摘要
提出了一种基于盲提取算法的冷却塔故障诊断方法。该方法以采集到的冷却塔的声信号为基础,通过计算各测点测得声信号的联合熵达到最小为目标,首先对输入信号进行去均值和去相关的预处理,使信号能够适用于盲提取算法,然后利用随机梯度算法从处理过的输入信号中提取出故障声信号,并根据故障信号进一步判断故障类型,从而实现声学诊断。实验结果表明,该方法能有效提取冷却塔的故障噪声信号,与其它盲提取方法相比具有明显的优越性。
It presents fault diagnosis method of cooing tower based on blind source processing. This method is based on the acoustic signal of the cooling tower, and its target is to reach the minimum of the joint entropy of the acoustic signal measured by each measuring point. Firstly, removing mean and decor relation are carried out to the input signal. Then stochastic gradient algorithm is used to extract the fault acoustic signal in order to achieve acoustic diagnosis. Through the experiment it is shown that this method can extract fault signal of cooing tower effectively, which has obvious superiority compared with other blind source processing method.
出处
《机械设计与制造》
北大核心
2013年第9期166-168,共3页
Machinery Design & Manufacture
基金
国家自然科学基金(50975180)
国家自然科学基金(51005159)
辽宁省教育厅基金(L2010401)
关键词
盲提取
冷却塔
声信号
故障诊断
Blind Extraction
Cooling Tower
Acoustic Signals
Fault Diagnosis
