Research On NMR Signal Detection And Denoising Algorithm

Nuclear magnetic resonance (NMR) method has wide measurement range, high precision, and is one of the most widely used methods in magnetic field measurement. The marginal oscillator, acting as RF generator and NMR signal detector, is the core device to detect NMR signal in traditional NMR signal detection. The magnetic field strength can be calculated according to the RF frequency when the peak space of the NMR signal monitored by an oscilloscope gets equal. First, because the variable capacitance diode of marginal oscillator is easily affected by operating ambient, the RF frequency is unstable. Second, the oscilloscope just roughly estimate the peak space of NMR signal. Third, the NMR signal has too much noise. These result in great error in practical measurement. So, this thesis emphases on research of precise NMR detection and denoising method.An improved detection method is proposed according to defects of tradition method. The RF signal is generator by direct digital frequency synthesis (DDS) and phase lock loop (PLL). The RF emitting coil is placed in the coil of the marginal oscillator that is used only as NMR detector. Experiments show that this method can get stable NMR signal. A single chip is used to sample the NMR signal using Analog Digital (AD) converter. The sampled data is transferred to a PC for post processing. The PC adjusts the peak space of NMR signal until it strictly gets equal, then calculates the value of magnetic field. The testing results show that the precision of magnetic field strength may reach10"7.The strength of the sampled NMR signal is weak. So when being detected, the NMR signal usually is affected by electromagnetic field from the operating ambient. It is one of the key points to research how to deprive the noise and enhance the Signal Noise Ratio (SNR). This thesis uses wavelet threshold denoising algorithm to denoise the actually sampled NMR signal. An improvement is made to this algorithm, and a new threshold denoising algorithm is proposed based on the improvement. It syncretizes traits of soft-threshold and hard-threshold denoising methods by weighted average and the parameters can be adjusted properly to produce the best estimations of the wavelet coefficients. The experimental results show that the quality of the denoised NMR signal is significantly improved. The detail of NMR signal is clear, which brings convenience for analyzing peak spacing. At last, the summary on this research is described and the prospect of the novel method is also proposed.