Abstract: The rolling bearing is an important part of the wheel system of the harvester, and it is also a vulnerable component. It is easy to cause the failure of the rotating machinery for the damage of the rolling bearing, so the detection of the bearing failure is very important. According to the mechanism and characteristics of vibration signal in the process of bearing operation, it proposed the wavelet neural network pattern recognition algorithm based on the analysis of signal time domain and frequency characteristics, which can intelligently identify bearing fault, reduce the error of fault diagnosis and improve the accuracy of fault type judgment. Finally, it realized the visual display function of the fault detection process by combining the virtual instrument development software LabVIEW, which makes the monitoring of the harvester bearing fault more efficient and improves the intelligent level of fault detection.