Research On Fault Detection Method Of Marine Induction Motor Bearing

Induction motor is widely used in the field of ship because of its low cost,high reliability,easy maneuverability and simple structure.When the induction motor has serious defects,it will lead to abnormal work of equipment concerned on board,and even directly threaten to the safety of ship.As an important component of induction motor,bearing is also one of the components with the highest probability of induction motor failure.With the development of intelligent ship management and intelligent ship research,there is an urgent need for the mobile diagnosis equipment and online diagnosis of the bearing fault of induction motor.In this paper,according to the characteristics of ship vibration environment and harmonic noise of power grid,which is worse than that on the land,the current characteristic analysis method with non-invasive measurement characteristics is used to find a way which has relatively low computational complexity but effectively identify ability of the fault in the induction motor bearing.It can be applied to portable induction motor bearing fault diagnosis equipment or online fault monitoring method.In this paper,the methods of fault extraction,such as wavelet transform,Hilbert transform,WELCH method and empirical mode decomposition,and three pattern recognition methods,such as neural network,random forest and support vector machine,are analyzed and studied in the aspect of fault diagnosis of induction motor bearing.According to the trade-off of fault information extraction,recognition effect and calculation complexity,a new method is put forward through experimental comparison based on Blackman's window function,the WELCH method is combined with support vector machine(SVM)to extract and identify bearing fault information.In terms of fault identification,the feature parameter selection and weighting algorithm are designed to optimize the structure of the fault feature vector input by the support vector machine.And five parameters(kurtosis,skewness,crest factor,clearance and shape factor)are proposed as the input fault eigenvector construction method of SVM.The kernel function parameters are optimized by using grid search method and applied to support vector machine training and rolling bearing fault recognition.Based on the above method,the hardware and software of marine induction motor bearing fault monitoring system is designed,including current sampling circuit,A/D conversion circuit,DSP-PC communication circuit,key input circuit,system display circuit,system training,fault extraction and identification software program.The experimental results show that the fault diagnosis method for induction motor bearings proposed in this paper is feasible and effective in rolling bearing fault diagnosis applications.And the method is relatively low in computational complexity and cost,which can be applied to portable diagnosis equipment and on-line and off-line induction motor bearing fault diagnosis.