CN116755165A - Ocean wave induction magnetic interference suppression and evaluation method based on time sequence matrixing - Google Patents

Abstract

The invention provides a method for suppressing and evaluating sea wave induction magnetic interference based on time sequence matrixing, which comprises the following steps: reading a marine magnetotelluric field time sequence; estimating noise by adopting a combination of variational modal decomposition and band-pass filtering; constructing a time sequence matrix and a noise matrix based on a phase space reconstruction matrixing method; MNF transformation is carried out on the data matrix; selecting a high-order MNF component to reconstruct a noise matrix and performing inverse matrixing to obtain a noise time sequence; band-pass filtering is carried out on the noise time sequence, and sea wave induction magnetic noise is extracted to realize signal-noise separation; performing noise suppression effect evaluation based on the autocorrelation function; judging whether the requirements are met; a time series of the suppressed noise is output. The interference suppression method provides an effective solution for suppressing noise which is aliased with effective signals in time domain and frequency domain, and improves the signal-to-noise ratio of data. The noise suppression effect evaluation method based on the autocorrelation function can be used for actually measured data, and has wider application range and practicability compared with the traditional method.

Description

Wave-induced magnetic interference suppression and evaluation method based on time series matrixing

Technical field

The invention relates to the technical field of marine geophysical signal processing, and specifically relates to a method for suppressing and evaluating wave-induced magnetic interference based on time series matrixing.

Background technique

Marine magnetotellurics (MT) obtains the electrical distribution of the seafloor medium by observing the induced electromagnetic field generated by natural field sources on the seafloor. It is an important geophysical method for studying the electrical structure of the oceanic crust and upper mantle as well as deep geological processes. In a shallow water environment, the frequency range of the induced electromagnetic field interference generated by the movement of waves overlaps with the frequency of the MT signal, and it is difficult to separate it from the MT signal in the time domain and frequency domain; the electromagnetic interference induced by the waves has strong energy, which greatly reduces The signal-to-noise ratio of MT data is reduced, resulting in severe distortion of apparent resistivity and phase, which is the main interference of shallow water MT data. Therefore, it is necessary to suppress the wave-induced electromagnetic field to improve the signal-to-noise ratio of MT data in shallow water areas.

Traditional processing methods include wavelet transform, SVD decomposition, etc. Due to the aliasing of noise and effective signals in the time domain and frequency, the above methods are prone to damage the effective signals, resulting in unsatisfactory denoising effects. Yu Caixia et al. (2010) and Zhang Baoqiang (2018) respectively used EMD and wavelet threshold denoising methods to suppress wave-induced magnetic noise in MT data. This type of signal decomposition algorithm can effectively improve the apparent resistivity curve, but the correction effect of the phase curve is still Less than ideal. Feng Changqing et al. (2022) used the K-SVD dictionary learning algorithm to extract wave-induced magnetic noise. The improvement effect on the apparent resistivity distortion was obvious, but the phase distortion still needs to be corrected twice with the corrected apparent resistivity, and the processing effect also needs to be Further improve. In terms of denoising effect evaluation, existing methods usually calculate the Signal-to-Noise Ratio (SNR), Root-Mean-Square Error (RMSE), and Correlation Coefficient (Correlation Coefficient) in the time domain. CORC) and other indicators to evaluate the denoising effect (Wang et al., 2017; Feng Changqing et al., 2022). Such indicators need to be obtained by calculating the noise-free signal and the denoised signal. However, the noise-free signal of the measured data is unknown , so this type of index cannot be used to evaluate the denoising effect of measured data. How to evaluate the application effect of noise suppression methods in measured data has always been a hot issue discussed by geophysicists.

Contents of the invention

The purpose of this invention is to provide a method for suppressing and evaluating wave-induced magnetic interference based on time series matrixing. This method can realize matrixing of single-channel time series data and improve the signal-to-noise ratio of ocean magnetotelluric data. The proposed method is based on The noise suppression effect evaluation method of autocorrelation function can be used to evaluate the application effect of wave magnetic noise processing method in measured data.

In order to achieve the above objectives, the present invention provides the following technical solutions:

1. A method for suppressing and evaluating wave-induced magnetic interference based on time series matrixing, which is characterized by including:

S1. Read the evaluation requirements and the original time series x(N) of the ocean magnetotelluric field, where N is the number of data;

S2. Use the method of combining variational mode decomposition and band-pass filtering to estimate the noise of the time series x(N), and obtain the noise time series x _n ;

S3. Based on the phase space reconstruction matrix method, samples are extracted from the original time series x(N) as one-dimensional phase space vectors to construct the original time series matrix X _S :

In the above formula, the relationship between the number of rows m, the number of columns n and the time delay τ of the matrix and the total number of data points N of the time series satisfies N=(m-1)τ+n. For the time delay τ, its value range is means rounding down;

S4. Matrix the estimated noise time series x _n to obtain the noise matrix X _N ;

S5. Perform MNF transformation based on the original time series matrix X _S and noise matrix X _N to obtain the MNF component Z;

S6. Select appropriate high-order MNF components to reconstruct the noise matrix based on noise characteristics.

