Research On Applications Of Data Mining Technology To Pressure Pipeline Safety Management Work

Enterprising safety management is very important for normal functions and productionsof an enterprise. Enterprising safety management almost always involves safety managementfor mechanical equipments. As the design of these equipments becomes more complicatedand their applications become more diversified, enterprises have accumulated a large amountof data records about quality and functions of these equipments. Efficiently managing dataand extracting useful safety management information become important and challengingproblems in safety management.This thesis applies a new emerging technology of data mining to solving the problem ofefficient analysis of a large amount of data, and hence reveals useful information embedded indata. We choose pressure pipelines as a class of representative equipments for ourinvestigation, and reveal insightful models and factor correlations in data about pressurepipeline accidents by applying data mining technology. These results cannot be obtained fromtraditional statistical approaches.More specifically, this thesis applies three data mining algorithms: decision treeanalysis, association rule analysis, and regression analysis, and investigates applications ofdata mining to safety management on pressure pipeline accidents from different perspectivesin order to fully reveal useful information represented by data. We first apply decision treeID3 algorithm to establishing the decision tree for pressure pipeline accidents, and analyze theimpact of the pressure pipeline type and the accident cause on accident classification. Next,we apply association rule analysis to investigating the correlation between the pressurepipeline type and the accident cause. By quantitatively analyzing the lifts for the pressurepipeline type and the accident cause, we reveal that for a given pressure pipeline type, somespecific causes result in accidents more easily. Finally, we apply regression analysis andestablish a functional model of the relationship between the number of pressure pipelineaccidents and the corresponding year. We also use the model to obtain an accurate predictionof the number of accidents in a future year. This also demonstrates the accuracy of theestablished model. This thesis not only uses an example to demonstrate power of data mining technology onenterprising safety management, but also provides a specific application context forimprovement of data mining technology, thus helping the development of both the area ofenterprising safety management and the area of data mining in a complementary manner.·...