Applications Of Vibration Fault Diagnosis For Rotary Machinery Based On Data Mining

The large rotary machineries are very important equipments for the factories. They alwayshave complex structure and high degree of automation, so their running intact or not directlyaffects the economic efficiency of enterprises. With the increasingly wide range of applicationsof computer technology, the use of computer management to first-line production equipmentbecomes possible. This is also in line with the trend of modern industry to the development ofinformation technology, high-speed and automation. Therefore, it is very necessary to theforefront of the computer technology to remote fault diagnosis of rotary machineries, instead ofthe traditional manual inspection mode by people. This is of great significance to reduce the costof fault detection, and improve the efficiency of production and operation of enterprises.In this thesis, we systematically expounded the relevant principles in the field of rotarymachineries fault diagnosis and research, and focus on the specific method of using the vibrationdata to the device for diagnosis. Vibration fault as a fault occurs most frequently in rotarymachineries, attaches great importance to the problem of manufacturing enterprises in theday-to-day patrol. Because of the amount of data is too large, the artificial analysis method is soimpractical. Therefore, data mining techniques has become the first choice to solve theseproblems.Data mining techniques can be taken as a new type of data processing. It is a method thatthrough the analysis of existing data from the vast amounts of data to find implicit informationthat is useful to humans, then show the result of mining. This study determined technologyroadmap for the vibration of rotary machinery fault diagnosis based on data mining. Focus of thisstage the field of data mining algorithm. According to the actual situation of the existing datasource, select a algorithm from the data mining algorithms as a research-based. Treat the rotatingmachinery vibration fault as the research object, and then find the fault diagnosis rules by theselected data mining algorithm from a large number of production data. Then Compare the rulesin the rule base to the existing rules. The results show that, the correct rate of the rules which found by the data mining algorithm is more than85%. Create a rules knowledge base composedby failure phenomenon, failure reason and expert advice. It can provide a necessary theoreticalfoundation for fault diagnosis of rotary machineries.We use the fault diagnosis rules that found by the algorithm of data mining to develop asoftware which can diagnose the fault of rotary machineries. This software is a windows7platform, developed by application Myeclipse development tools, Mysql database, Navicat forMySQL8.0database management software and java programming language. It is a B/S structuresoftware using object-oriented program design thinking.In the process of the development of this software, we use the software engineeringtechnology to detail the requirement analysis, system design, coding, software testing andsoftware implementation.After a series of tests and commissioning work, the software achieves the desired objectives.Currently, the software has been delivered to the user, it is very helpful for the enterprises indaily rotary machineries vibration fault diagnosis, improving the degree of informationmanagement, and reducing excess maintenance costs.With the data mining technology and machinery fault diagnosis is becoming increasinglymature, and using computers to diagnose the fault of rotary machineries is more and morepopular. The result of this thesis precisely follows the development trend of the immediate fieldof mechanical fault diagnosis. It can provide some reference and reference for the study of datamining techniques.