Noise Removal Method For MRS Signal Based On Higher Order Cumulants

Groundwater is one of the necessary factors for the survival of mankind, to scientifically grasping the time and space distribution and ensuring the sustainable utilization of water resources development, detection and evaluation of groundwater are very pregnant. In recent years, nuclear magnetic resonance sounding Magnetic Resonance Sounding( MRS) geophysical exploration technology has achieved unprecedented rapid development. So far, many countries, including France, the United States, China and other countries, have launched the development of groundwater detection instrument and research on relevant methods. Whether it is in the detection depth, detection accuracy, or in the commercialization of the instrument has been a qualitative leap. However, due to the high sensitivity of the magnetic resonance, full wave MRS signal measured usually nanovolt level, are easily subjected to random noise, power frequency harmonic interference, thus affecting the Mrs signal characteristic parameters extraction, leading to reduce the accuracy of a subsequent inversion interpretation. Therefore, it is very important to choose a kind of direct and effective method to remove the noise.As a new method of mathematical statistics, High Order Cumulants( HOC) contains more abundant characteristic parameter information than power spectrum. That is the season why HOC method has been widely used in signal processing, system theory and time series analysis, etc. In this paper, based on consulting a large number of documents and taking examples of those successful application experiences in other domains, a research on noise removal method for MRS signal based on higher order cumulants is proposed. According to the different signals of different higher-order cumulant features for noise filtering method of the relative research, so as to achieve directly and effectively suppress the noise of.First of all, through by theoretical analysis, calculation and simulation, analyze the higher-order cumulant features of MRS signal and its environment by electromagnetic noise. The random noise and the peak noise can be restrained effectively, but the harmonicpower frequency noise, cannot be removed via of the same feature with MRS signal.Secondly, according to the removal of the harmonic power frequency noise from MRS signal, using by the feature of Quadratic Phase Coupling(Quadratic Phase Coupling,QPC), a1(1/2) spectrum theory is proposed. By the theoretical calculation and algorithm simulation, the results show that,1(1/2) spectrum method is effectively in distinguishing the harmonic characteristic and MRS signal.Finally, a MRS signal recovery method based on nonlinear fitting method is studied.Compared with the traditional filter,1(1/2) spectrum method has the advantage to extract the feature parameters. And the average fitting error of initial amplitude is 2.72%, the average fitting error of relaxation time is 4.08%. The measured data processing experiment further verifies the validity and practicability of the algorithm.