S7. Reconstruct the noise matrix based on the time delay τ pair Perform inverse matrixization to obtain the extracted noise time series;

S8. Estimate the wave frequency band, perform band-pass filtering on the extracted noise time series, extract the wave-induced magnetic noise, and subtract the extracted wave-induced magnetic noise from the original time series;

S9. Calculate the autocorrelation function based on the time series after suppressing the noise, and evaluate the suppression effect of wave-induced magnetic noise based on the autocorrelation function. The calculation method of the autocorrelation function is:

In the above formula, y _t is the observed value of the time series at time t, is the average of all observations in the time series, k is the lag time, and its value range is [0, N-1]. The autocorrelation function ACF refers to the sequence composed of autocorrelation coefficients r _k ;

S10. Determine whether the evaluation requirements are met based on the autocorrelation function. If so, go to S11; if not, go to S3;

S11. Output the noise-suppressed magnetotelluric time series.

2. The wave-induced magnetic interference suppression and evaluation method based on time series matrixing as claimed in claim 1, characterized in that the calculation method of MNF component Z is:

Z ₌ ^RTXS

In the formula, X _S is the original time series matrix, and R ^T is the structural rotation matrix calculated based on the noise matrix X _N.

3. The wave-induced magnetic interference suppression and evaluation method based on time series matrixing as claimed in claim 1, characterized in that the reconstructed signal The calculation method is:

In the formula, B is the angular truncation matrix of m rows and m columns, which is expressed as 1 on the corresponding row of the MNF component participating in the reconstruction, and the remaining parts are all 0.

4. The wave-induced magnetic interference suppression and evaluation method based on time series matrixing as claimed in claim 1, characterized in that the reconstructed noise matrix The calculation method for inverse matrixing is based on the time delay τ, sequentially eliminating duplicate data points for each sample in the data matrix, and sequentially splicing them into a single-channel time series.

5. The wave-induced magnetic interference suppression and evaluation method based on time series matrixing as claimed in claim 1, characterized in that the noise suppression effect evaluation method based on the autocorrelation function can be applied to evaluate specific frequency characteristics in unstable signals. The suppression effect of the signal can be applied to the evaluation of measured data.

Compared with the prior art, the beneficial effects of the method provided in some embodiments of the present invention are:

This invention mainly aims at the problem of strong interference to MT magnetic field signals caused by ocean wave motion in ocean magnetotelluric detection, which leads to low signal-to-noise ratio of magnetotelluric data. It proposes a wave-induced magnetic interference suppression method based on time series matrixing. This method provides an effective technical solution for improving the signal-to-noise ratio of marine MT data in complex shallow water environments, and can more effectively correct phase distortion. Compared with the traditional electromagnetic noise suppression method, this method is based on the phase space reconstruction matrixing method to achieve the matrixing of single-channel time series, achieves the separation of signals and noise through the minimum noise separation method, and selects appropriate high-order MNFs based on noise characteristics. The components reconstruct the noise matrix, and through inverse matrixization, the extracted noise time series is obtained. The proposed wave magnetic interference suppression method provides an effective solution for suppressing the noise that overlaps with the effective signal in the time domain and frequency domain, and improves the signal-to-noise ratio of the electromagnetic data.

Aiming at the problem that common indicators cannot be used to evaluate the denoising effect of measured data, a noise suppression effect evaluation method based on autocorrelation function is proposed. This method combines the non-stationary characteristics of the MT signal and the periodicity of wave-induced magnetic noise, and uses two The effect of signal-to-noise separation is judged by the periodic difference between them. The proposed noise suppression effect evaluation method based on autocorrelation function does not require a known noise-free signal, and the denoising effect is judged by the oscillation shape of the curve. Compared with the traditional evaluation method, this type of index can be used to evaluate the denoising effect of measured data. The application scope and practicality of the evaluation method are improved.

Description of the drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings in the following description are only illustrative of the present invention. For some embodiments, those of ordinary skill in the art can obtain other drawings based on these drawings without exerting any creative effort.

Figure 1 is a program flow diagram of the method of the present invention;

Figure 2 is a schematic structural diagram of the one-dimensional geoelectric model;

Figure 3 shows the time series and power spectral density diagram of simulated MT data before and after adding noise;

Figure 4 shows the time series and power spectral density diagram before and after denoising the simulated MT data;

Figure 5 is a diagram of the autocorrelation function of the simulated MT data denoising effect;

Figure 6 shows the apparent resistivity and phase curves after interference suppression.

