摘要
提出了一种阀厅中设备声音故障识别算法,实现了在阀厅智能巡检机器人下对设备的智能检测和状态识别。首先,使用机器人携带的拾音器采集待检测设备的声音信号。其次,使用小波分解将信号由时域转化为频域,获取信号的第五层细节系数作为新声音信号。最终,提取新声音信号的质心、方差、能量和熵作为特征向量,使用BP神经网络来识别声音信号判断设备的运行状态。实验结果显示,该算法能实现设备故障诊断功能,算法简单、准确率高的特点。
An algorithm of sound equipment fault identification based on valve hall intelligent inspection robot is put forward in this paper, which realizes intelligent detection and pattern recognition. Firstly, robot was collected the device sound signal.Secondly, the signal is transformed into frequency domain by wavelet decomposition, and the fifth layer detail coefficient of signal is used as the new sound signal. Finally, the centroid, variance, energy and entropy of the new sound signal was extracted as feature vector, and BP neural network was used to decided the equipment running status. Experimental results show that this method can realize the fault diagnosis function, high accuracy and simple.
出处
《电子设计工程》
2016年第21期63-65,68,共4页
Electronic Design Engineering
关键词
阀厅巡检机器人
声音识别
小波分解
BP神经网络
valve hall inspection robot
voice recognition
wavelet decomposition
BP neural network
