A Bayesian Approach to Mine Accident Causes Association Rules in Petroleum Drilling

摘要

The causes of major accidents during drilling operation are very complex. Association rule has been applied to analyse the cause-and-affect relationships of accidents in petroleum drilling. Bayesian Network (BN) is a graphical model which represents probabilistic relationships among variables in a database. In this paper, we propose a two-stage approach to mine meaningful association rules, the approach combines Apriori algorithm and BN. These meaningful rules are applied to petroleum drilling accidents analysis. First, simple association rules are mined through setting minimum support and minimum confidence. Second, these rules are synthesised with BN to obtain more complex and meaningful rules. Third, those complex and meaningful rules are applied to operation suggestion. The experimental results show that our approach can obtain rules accurately and efficiently.