Noise suppression method for time-frequency electromagnetic detection data

Abstract

The invention discloses a noise suppression method for time-frequency electromagnetic detection data. Corresponding signal processing work is carried out on analog signals through a front-end signal conditioning circuit and digital signals obtained after AD sampling on an FPGA platform so as to improve the signal-to-noise ratio of input signals. And in a time domain electromagnetic detection mode, denoising processing is carried out on the transient sampling signal by using a superposition average algorithm. In a frequency domain electromagnetic detection mode, a harmonic sampling signal is subjected to digital orthogonal locking amplification algorithm processing, frequency mixing is carried out on the harmonic sampling signal, and the frequency spectrum of the harmonic sampling signal is shifted to be near zero frequency; low-pass filtering is realized through a narrowband filtering extraction module, out-of-band noise is suppressed, and the signal-to-noise ratio of the signal is improved; and further improving the signal-to-noise ratio by using an FFT algorithm, and obtaining the amplitude and phase of the target signal through a CORDIC algorithm.

Description

Technical Field

[0001] The present invention relates to the technical field of underground space electromagnetic detection data processing, and particularly to a method for suppressing noise in time-frequency electromagnetic detection data. Background Technique

[0002] The data processing and detection activities of existing time-domain electromagnetic detection systems and frequency-domain electromagnetic detection systems are in a separated working mode. The data collected during the detection process is first stored in the device and then transmitted to the signal processing system for data analysis after the detection is completed. The real-time detection and positioning capabilities still need to be improved. Using FPGA for the acquisition of received signals, large-capacity storage, and efficient filtering and preprocessing of data greatly reduces the workload of subsequent data processing, thereby improving the detection efficiency. Summary of the Invention

[0003] The purpose of the present invention is to provide a method for suppressing noise in time-frequency electromagnetic detection data.

[0004] The technical solution for achieving the purpose of the present invention is as follows: In the time-domain electromagnetic detection mode, the transient superposition of sampling signals is used to improve the signal-to-noise ratio by utilizing the randomness of noise, and then the trace superposition is used to further improve the signal-to-noise ratio while compressing the data volume. In the frequency-domain electromagnetic detection mode, a digital decimation filter is used to decelerate the mixed signal, and parameter control is introduced to flexibly select the decimation rate according to detection requirements, greatly improving the detection performance of the system; then, the out-of-band noise is suppressed by a corresponding digital low-pass filter to improve the signal-to-noise ratio of the signal; then, according to Parseval's theorem, the continuous energy in the time domain is transformed into discrete energy in the frequency domain by using FFT; the amplitude and phase of the harmonic induction signal are the final required information, and mathematical formulas of trigonometric functions and square roots are required. Using the CORDIC algorithm, the amplitude and phase can be obtained only by addition and shift operations, saving on-chip resources.

[0005] The entire signal processing algorithm for time-domain and frequency-domain electromagnetic detection data is implemented inside the FPGA, achieving efficient processing of detection data and reducing the workload of subsequent data inversion.

[0006] In the time-domain electromagnetic detection mode, an average processing algorithm of transient superposition and trace superposition is adopted to eliminate the influence of random noise and improve the signal-to-noise ratio of transient detection data.

[0007] In the frequency-domain electromagnetic detection mode, compared with the traditional quadrature lock-in amplifier, the digital method is used to generate more accurate quadrature signals; mixing to near zero frequency instead of zero frequency reduces the influence of DC interference.

[0008] In the frequency-domain electromagnetic detection mode, the narrowband filtering and decimation module adopts a multi-stage decimation structure with configurable parameters, which can independently select decimation processing with different decimation factors to meet different detection requirements; according to Parseval's law, the FFT can be used to transform the continuous energy in the time domain into discrete energy in the frequency domain, and at the same time, the FFT is equivalent to a group of narrowband filters, thereby further improving the signal-to-noise ratio of the signal; finally, the amplitude and phase are calculated by the CORDIC algorithm, saving on-chip resources.

[0009] Compared with the prior art, the present invention has the following remarkable advantages: 1) The present invention can simultaneously perform filtering preprocessing on electromagnetic detection data in both the time domain and the frequency domain, improving the detection efficiency of the system and reducing the workload of subsequent data inversion; 2) The digital quadrature lock-in amplification algorithm used in the frequency-domain electromagnetic detection mode of the present invention mixes the sampled signal to near zero frequency instead of zero frequency, greatly reducing the influence of DC interference. Brief Description of the Drawings

[0010] Figure 1 It is a flowchart of the noise suppression technology for the time-frequency electromagnetic detection data of the present invention.

[0011] Figure 2 It is a specific implementation diagram of the noise suppression method for frequency-domain electromagnetic detection data.

[0012] Figure 3 It is a specific implementation diagram of the transient superposition of time-domain electromagnetic detection data. Specific Embodiments

[0013] The present invention will be further described below with reference to the accompanying drawings. A noise suppression method for time-frequency electromagnetic detection data of the present invention. The overall block diagram of the algorithm is as Figure 1 shown. In the time-domain electromagnetic detection mode, the AD sampling signal is a transient induction signal, and transient superposition and trace stacking are performed on it to improve the signal-to-noise ratio of the transient induction signal. In the frequency-domain electromagnetic detection mode, the AD sampling signal is a transient induction signal, which is mixed and shifted, and then the decimation factor is adjusted according to actual needs for narrowband filtering and decimation processing on the I / Q two channels of signals to improve the signal-to-noise ratio of the induction signal; then the filtered and decimated signal is transformed from continuous energy in the time domain to discrete energy in the frequency domain by FFT; finally, the amplitude and phase of the induction signal are obtained by using the CORDIC algorithm.

