关键词: 冰箱, 时频域, 聚类分析, Kmeans++, 特征识别

Abstract: Aiming at the identification of refrigerator fault acoustic signals under compound sound sources, a method for identifying fault acoustic signals of refrigerators based on time-frequency domain clustering analysis is proposed to realize the identification of fault acoustic signals under strong noise. Firstly, the data acquisition instrument is used to obtain the acoustic signals of the refrigerator in the normal operation state and the fault state, and then process them into frames; then the normal n-frame acoustic signals are decomposed by IEEMD, and the full characteristic parameters of each modal component are calculated; then the kmeans++ algorithm is used to analyze the perform clustering analysis on all feature parameters to obtain the optimal features of the expression signal; finally, using the optimal features of n frames as the clustering benchmark, identify the fault signal (including noise), and complete the identification problem of the fault sound signal under strong noise. Acoustic and vibration faults provide pre-processing data support.

Key words: Refrigerator, Time-frequency domain, Cluster analysis, Kmeans++, Feature recognition