自适应广义形态滤波方法在介损在线监测数据处理中的应用研究

STUDY ON APPLICATION OF ADAPTIVE GENERALIZED MORPHOLOGICAL FILTER IN PROCESSING ON-LINE MONITORING TANδ DATA

该文基于数学形态学的基本形态变换及其组合形式,提出了一种采用不同结构元素级联而成的广义形态滤波器,将其应用于介损在线监测数据处理,以保证进一步分析诊断的准确性。考虑到广义形态滤波器仍存在统计偏倚现象,造成开一闭滤波器的输出幅值偏小,而闭一开滤波器的输出幅值偏大的问题,结合了自适应方法实现形态开、闭自适应加权组合滤波以保证输出的准确性。仿真分析和现场处理结果表明,广义形态滤波对于数据中存在的各种加性白噪声和正负脉冲性噪声有较好的抑制作用,能有效地还原数据的基本变化规律,同时采用自适应组合形态滤波的效果要优于单一形态滤波方法和其它非形态滤波方法。

Based on basic morphological transforms and their combination modes, a generalized morphological filter cascaded by different structure elements is proposed. The filter is applied in processing on-line monitoring tanδ to guarantee the accuracy of further fault diagnosis. Due to statistic bias of morphological filter, the output of open-closing filter is relatively small, but that of clos-opening filter is contrary. Therefore, adaption approach is employed to modify weights to realize the combined morphological filter. Results of simulation study and field data processing show the generalized morphological filter can suppress different kinds of white noise and pulse noise, restore the general regularity of the data effectively, and adaptive combined morphological filter has better performance compared with single morphological filter and other filtering approaches.