Based On The Data Mining Technology In The Application Of Gearbox Fault Diagnosis

Gear transmission is the most common way of transmission of mechanical equipment, abnormal gear is the important cause of machine failure,. In today’s society, computer application in various enterprises in the production of more and more widely, a high degree of automation of this for various enterprises at present complex operation data record processing equipment has the very big help. But the more complex the equipment and high-end, its data the more big. The data mining technology is applied to the equipment of computer remote fault diagnosis, will improve the previous traditional diagnosis method of hysteresis, high cost and unpredictable situation. Data mining technology has the function of data analysis and processing, it has the advantage of the intelligent diagnosis widely promotion and application in various industries.This paper focus on the gear box fault features of various parts and explained the characteristic parameters. And then several kinds of feature parameter extraction method was introduced in detail, at the same time through the gearbox vibration signal of time domain and frequency domain analysis for gearbox fault samples, and some important characteristic index parameters. In gearbox fault diagnosis, through to the traditional methods for the diagnosis of gear box and intelligent methods were compared. Then starting from the concept and characteristics of data mining technology, put forward the idea of fault diagnosis based on data mining.Finally, the paper focuses on the classification of the data mining method and its principle, on the basis of the existing using rough sets theory to reduction of knowledge ability and rapid classification of combined with C4.5 decision tree algorithm, and optimization of the system, and then build a new kind of improved rough set decision tree. Finally through the simulation test of gear box fault samples, show that : first of all, can effectively reduce the workload characteristics of data acquisition, and can rapidly and accurately for different fault identification, has stronger than direct use of decision tree C4.5 algorithm of engineering practicability.