[0014] The present invention will be further described below with reference to embodiments.

[0015] First, select and set the parameters required for the time-domain electromagnetic detection mode or the frequency-domain electromagnetic detection mode in the host computer and send them to the FPGA through the network port.

[0016] In the time-domain electromagnetic detection mode, a square-wave pulse width of 2 ms is configured, the acquisition duration is 50 ms, and the number of trigger times is selected as 100 times, that is, the acquisition is repeated 100 times. During the period when the signal changes from high level to low level, the FPGA is started for acquisition and data processing. The FPGA performs transient superposition and averaging on the 100 acquisition data and stores them in the RAM. The schematic diagram is as shown in Figure 2 ; then the data is segmented according to logarithmic time, and the data within each time period is superposed and averaged, and the result is transmitted to the host computer. The actual measurement shows that the induced signal voltage that can be resolved by a single acquisition data without preprocessing is about 10 -4 V, and the induced signal voltage that can be resolved after superposition and averaging processing is about 10 5 V, and the signal-to-noise ratio is increased by 10 times.

[0017] In the frequency-domain electromagnetic detection mode, continuous frequency sweeping from low frequency to high frequency is configured at 20000 Hz - 49500 Hz (step size 500 Hz, a total of 60 frequency points). The FPGA can only calculate the amplitude and phase corresponding to one frequency point at a time. The AD sampling rate is 1.25 MHz, and a narrowband filter decimation module with a decimation factor of 1000 is selected. Taking f = 20000 Hz as an example to analyze the principle of the digital quadrature lock-in amplification algorithm, the algorithm flow chart is as shown in Figure 3 ; First, the sampling signal is mixed with the local oscillator signal with a frequency of 20019.53125 Hz, and a difference frequency component (useful signal) of 19.53125 Hz and a sum frequency component (high-frequency interference) of 40019.53125 Hz can be obtained; the high-frequency interference is filtered out by the narrowband filter decimation module with a decimation factor of 1000, and the in-phase component I and the quadrature component Q are sent to the FFT module; the FFT is used to convert the continuous energy in the time domain into discrete energy in the frequency domain to further improve the signal-to-noise ratio. Finally, the amplitude and phase of the FFT output result are calculated by the CORDIC algorithm. The actual measurement shows that after digital quadrature lock-in amplification processing, the resolution of the induced signal voltage can reach 1 μV.

Claims

1. A noise suppression method for time-frequency electromagnetic detection data, characterized in that: The FPGA receives the instructions from the host computer and parses them to perform corresponding signal processing on the digital signals after the analog signals are sampled by the AD. In the time-domain electromagnetic detection mode, transient superposition and trace superposition of the transient sampling signals can eliminate the influence of random noise on the signals, and the final output result is sent back to the host computer through the network port for graphical display. In the frequency-domain electromagnetic detection mode, the harmonic sampling signal is down-converted to shift its spectrum to a fixed position near zero frequency; then a narrowband filtering and decimation module composed of a cascaded structure of an integrator-comb filter (CIC), a half-band filter (HB), and a low-pass filter (LPF) is used to perform low-pass filtering on the mixed-frequency signal; then the low-pass filtered output signal is subjected to FFT processing to further improve the signal-to-noise ratio; finally, the amplitude and phase are calculated according to the FFT output result through the CORDIC algorithm and sent back to the host computer for display.

2. The noise suppression method for time-frequency electromagnetic detection data described in claim 1, characterized in that: The transient superposition and trace superposition in the time-domain electromagnetic detection mode and the digital quadrature lock-in amplification algorithm in the frequency-domain electromagnetic detection mode are all implemented inside the FPGA, providing high-quality detection data for subsequent inversion work and greatly improving the detection work efficiency.

3. The noise suppression method for time-frequency electromagnetic detection data described in claim 1, characterized in that: The transient superposition and trace superposition in the time-domain electromagnetic detection mode greatly improve the signal-to-noise ratio of the transient detection data.

4. The noise suppression method for time-frequency electromagnetic detection data described in claim 1, characterized in that: In the frequency-domain electromagnetic detection mode, compared with the traditional quadrature lock-in amplifier, the digital method can generate more accurate quadrature signals; mixing to near zero frequency instead of zero frequency reduces the influence of DC interference.

5. The noise suppression method for time-frequency electromagnetic detection data described in claim 1, characterized in that: In the frequency-domain electromagnetic detection mode, the narrowband filtering and decimation module adopts a parameter-configurable multi-stage decimation structure, which can self-select the decimation processing with different decimation multiples to meet different detection requirements; according to Parseval's theorem, the continuous energy in the time domain can be converted into discrete energy in the frequency domain by using FFT to further improve the signal-to-noise ratio; finally, the amplitude and phase are calculated through the CORDIC algorithm, saving on-chip resources.

Images