基于LMD-LSSVM的扬声器异音故障诊断方法研究

Research on Loudspeaker Abnormal Sound Fault Diagnosis Method Based on LMD and LSSVM

文章采用局部均值分解(Local mean decomposition,LMD)和最小二乘支持向量机(Least squares support vector machine,LSSVM)相结合的方法进行扬声器异音故障诊断的研究。首先,选取正常扬声器与漏气、铁粉杂质、小音三种故障类型的扬声器作为研究对象,在实验平台上对上述四种类型扬声器施加激励信号。然后,获得不同类型的响应信号并对其进行LMD分解,得到一系列乘积函数(Product function,PF),求出它们的能量熵,构成特征向量。最后,将提取的特征值作为LSSVM模拟分类器的输入量进行训练和识别,诊断出扬声器的故障类型。实验结果表明,该方法可以有效地对扬声器异音进行故障诊断分类,诊断准确率平均达93.42%。

In this paper,the method of combining local mean decomposition(LMD)and least squares support vector machine(LSSVM)is used to study the speaker abnormal sound fault diagnosis.Firstly,the normal speakers and the speakers with air leakage,iron powder impurities and small sound faults are selected as the research objects,and the excitation signals are applied to the above four types of speakers on the experimental platform.Then,different types of response signals were obtained and LMD decomposition was carried out to obtain a series of product functions(PF),and their energy entropies were calculated to form eigenvectors.Finally,the extracted eigenvalues are used as the input of LSSVM simulation classifier for training and recognition,and the fault types of loudspeakers are diagnosed.The experimental results show that this method can effectively diagnose and classify speaker abnormal sound,and the average diagnostic accuracy is 93.42%.