Application Investigation Of Data Mining On Fault Diagnosis For Rotary Machinery

Under the impetus of the computer and information technology, the modern industry develops gradually to this direction that production equipments become larger, more high-speed and automated. In this case, significant advancement is made in these aspects to enhance the production efficiency, reduce cost, saving energy and improve products quality. But the more complex the structure of machines becomes the more difficult the fault diagnosis also is. If equipments go down for the failure to discover and solve faults, it will bring more loss to enterprises and make the service more difficult. Therefore, it is necessary to develop a system that can monitor conditions of equipment and diagnose the style of the fault automatically.Actually, the fault diagnosis of equipment is a kind of induction classification processes. This process includes these following steps: to extract fault signals' characteristic frequencies, to arrange and combine fault signals according to these frequencies, and to classify different fault categories. The decision tree in the data mining can solve these questions very well. Techno-way is ascertained for rotary machinery fault diagnosis on line based on Data Mining. This paper mainly studies four kinds of typical rotating machinery faults including rotor misalignment, rotor unbalance, bear oil whip and shaft friction and hitting. Four kinds of typical fault principles and characteristic behavior are analyzed. Vibration analysis method is a very important fault diagnoses way for rotating machinery. The different fault samples tables are gotten by analysis of time domain and frequency domain. State information from monitoring rotating machinery is matched with decision tree to confirm state mode, and then the decision of fault diagnosis is made out.Under the Window XP, applying the Eclipse development kit, JAVA, Oracle 8i database, using the object-oriented programming thought, Knowledge Discovery is designed and realized, at the same time decision tree is constructed. Regarding of this system Web application, the frame Java Applet/Servlet is used. The model level obtains the decision tree through the JDBC visit database and makes the pattern matching according to the submission data. At the same time, system returns to user the result of requested fault diagnosis.This system is delivered to YanShan Petrol-Chemical Co., and it can make the quite accurate distinguish to several typical faults of some large-scale machines. It proves that this system can be used practically. And it is also helpful in cutting short the produce detect budget and advancing the produce stabilization.