The Fault Diagnosis System Based On BP Neural Network For Steel Convertor Roller Bearing

Faults diagnosis of large-scale power equipment is closed to production safety and economic efficiency of the enterprises. Roller bearing is the common component in power machinery, which is the major equipment in power generation, metallurgy and chemical industry, the state of their work is related to the normal operation of the entire system. At present, the means of monitoring converter bearing in steel mills is uncultured, generally depending on worker's experiences, which cause energy consumption and low efficiency. So it is necessary to design and develop a set of reliable, applied, and advanced system of online monitoring and faults diagnosis which ensure the safety of equipment, energy-saving operation.LabVIEW is a standard software of data acquisition and instrument control. The environment of MATLAB contains neural network toolbox, But it is far less than LabVIEW in interface development, equipment control and network connections. So this paper made a mixed MATLAB and LabVIEW program in the MATLAB Script node provided by the LabVIEW.The drive mechanism of Steel converter is trunnion bearing, which is the key component of steel converter, Trunnion bearing is rolling bearing with low speed and high load, its fault diagnosis is a very complicated issue, there are non-linear mapping relations between fault and signs because of the location of the installation, operation condition and other factors. Recently, ANN has been widely applied in fault diagnosis of machinery because of its associative memory and non-linear pattern recognition.Following computer techniques fast developing, the fault diagnosis system built by the microcomputer composed with proper software and hardware is easy to patulous, the BP (Back Propagation) algorithm was made by a mixed MATLAB and LabVIEW program, so this paper developed the fault diagnosis system based on BP neural network for Steel Convertor roller bearing by LabVIEW8.2.As the trunnion bearing has a vibration with the rotation, the state of the operation is often directly reflected in the vibration signal which is tested simply and directly. The vibration signal picked up by the sensors reflects the work of the bearing itself, but also contains a large number of other parts of the operation such as noise vibration information. So fetching the fault signals and suppressing background noise and effectively diagnosing faults become the important parts of roller bearing vibration monitoring and diagnosis techniques. This paper analysed roller bearing vibration signal in both time and frequency domain, acquired characteristics, then input BP neural network to get finally result after the analysis of the network.Finally, this paper verified the conclusion by using actual data, The results show that fault diagnosis system based on BP neural network for Steel Convertor roller bearing is reliable and practical, as a kind of pattern recognition technology, this system has a good applicative expectation in fault diagnosis of equipment.