Abnormal motor detection based on vibration signal

This paper presents an anomaly detection method of motor which combined time-frequency characteristics and fractal dimension. The energy of the original signal is characterized by the time-domain characteristics, and the multiple frequency characteristics describe the components of the frequency-domain signal which is after the Fourier transform of the time-domain signal. Fractal represents the similarity of whole and local existence, and completely describes the characteristic information in the process of motor operation; Then, the genetic algorithm is used to select the features, and the optimal feature subset in the anomaly diagnosis model is selected and sent to the classifier to detect the equipment anomaly. In this paper, the experimental data are used for simulation, support vector machine (SVM) and support vector data description (SVDD) classifiers are used for experiments, and the accuracy is 98.92% and 97.04% respectively. The algorithm in this paper is comprehensive feature extraction, by eliminating redundant features through genetic algorithm, motor anomaly detection can be realized efficiently and accurately, which is more suitable for the application and implementation of detection equipment.