基于Hash和倒排项集的海上钻井平台隐患关.联分析

Hazard correlation analysis of offshore drilling platform based on hash and inverted itemset

为了充分挖掘海上平台隐患案例中隐患属性与致因之间的关联性,提高对平台隐患风险预测的准确性和时效性,提出了一种基于Hash技术和倒排项集的关联规则挖掘模式。首先针对292条平台结构现场数据的每个隐患属性进行分析提取;其次在多支持度下,按照多维、多层关联规则挖掘模式挖掘隐患设备、隐患位置、隐患现象等属性的关联规则;然后提出用信息增益来衡量关联规则的有效性;最后,根据挖掘结果分析并总结海上钻井平台隐患特征。结果表明:春季油水分离器常出现缺少保养的隐患,护管常出现未封堵的隐患;一般腐蚀和锈蚀的致因为操作维保不当、防护层剥落,以及防护装置锈蚀严重、脱焊、不符合规范,提高海上平台管理水平可减小现场隐患的发生概率。

An association rule mining model based on hash technology and inverted itemset was proposed to fully mine the correlation between potential unsafe attributes and causes in offshore platform hidden danger cases,and improve the accuracy and timeliness of platform hidden risk prediction.Firstly,292 on-site hidden danger data about the platform structure are cleaned,the typos in the data are corrected,and each hidden danger attribute is analyzed and extracted.Secondly,the association rules of hidden equipment,hidden location,hidden phenomenon,and other attributes are mined according to the multi-dimensional and multi-layer association rule mining mode under multi-support to mine the high-frequency and low-frequency hidden information at the same time.Then the problem that the traditional Apriori algorithm needs to traverse the data set many times and produce a large number of redundant itemsets is solved by using a hash combined with an inverted set.Besides,the time in the intersection of inverted sets is reduced by using hash technology.Furthermore,K(Kulc),imbalance ratio(I),and information gain are used to measure the effectiveness of association rules,and reliable association rules are obtained.Finally,the potential hazard features of the offshore drilling platform are analyzed and summarized according to the excavation results.The experimental results show that the oil-water separator often has the hidden danger of lack of maintenance in spring,and the protective pipe often has the hidden danger of not plugging.The causes of general corrosion and corrosion are improper operation and maintenance,peeling of a protective layer,serious corrosion,desoldering,and non-compliance with the specifications of protective devices.The improvement of the offshore platform management level can reduce the probability of on-site hidden dangers.In addition,the time-consuming results also show that the running time of the improved algorithm is about one-seventh of that of the traditional algorithm.