Research On Fault Feature Extraction Method For Bearings Of Rotating Equipment In Initial Oil And Gas Processing Unit

Nowadays, Hilbert-Huang transform (HHT) as an adaptive time-frequency analysis method has been widely used in the field of fault diagnosis for rotating machinery. With the development of the advanced and intelligent fault diagnosis technology, HHT also come to mature gradually, but it also exposes some problems.According to the existing problems and research status of HHT through the intensive study of HHT theory, the corresponding solutions are put forward in consideration of the actual characteristics of vibration signals for bearings in initial oil and gas processing unit. Meanwhile, the effectiveness of this method is verified through the simulation and experiments.Firstly, the basic principle of HHT including HHT flow, EMD decomposition process, Hilbert transform and the corresponding algorithm and spectrum are researched. The research status of the problems which exist in HHT, such as mode mixing, end effect, false component, lack of theoretical support and so on are collected and analyzed. Considering the application of HHT in the field of fault diagnosis for rolling bearings, the corresponding improvement ideas of HHT are proposed which contains two aspects exhaustive study for signal pretreatment and the selection of EMD method and conditions.Secondly, basing on the research of existing reduction method for signal noise, an average morphological filter with multi-scale is constructed which also considering the characteristics of environment and vibration signals for rolling bearings. The vibration signals are pretreated by the filter to suppress the surrounding noise and facilitate the extraction of characteristic frequency. Meanwhile, considering the actual characteristics of vibration signals for rolling bearings, a symmetric extension method based on feature matching is suggested through analyzing the existed extension methods. The symmetric extension method can be combined with the interpolation method in EMD to suppress the mode mixing and end effects etc.Then, the interpolation method and IMF criterion and sifting stop criterion in EMD are discussed. The improved HHT method and the corresponding feature extraction flow are proposed through the combination of signal pretreatment and the selection of EMD interpolation method and criterions.At last, the simulation and experiments for feature extraction based on traditional and improved HHT are carried out. The improved HHT method suggested in this paper is verified to be effective in extracting fault characteristic frequency of rolling bearings through the comparison and analysis of the results.