Platform Development Of Bearing Life Prediction Based On Bp Neural Network

Bearing plays a significant role in carrying and transiting load, and it has become an indispensable part in the mechanical industry. There are lots of major accidents caused by the fault of bearing which works under bad conditions. Therefore, mastering the residual working life of bearing can help to prevent the unpredictable bearing fault, prompt worker to do some reasonable performance check and timely replacement to avoid the damage of the mechanical equipment caused by bearing fault, and consequently, reduce the cost of industrial production and decrease mortality.In this paper, a6309bearing from locomotive fan motor was studied. At first, the vibration signal of this bearing was analyzed and processed, some important parameters such as effective value, peak value and kurtosis etc, were extracted. Subsequently, the bearing network structure which can predict the residual life of bearing was designed based on the BP neural network, and MATLAB toolbox was conducted to realize the network structure. Training function and hidden layer neurons were performed to optimize the BP neural network. After learning and training, the most suitable training function and the optimal number of hidden layer neurons can be selected by comparing the performance of various networks. After optimization, a large number of bearing data samples were used for learning and training, to achieve structure weights and threshold values with smaller prediction error.Comparative analyzing the advantages and disadvantages of the Visual Basic and MATLAB programming methods was performed, to select the most appropriate interface method. A structure and algorithm formula based on the BP neural network was proposed. Subsequently, an M file which can predict the bearing life by combining structure weights and threshold values was wrote and replaced the neural network toolbox function. Finally, mixed programming method of COM component was conducted, to achieve the invoking of the DLL file in the user interface. Here, the COM components were generated from the M function file,The operating interface of prediction platform of bearing life was designed and wrote in Visual Basic6.0environment. The main modules were included the landing module, bearing life prediction module, data statistics module, and auxiliary module. These modules were used to achieve the function such as system landing, bearing life prediction, data processing, user registration and password changing. Therefore, developing a bearing life prediction and bearing data integration platform is significant to predict bearing lift at any time in the actual production.