International Journal of Scientific & Engineering Research, Volume 5, Issue 3, March‐2014 ISSN 2229‐5518 IJSER © 2014 http://www.ijser.org Grey Cast Iron Categorization using Artificial Neural Network Amany Khaled, Mostafa Rostom A. Atia,Tarek Moussa Abstract—Grey cast iron (GCI) takes part in a wide range of applications in industry specially automotive one due to its unique properties like castability, machinability, low melting point and low cost as well. It’s used in manufacturing engine block, clutches, cylinder head, drum brakes, etc. The cooling rate of GCI affects its microstructure. Consequently, mechanical properties of GCI show strong deviation with the change in the texture of its microstructure. The main challenge with GCI is that surfaces are section sensitive i.e. cutting direction of sample gives different shape of microstructure. Although manual assessment to images gives accurate results, it's susceptible to human error, lack of experience and variation of the operators’ performance. Thus, automated image processing has a great contribution in this area. It reduces the amount of time required and increases the accuracy of extracted data. Since artificial neural networks (ANN) are always used in cases that are prone to uncertainty and decision making, software for image processing based on artificial neural networks will be introduced to categorize grey cast iron samples. Index Terms— grey cast iron, neural networks, image processing, GLCM, co-occurrence matrix, categorization, CLAHE ——————————  —————————— 1 INTRODUCTION y the late 1780s and early 1800s, within the start of the industrial revolution, several types of materials have be- come of great importance to the industry and therefore, their study became a real necessity. Grey cast iron (GCI) is an attractive material used in industrial applications due to its some advantageous properties such as good castability, corro- sion resistance, machinability, low melting point, low cost, and high damping capacity [1]. The microstructure of GCI is characterized by graphite flakes dispersed into the ferrous matrix. The amount of graphite, it's size, morphology and dis- tribution are critical in determining the mechanical behavior of GCI [2]. The widely used method to determine GCI me- chanical property based on its microstructure is manual visual inspection by a metallurgical expert. Manual assessment of microstructures is a time consuming task and unbiased deci- sion can’t be guaranteed. Thus, automation of GCI image analysis is highly recommended. 2 LITERATURE REVIEW Cast irons and cast steels make a huge family of ferrous al- loys[3]. The commercial production of cast iron in the west did not commence until the 13th century A.D., considerably later [4]. The main advantages of cast irons are their low price and ability to originate products of complex shapes, frequently in a single production step [5]. Therefore, These materials have been selected because they are commonly used by industries, as in, for example, structures of machines, lamination cylin- ders, main bodies of valves and pumps and gear elements [6]. Graphitic cast irons, including those that contain small amounts of alloying elements, are classified as gray, ductile, and malleable according to graphite shape and method of graphite production [7]. GCI is most commonly used materials out of all other cast irons. What makes cast irons applicable in industry is their microstructure under slow cooling. Graphite flakes found in GCI internal structure reduces the amount of volume reduction in case of slow cooling to zero. General Motors DAEWOO Auto & Technology Company (GMDAT) has offered the standard EDS-T-7101 to apply mi- croscopic test over GCI microstructures and grade and rate GCI samples. The GMDAT standard presents five classes to characterize the different shapes of graphite particles named from A to E [8]. Details of GMDAT standard pattern classifi- cation are illustrated through Fig. 1 and summarized in [9]. Fig. 1 Reference images for the five classes of graphite particles accord- ing to standard EDS-T-7101 The morphology and the distribution of graphite grains are the decisive factors in judging the properties of cast iron [10]. Therefore, the assessment of microstructures is an important approach in GCI property analysis. The conventional way of analysis is visual inspection and decision making through a metallurgical expert. This operation can be misleading if the metallurgist was not knowledgeable enough. Therefore, au- tomation of this operation through textural image analysis and artificial neural networks has become of great importance. Several research areas have adopted statistical image analy- sis through gray level co-occurrence matrix (GLCM) feature extraction then decision making through (ANN) like in [11] [12] [13]. This research work illustrates the application of using image processing techniques to analyze GCI texture then apply this analysis to an ANN to categorize the grades of GCI. To start B ———————————————  Eng.Amany Khaled is currently pursuing masters degree program in Me- chanical Engineering in Arab academy for science,technology and maritime transport, Egypt, E-mail: amanykhaled@aast.edu  Assoc. Prof. Mostafa Rostom A.Atia is currently head of mechanical engi- neering department in Arab academy for science, technology and maritime transport. , Egypt E-mail: mrostom1@aast.edu  Dr.Tarek Moussa is the supervisor of materials science laboratory at Ain university- Faculty of Engineering, Egypt. Email: tarekmoussa2001@yahoo.com