Study On Fault Diagnosis Of Mine Hoist On Basis Of Data Mining Technology

Playing a crucial role in ensuring work safety in coal mine, mine hoist is the key equipment of coal mine. In the process of the hoist daily monitoring, vast amounts of data, implying a large number of potential information, are produced. Therefore, the paper introduces data mining technology to improve the ability of automatically acquirement of mine hoist fault diagnosis knowledge, and designs hoist fault diagnosis system based on data mining technology, which is of great importance to ensure coal mine production safety.The paper choses mine hoist system as the research object, draws a conclusion about characteristics of the common mine hoist mechanical failure, researches the theory of rough sets and decision tree classification algorithm, introduces the diagnostic model of decision tree classification algorithm based on approximate accuracy and designs the mine hoist fault diagnosis system based on data mining technology.The main research contents include five points.(1) The common failure types of mine hoist and vibration mechanism are analyzed, and the common fault characteristics are summarized. To make a clear aim at monitoring signal for the system, characteristic parameters containing fault information are analyzed in time and frequency domain.(2) The basic theory of decision tree algorithm is introduced as theoretical foundation. The generation process of decision tree, evaluation criteria and common classification algorithms are analyzed. For the sake of verifying the practicality of data mining technology applying in machine fault diagnosis, diagnostic experiment based on decision tree classification algorithm using hoist monitoring data is achieved.(3) The basic theory of rough sets is studied for further research. The model of decision tree classification algorithm based on approximate accuracy and its extended model are established and tested by instance data. While the principles and construction process of the two algorithm models are gave.(4) The model of decision tree classification algorithm, the model of decision tree classification algorithm based on approximation accuracy and its extension model are built on professional development platform for data mining, realizing the analysis and evaluation of those models.(5) The acquisition and processing methods of monitoring signal, data mining and fault diagnosis procedures are analyzed. From the perspectives of the signal acquisition module design, database design, system management module design, online monitoring module design, data mining and the fault diagnosis module design, the hoist fault diagnosis system based on data mining technology is established. The system has the advantages of multiple functions, excellent expending capacity and low-cost, and fault diagnosis rules generated by data mining are effective. The study lays a good foundation for the future study and practical applications.There are 50 figures, 30 tables and 87 references in the thesis.