Detailed ways

In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention and are not intended to limit the present invention.

Example 1

Referring to Figure 1, a wave-induced magnetic interference suppression and evaluation method based on time series matrixing includes the following steps:

S1. Read the original time series x(N) of the oceanic magnetotelluric field, N is the number of data, read the evaluation criteria, and the periodic characteristic threshold based on the autocorrelation function;

S2. Use the method of combining variational mode decomposition and band-pass filtering to estimate the noise of the time series x(N), and obtain the noise time series x _n ;

S3. Based on the phase space reconstruction matrix method, samples are extracted from the original time series x(N) as one-dimensional phase space vectors to construct the original time series matrix X _S :

In the above formula, the relationship between the number of rows m, the number of columns n and the time delay τ of the matrix and the total number of data points N of the time series satisfies N=(m-1)τ+n. For the time delay τ, its value range is means rounding down;

S4. Matrix the estimated noise time series x _n to obtain the noise matrix X _N ;

S5. Perform MNF transformation based on the original time series matrix X _S and noise matrix X _N to obtain the MNF component Z;

S6. Select appropriate high-order MNF components to reconstruct the noise matrix based on noise characteristics.

S7. Reconstruct the noise matrix based on the time delay τ pair Perform inverse matrixization to obtain the extracted noise time series;

S8. Estimate the wave frequency band, perform band-pass filtering on the extracted noise time series, extract the wave-induced magnetic noise, and subtract the extracted wave-induced magnetic noise from the original time series;

S9. Calculate the autocorrelation function based on the time series after suppressing the noise, and evaluate the suppression effect of wave-induced magnetic noise based on the autocorrelation function. The calculation method of the autocorrelation function is:

In the above formula, y _t is the observed value of the time series at time t, is the average of all observations in the time series, k is the lag time, and its value range is [0, N-1]. The autocorrelation function ACF refers to the sequence composed of autocorrelation coefficients r _k ; if there is wave-induced magnetic noise If affected, the ACF curve of the MT time series shows certain periodic fluctuations; if the wave-induced magnetic noise is suppressed, the periodic fluctuations in the ACF curve are reduced and appear as a smoother curve. Based on the above characteristics, ACF can be used to evaluate the denoising effect of measured MT data;

S10. Determine whether the evaluation requirements are met based on the autocorrelation function. If so, go to S11; if not, go to S3;

S11. Output the noise-suppressed magnetotelluric time series.

Example 2

Refer to Figure 2, which is a one-dimensional geoelectric model diagram. In order to verify the feasibility of suppressing the measured wave-induced magnetic interference proposed in this invention, the magnetic field simulated by the one-dimensional ocean geoelectric model shown in Figure 2 is used as the effective magnetic signal, and the synthesized wave-induced magnetic field is used as noise and mixed into the effective signal to achieve Wave-induced magnetic interference affects the aliasing of effective signals and noise in the frequency band. The two-dimensional random wave-induced magnetic field formula is used to simulate wave-induced magnetic noise. Assume that the wave spectrum is a PM spectrum, the geomagnetic field intensity is 48000nT, the seawater layer thickness is 25m, the magnetic declination angle is 60°, the magnetic inclination angle is 45°, the seawater resistivity is 0.3Ω·m, and the wind speed is 13m/s, the simulation results in a sampling rate 10Hz, 1h wave-induced horizontal magnetic field time series. The simulated wave-induced magnetic noise is added to the simulated ocean magnetotelluric component to obtain synthetic MT data. Figure 3 shows the MT time series and power spectral density before and after adding noise. It can be seen from the figure that the power spectral density of the MT data after adding noise has a strong energy anomaly bulge near 0.1Hz. This anomaly is the wave-induced magnetic noise. Figure 4 is the denoising result of the method of the present invention. It can be seen from the figure that the time series denoised by MNF has a high degree of fit with the noise-free MT signal. MNF can better protect the MT signal while suppressing the magnetic noise induced by waves, showing strong signal-to-noise separation capabilities.

Figure 5 is an autocorrelation function ACF diagram to evaluate the denoising effect of synthetic MT data. The diagram shows the ACF curve of MT data before and after denoising and the simulated wave-induced magnetic noise. The dotted line in the diagram indicates the peak of the ACF curve of the synthetic MT data before denoising. The lag time k, the sampling rate of the synthetic MT data is known to be f = 10 Hz, and the time interval between the two peaks Δt = Δk/f is calculated. It can be seen from the figure that the ACF curve simulating the magnetic noise induced by sea waves shows obvious periodic fluctuation characteristics and is more obvious at small lag times, while the ACF curve of the noiseless MT signal is a smooth curve that decreases gently. Under the influence of wave-induced magnetic noise, the ACF curve of the synthesized MT data shows corresponding periodic fluctuation characteristics, and the period of peak occurrence is within the period range of wave-induced magnetic noise (6-16 s). The ACF curve of the MNF processing result almost completely coincides with the ACF curve of the noiseless MT signal, and the periodic fluctuation characteristics disappear, indicating that MNF suppresses the magnetic noise induced by sea waves while causing less damage to the MT signal.

