Comparison of the cbr methodology and the cart algorithm applied to problem of casting defects classification

Krzysztof Regulski , Gabriel Rojek

AGH University of Science and Technology, Department of Applied Computer Science and Modelling, Al. Mickiewicza 30, Krakow, Poland.

DOI:

https://doi.org/10.7494/cmms.2014.3.0489

Abstract:

The main scope of presented in this article research is the analysis of application of artificial intelligence methodologies at building of a computer system that should aid at problem of casting defects classification. The computer system is designed as a decision support tool in the diagnosis of casting defects for small and medium-sized plants, which implies restrictions according to the usage of data that, in this case, are not measured in real time of production. Without access to control data, the diagnosis of casting defects can be based on observations made by a technologist responsible for the inspection of ready castings. Those observations concern usually the type of damage, its distribution, location, occurrence or even color of surface. The problem of such observation based diagnosis can be resolved by building of a computer tool, which uses classification methodologies in order to give aid at casting defects classification. Presented research focus on two methodologies within artificial intelligence: Case-Based Reasoning (CBR) and Classification And Regression Trees (CART). The CBR methodology enables to use knowledge according to previously made classifications in order to help predict the present classification problem. The decision support system with the applied CBR methodology is able to learn basing on the knowledge which is acquired in the result of classifications performed by this system. The CART algorithm enables to generate classification tree, which can be easily used by a technologist or by an expert system, giving support at defect diagnosis. Presented in this article research concerns comparison of those two methodologies in terms of their usefulness at designing the system operating in conditions of small and medium-sized casting factory.

Cite as:

Regulski, K., & Rojek, G. (2014). Comparison of the cbr methodology and the cart algorithm applied to problem of casting defects classification. Computer Methods in Materials Science, 14(3), 180 – 189. https://doi.org/10.7494/cmms.2014.3.0489

Article (PDF):