Characterization of Superalloys by Artificial Neural Network Method

In this study, the use of artificial neural networks in the classification of a superalloys whose chemical analysis is performedin the quality process is investigated. In general, chemical spectro analysis method alone is not sufficient to determine which class a steelbelongs to. In addition to the chemical analysis method, tests such as tensile test, hardness test or notch impact test are applied. The testsperformed in addition to the chemical analysis both take time and destroy the material. The fact that an algorithm that classifies steelonly according to the results of chemical analysis is not used has made destructive tests mandatory. Artificial neural networks (ANNs),usually simply called neural networks (NNs), are computing systems inspired by the biological neural networks that constitute animalbrains. An ANN is based on a collection of connected units or nodes called artificial neurons, which loosely model the neurons in abiological brain. Each connection, like the synapses in a biological brain, can transmit a signal to other neurons. In our study, a total of34 superalloy materials belonging to 6 different classes were used. Chemical composition values were determined for each superalloysample. The appropriate artificial neural network model was determined according to the chemical composition values. A model thatcan predict superalloy material based on chemical composition value has been created. Weka 3.9.5 package program was used to createthe artificial neural network model. The high success rate of the prediction model gave hope for the determination of the material classonly with the chemical analysis method.