Signal Analysis And Processing Based On Wavelet Analysis In Combination Station

Based on the idea of wavelet transform, this paPer malnly studies theaPplication of wavelat in de-noising and singUldrity detection, Which aims at thepractical problems in the combinaion station.The oil-dehydration process is a complex, MIMO, variable couPlingseriously, disturbance vnying frequently manufacturing process, which practicallyhas less measurable information. In this process, signals of oil-water interface andinner pressure are imPortant parameters that reflect the system working status andguarantee the systCm to run safely. Analysis shows that they can be considered asthe stochastic process around some working state under the disturbance inPllt ofgauss White noise.The ever-growing wavelet analysis method has been a poweffol tool inanalyzing and processing signals. WaVelet transform becomes a superiortimefrequency localization method due to its mathematical property itself It hasmuIti-resolution characteristics. By flexing and moving a base wavelet, whichmust satisfy some condition, we can analyze signals in multi-scale finely, so it iscalled "math microscope". In this aticle, we reconstruct the signals of oil-watCrinterface and inner pressure through Mallat algorithIn. The reconstruction ermrsatisfies the syStem resollltion perfectly, Which shows that the adaptability ofwavelet transfOrm in the analysis of unstable signals.De-noising is a great part in the aPplication of wavelet tfansfOrm. It is moreflexible than the classic filtering method in frequency domain. Wavelet methodcan not only obtain a higher SNR, but also hold a good resolution. Substaniallyspeaking, De-noising on the wavelet transfOrm modulus maxima comes from thefact that there are differences on the spread quality in the space of multi-scalebetween signals and noises, it can gain satisfactory effects as long as there isenough difference in singularities between signals and noises. While de-noisingby threshold is a self adaptation solution comParably, which is based on thedifference betWeen signals and noises in the distribution charcteristic underwavelet transform domain by the degree of energy centraization. This method istheoretically simPle, and has small calculation quantity Especially the thresho1dselection presents good adaPtability and the result after de-noising has finesmoothness. The Matlab simulation results, which is on the filtering fruits ofoil-Water interface and inner pressure signals by wavelet show that waveletanalysis is suPerior and useful in de-noising.