Robust impedance estimation was performed on the synthetic MT data before and after suppressing the wave-induced magnetic noise, and the apparent resistivity curve and phase curve were calculated as shown in Figure 6. It can be seen from the figure that for the synthetic MT data with added wave-induced magnetic noise, the apparent resistivity and phase curves are distorted in the 6-16s period range, significantly deviating from the true values. The apparent resistivity curve and phase curve obtained after MNF processing are smoother and more continuous, and in the frequency band affected by wave interference, they are numerically closer to the true value. The simulation data processing results show that the algorithm proposed in the present invention can effectively suppress the measured wave-induced magnetic field mixed in the magnetic field component, and finally obtain more reliable apparent resistivity and phase curves by estimating the impedance. This also shows that the algorithm proposed by the present invention is effective.

The above descriptions are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention shall be included in the protection of the present invention. within the range.

Claims (5)

1. The method for suppressing and evaluating the sea wave induction magnetic interference based on the time sequence matrixing is characterized by comprising the following steps of:

s1, reading an evaluation requirement and an original time sequence x (N) of an ocean magnetotelluric field, wherein N is the number of data;

s2, carrying out noise estimation on the time sequence x (N) by adopting a method of combining variation modal decomposition and band-pass filtering to obtain a noise time sequence x _n ；

S3, based on a phase space reconstruction matrixing method, extracting samples from an original time sequence X (N) as a one-dimensional phase space vector to construct the original time sequence matrix X _S ：

In the above formula, N is satisfied between the number of rows m, the number of columns N, the time delay τ and the total data point number N of the time sequence= (m-1) τ+n, which has a value in the range of time delay τ Representing a downward rounding;

s4, estimating noise time sequence x _n Matrixing to obtain a noise matrix X _N ；

S5, based on original time sequence matrix X _S And noise matrix X _N Performing MNF conversion to obtain MNF component Z;

s6, selecting proper high-order MNF components to reconstruct a noise matrix based on noise characteristics

S7, reconstructing a noise matrix based on the time delay tauPerforming inverse matrixing to obtain an extracted noise time sequence;

s8, estimating an ocean wave frequency band, carrying out band-pass filtering on the extracted noise time sequence, extracting ocean wave induced magnetic noise, and subtracting the extracted ocean wave induced magnetic noise from the original time sequence;

s9, calculating an autocorrelation function based on the time sequence after noise suppression, and evaluating the suppression effect of the sea wave induction magnetic noise based on the autocorrelation function, wherein the autocorrelation function calculating method comprises the following steps:

in the above, y _t Is the observation of the time series at time t,is all views of a time sequenceThe average value of the measured values, k is the lag time, and the value range is [0, N-1]The autocorrelation function ACF is defined as the autocorrelation coefficient r _k A sequence of constructs;

s10, judging whether the evaluation requirement is met based on the autocorrelation function, if so, turning to S11, and if not, turning to S3;

s11, outputting a magnetotelluric time sequence for suppressing noise.

2. The method for suppressing and evaluating sea wave induced magnetic interference based on time series matrixing as recited in claim 1, wherein the method for calculating the MNF component Z is as follows:

Z＝R ^T X _S

wherein X is _S R is the original time sequence matrix ^T Is based on noise matrix X _N And (5) calculating a construction rotation matrix.

3. The method for suppressing and evaluating sea wave induced magnetic interference based on time series matrixing as recited in claim 1, wherein the signals are reconstructedThe calculation method of (1) is as follows:

where B is an angular truncated matrix for m rows and m columns, and is represented by 1 on the corresponding row of MNF components involved in reconstruction, and the rest is all 0.

4. A method of suppressing and evaluating sea wave induced magnetic interference based on time series matrixing as recited in claim 1, wherein the reconstructed noise matrixThe calculation method for inverse matrixing is based on time delay tau, and each data matrix is eliminated in turnThe repeated data points of the sample are spliced into a time series of single channels in turn.

5. The method for suppressing and evaluating sea wave induced magnetic interference based on time series matrixing as recited in claim 1, wherein the method for evaluating noise suppressing effect based on autocorrelation function is applicable to evaluating suppressing effect of signals with specific frequency characteristics in unstable signals and to evaluating measured data.