştirilen modelin testi için kullanıldı. ANFIS model sonuçları ile deneysel sonuçlar arasındaki ilişkiyi X2 testi ile karşılaştırdık ve anlamlı bir ilişki bulunmuştur (p <0.001 ve sırasıyla, taban ve yan için κ = 0.3, 0.3). Fakat ANFIS model sonuçları ile ağız deformasyonu için deneysel sonuçlar arasında anlamlı ilişki bulunamamıştır (κ = 0.06).

Examining the Formal Deformation of Ceramic Objects by Artificial Intelligence Application: ANFIS

It is necessary to prepare many prototypes in order to obtain a suitable form in the development and design of new products in the ceramic industry. Numerous trials have led to increased cost and labor loss due to deformation in the production process. In this study is aimed to develop an artificial intelligence model by using software to reduce these losses in the ceramic industry. In order to investigate deformation of ceramic cylindrical objects, ceramic objects were produced in different chemical compositions, sintering temperature and sintering time. Initially, the cylindrical ceramic objects were poured into a hollow form. After evaporation of the water of these samples, this samples were scanned by using the geometric measurements (in the base, side and mouth regions) were made using classical method. Later, these samples were fired at different sintering times and temperatures in ceramic kiln. The deformations in the base, side and mouth regions of the cylindrical ceramic samples are then re-measured by using classical method. By using the data obtained from these experimental results, ANFIS (Adaptive Neuro Fuzzy Inference System) model was developed by using MatLab Toolbox. While the temperature, sintering time and composition of ceramic specimens are determined as input parameters in the developed ANFIS model, the amount of deformation is determined as output parameters. While the results of the Fifty-eight experiments were used for the training of the developing model, while the results of twenty-two experiments were used for the test of the developed model.   We compared relation between ANFIS model results and experimental results with X2 test and founded a significant correlation (p<0.001 and for base and side κ=0.3, 0.3, respectively).  But it is not found significant relationship between ANFIS model results and experimental results for mouth deformation (κ=0.06).  

___

  • Wikipedia. Deformation (Engineering). 2016 February, 2016; Available from: https://www.wikizero.pro/index.php?q=aHR0cHM6Ly9lbi53aWtpcGVkaWEub3JnL3dpa2kvRGVmb3JtYXRpb24.
  • HKMO. FİG. 2016 February; Available from: http://www.hkmo.org.tr/resimler/ekler/G14C_940ab4746839656_ek.pdf.
  • Jančíková, Z., et al., Artificial Neural Network Modelling of Glass Laminate Sample Shape Influence on the ESPI Modes. 2012. 61-69.
  • Fukuda, T. and O. Hasegawa. 3-D image processing and grasping planning expert system for distorted objects. in 15th Annual Conference of IEEE Industrial Electronics Society. 1989.
  • Xianming, L., X. Yougang, and L. Xuejun, Shell Construction Optimisation of Large Rotary Kiln with Milti-Supports. Guisuanyan Xuebao(Journal of the Chinese Ceramic Society), 2006. 34(2): p. 215-219.
  • Jančíková, Z., et al., Prediction of internal defects in rolled products from cr-mo steels using artificial intelligence methods. 2013.
  • Ishimaru, I., H. Hirabayashi, and M. Morisato. A proposal of probability type fuzzy inference and its application to precision plastic deformation algorithm. in IEEE SMC'99 Conference Proceedings. 1999 IEEE International Conference on Systems, Man, and Cybernetics (Cat. No.99CH37028). 1999.
  • Pataro, C. and H. Helman. Direct determination of sequences of passes for the strip rolling process by means of fuzzy logic rules. in Intelligent Processing and Manufacturing of Materials, 1999. IPMM'99. Proceedings of the Second International Conference on. 1999. IEEE.
  • Tsutsumi, K., et al. A study on the estimation method of target design engineering quantity considering client satisfaction. in IFSA World Congress and 20th NAFIPS International Conference, 2001. Joint 9th. 2001. IEEE.
  • Bielen, J., J.-J. Gommans, and F. Theunis. Prediction of high cycle fatigue in aluminum bond wires: A physics of failure approach combining experiments and multi-physics simulations. in Thermal, Mechanical and Multiphysics Simulation and Experiments in Micro-Electronics and Micro-Systems, 2006. EuroSime 2006. 7th International Conference on. 2006. IEEE.
  • Dinh, N.Q. and N.V. Afzulpurkar, Neuro-fuzzy MIMO nonlinear control for ceramic roller kiln. Simulation Modelling Practice and Theory, 2007. 15(10): p. 1239-1258.
  • Nanayakkara, N.D. and J. Samarabandu. Unsupervised model based image segmentation using domain knowledge based fuzzy logic and edge enhancement. in icme. 2003. IEEE.
  • Aali, K.A., M. Parsinejad, and B. Rahmani, Estimation of saturation percentage of soil using multiple regression, ANN, and ANFIS techniques. Computer and Information Science, 2009. 2(3): p. 127.
  • Jang, J.R. Input selection for ANFIS learning. in Proceedings of IEEE 5th International Fuzzy Systems. 1996.
  • Gulbandilar, E. and Y. Kocak, Application of expert systems in prediction of flexural strength of cement mortars. Computers and Concrete, 2016. 18(1): p. 1-16.