摘要
为了满足扬声器纯音检测的需要,提出了一种基于激光位移传感器的扬声器异音故障检测方法.该方法利用激光位移传感器采集扬声器纸盆中心振动信号,通过短时傅里叶变换将扬声器振动信号从一维的时域信号变换成两维的时频域信号.对时频图采用多级阈值分割的方法进行图像分割,统计其高频区域灰度值落于0~0.3的所有像素个数作为特征值,进而判断扬声器的好坏.实验表明,此方法对扬声器异音故障检测的正确率可达到97%.
In order to meet the needs for speaker pure tone detection, a sound fault detecting method based on laser dis- placement sensors was proposed. Using laser displacement sensors to collect the vibration signals at the center of the speaker cone, and by means of short-time Fourier transform, the vibration signals were transformed into two-dimensional time- frequency domain signals from one-dimensional time-domain signals. Multilevel threshold segmentation method was used to segment images. Finally the number of pixels whose gray value is between 0 and 0.3 in the high frequency area was collected, which was used to judge whether the speaker is good or not. Experiments show that, in this experiment, the accu- racy of this method is 97%.
出处
《天津科技大学学报》
CAS
2012年第3期64-67,78,共5页
Journal of Tianjin University of Science & Technology