【摘要】 研究了利用从扬声器响应信号中提取特征进行扬声器故障识别的方法.首先通过小波包分解及重构得到扬声器响应信号的初始特征;然后利用主分量分析(Principal Component Analysis,PCA)的方法对初始特征进行降维处理,并得到最终特征;设计神经网络分类器,并将得到的最终特征输入分类器进行识别.实验表明,该特征提取方法在满足扬声器故障检测识别率的同时,降低了特征提取过程中的计算量,为扬声器故障诊断提供了一种实用方法.更多还原

【Abstract】 The method of extracting feature form loudspeaker response signal was used to detect the loudspeaker.Initial characteristic was obtained from loudspeaker response signal by using the decomposition and reconstruction of wavelet-packet firstly.Then principal components analysis was used to get the final characteristics.The final charac-teristic was used by the designed Neural network to recognize the loudspeaker.Experiments show that the method of extracting feature can meet the recognition rate and reduce the computation during the process of extracting feature.It provides a practical method for diagnosing the loudspeaker.更多还原