【摘要】 针对基于Empirical Mode Decomposition(EMD)的风机振动信号异常检测中噪声污染、CSI拟合包络线导致的过冲问题,端点效应引发的端点飞翼现象三点问题,提出了一种改进的EMD算法.该算法首先引进小波方法对原始数据进行降噪处理,再用边界特征尺度匹配方法对原始信号两端进行端点延拓处理,降低端点效应,同时结合3次Hermite插值拟合法的良好柔性来拟合包络线,以获得均值曲线.实验表明,利用该改进的EMD方法得到矿井风机振动边际谱,能清晰地得出风机振动信号特性,消除了过冲的影响,对端点效应也有了明显改善,提高了风机异常检测的准确率.更多还原

【Abstract】 Focused on the issues that caused by the noise pollution of ventilator,the over shoot problem caused by CSI fitting Envelop curve,and the end swing problems caused by the end effect,an Improve-Empirical Mode Decomposition algorithm was proposed. Firstly,the algorithm introduces the wavelet method to denoise the original signal data,then uses matching boundary feature extension method to effectively restrain the end effect while combining with the good flexibility of Cubic Hermite to fit the envelope to obtain the average curve. Experiments shows,by analyzing the ventilator vibration marginal spectrum with the improved algorithm,ventilator vibration signal characteristics can be shown clearly,the improved algorithm can eliminate the overshoot problems,markedly improve the endpoint effect and improve the accuracy of the ventilator anomaly detection.更多还原