Discrimination of Grey Cast Iron morphology using integrated SVM and ANN approaches

Abstract

The internal structure of Grey Cast Iron (GCI) and its microstructure determines the acceptance or rejection of several mechanical parts in the inspection process. This is based on the change of GCI mechanical properties due to the variation of its cooling rate. Visual inspection by metallurgical experts has been the approved method to assess GCI types. However, such method has always been subject to human error, biased categorization, lack of experience and variations in performance level. Even though several commercial software is available for such discrimination approaches, multiple flaws and defects are detected in the way it assesses samples. This research introduces a new software that is capable of distinguishing between GCI and other types of cast irons based on Support Vector Machines (SVM). Moreover, the software can identify the GCI types according to international standards using a well-trained Artificial Neural Network (ANN).