关键词: 信号处理, 故障检测, 特征提取, 语谱图

Abstract: A method for identifying faults in air conditioning motors based on spectrograms was developed. By applying spectrograms and Mel spectrograms to simulate human auditory perception, features were extracted from motor audio samples to distinguish between normal and faulty samples. The artificial features calculated included spectrogram harmonic noise ratio, adaptive gradient filtering features, and a newly proposed higher-order spectrogram feature. These features were used to identify four common types of motor faults: looseness, friction, bearing and lubrication faults.
Experimental results demonstrated that this method was effectively applied to a specific model of air conditioning motor, accurately identifying the aforementioned fault types. The study demonstrated that using audio analysis combined with auditory simulation enhances the accuracy and efficiency of fault diagnosis, providing a new technical approach to fault diagnosis in air conditioning motors.