Algorithm for diagnosing and warning of vibration caused by loosened screws on motor base using wavelet packet and neural network

Motor base screw loosening is a common problem in motor operation, which, if not dealt with in time, may lead to motor failure and damage. However, few studies have focused on the diagnosis and warning of this problem. Based on wavelet packet and neural network analysis, this paper presents a new algorithm for monitoring, diagnosing, and warning vibration caused by loose screws in the motor base. The vibration signal generated by the base screw loosening is monitored and sampled with sensors, and the wavelet packet is used to decompose, reconstruct, and reduce the noise of the vibration signal to enhance the time–frequency characteristics of the signal. After analyzing the fault data by wavelet, the feature vector characterizing the fault is extracted, and then, the vector and the corresponding fault type are used as the input and output of the neural network, respectively, and the non-mapping relationship between them is built to complete the diagnosis and early warning of the fault. Finally, the method is used to compare the motor base screw loosening operation and normal operation. The experiments show that the new algorithm based on wavelet packet and neural network can complete the health diagnosis and early warning of motor in the early stage of motor base screw loosening, reduce the loss caused by subsequent faults, and provide a new reference scheme for the motor fault diagnosis field.