Application of bayesian network in the diagnosis of hot-dip galvanising process

Barbara Mrzygłód¹, Anna Adrian¹, Stanisława Kluska-Nawarecka^1,2, Robert Marcjan³

¹AGH UST, Department of Industrial Computer Science. ²Center of Competence for Advanced Foundry Technology in Cracow. ³AGH UST, Department of Computer Science.

DOI:

https://doi.org/10.7494/cmms.2007.2.0163

Abstract:

This study presents an output of the application of a probabilistic method of inference based on Bayes’ rule in the diagnosis of defects formed during hot-dip galvanising process. Bayesian cause-effect network for given group of surface defects and its causes was build. Many factors causing defects was taken into consideration in like: technological parameters, technological nodes and character of cause. The process of creating knowledge representation of the hot-dip galvanising process was disclosed on chosen defect (discontinuity of coating) and two causes (pH fluxing bath and surface contamination) along with a scheme of reasoning in Bayesian network and its implementation in a Norsys Netica packet. The advantages and drawbacks of a probabilistic method of representation of the incomplete and uncertain empirical knowledge were highlighted.

Cite as:

Mrzygłód, B., Adrian, A., Kluska-Nawarecka, S., & Marcjan, R. (2007). Application of bayesian network in the diagnosis of hot-dip galvanising process. Computer Methods in Materials Science, 7(2), 317 – 323. https://doi.org/10.7494/cmms.2007.2.0163

Article (PDF):