Distributed sound wave monitoring signal noise reduction method based on adaptive Wiener filtering

Abstract

The invention discloses a distributed sound wave monitoring signal noise reduction method based on adaptive Wiener filtering. The method comprises the following steps: extracting an effective sound wave signal of a current signal channel; preprocessing the effective sound wave signal of the current signal channel; converting the preprocessed time domain signal into a frequency domain signal; obtaining a logarithmic energy value of each frame of frequency domain signal; judging whether the current frame frequency domain signal is an initial frequency domain signal or not, if not, performing underground water flowing sound activity detection according to the logarithmic energy value and a calculated threshold value, and judging whether underground water flowing noise exists or not; calculating the posterior signal-to-noise ratio of the current frame frequency domain signal, and updating the prior signal-to-noise ratio by using the posterior signal-to-noise ratio; determining a gain function according to the prior signal-to-noise ratio, and performing filtering enhancement on the frequency domain signal by using the gain function to obtain a filtering enhanced frequency domain signal; and converting all the obtained frequency domain signals after filtering enhancement into time domain signals through inverse short-time Fourier transform, and outputting the obtained time domain signals as a noise reduction result.

Description

Technical Field

[0001] The invention belongs to the technical field of distributed underground fluid dynamics acoustic wave time domain signal noise reduction processing, and specifically relates to a distributed acoustic wave monitoring signal noise reduction method of adaptive Wiener filtering. Background Art

[0002] Due to the vertical heterogeneity of aquifers, the dynamic monitoring of groundwater in aquifers and the identification of water-producing layers still face great challenges. At present, the identification of water-producing layers is usually carried out through vertical borehole flow profile testing. Commonly used methods include segmented pumping / water pressure test method, flow logging method and salinization logging method. However, these existing vertical borehole flow profile test methods are not only complicated and time-consuming, but also unable to accurately identify the water-producing layer of the aquifer in the vertical direction. In order to facilitate long-term monitoring, it is necessary to use distributed fiber optic acoustic wave sensing to collect groundwater data under natural environment conditions. However, under natural environment conditions, distributed fiber optic acoustic wave sensing signals are often submerged in background noise, have time-varying characteristics and spatial variation characteristics, and have a low signal-to-noise ratio, resulting in a decrease in the accuracy of identifying water-producing signals under the submergence of background noise, which urgently needs to be improved. Summary of the invention

[0003] In view of the defects and shortcomings in the prior art, the present invention provides a distributed acoustic wave monitoring signal denoising method based on adaptive Wiener filtering to solve the technical problems of low signal-to-noise ratio and low accuracy in the prior art of collecting underground fluid dynamic acoustic wave data using distributed optical fiber acoustic wave sensors.

[0004] To achieve the above object, the present invention adopts the following technical scheme:

[0005] A distributed acoustic wave monitoring signal noise reduction method based on adaptive Wiener filtering comprises the following steps:

[0006] Step 1: using an acoustic wave sensor to collect underground fluid dynamic acoustic wave time domain signals from multiple signal channels distributed along the depth direction of the borehole, to obtain underground fluid dynamic acoustic wave time domain signals of the multiple signal channels; the interval between adjacent signal channels along the depth direction of the borehole is 0.1 to 3 meters;

[0007] Step 2: extracting the underground fluid dynamic acoustic wave time domain signal of the current signal channel to obtain the effective acoustic wave time domain signal of the current signal channel;

[0008] Step 3, performing frame-by-frame windowing processing on the effective acoustic wave time domain signal of the extracted current signal channel;

[0009] Step 4, initializing the effective acoustic wave time domain signal after the frame division and windowing process to obtain the noise spectrum mean and noise spectrum variance;

[0010] Step 5: Perform short-time Fourier transform on the current frame time domain signal, convert the current frame time domain signal into the current frame frequency domain signal, and obtain the amplitude spectrum of the current frame frequency domain signal;

[0011] Step 6: Perform square operation and frequency domain integration processing on the amplitude spectrum of the current frame frequency domain signal in sequence to obtain the energy value of the current frame frequency domain signal in the full frequency range; take the logarithm of the energy value to obtain the logarithmic energy value of the current frame frequency domain signal;

[0012] Step 7, determine whether the current frame frequency domain signal is the initial frame frequency domain signal, if not, proceed to step 8, if yes, proceed to step 9;

[0013] Step 8, performing underground fluid dynamic acoustic wave activity detection according to the logarithmic energy value and threshold value obtained in step 6, and judging whether there is underground fluid dynamic acoustic wave, if there is underground fluid dynamic acoustic wave, proceeding to step 9, if there is no underground fluid dynamic acoustic wave, recalculating the noise spectrum mean and noise spectrum variance, and then proceeding to step 9;

[0014] Step 9: Calculate the a posteriori signal-to-noise ratio of the frequency domain signal of the current frame according to the obtained noise spectrum variance, and update the a priori signal-to-noise ratio of the frequency domain signal of the current frame according to the calculated a posteriori signal-to-noise ratio of the frequency domain signal of the current frame to obtain an updated a priori signal-to-noise ratio of the current frame signal;

[0015] Step 10: determine the gain function according to the updated prior signal-to-noise ratio of the current frame frequency domain signal, and use the gain function to filter and enhance the amplitude spectrum of the current frame frequency domain signal to obtain the amplitude spectrum of the current frame frequency domain signal after filtering and enhancement; convert the current frame frequency domain signal after filtering and enhancement into a time domain signal through an inverse short-time Fourier transform, and use the obtained time domain signal as the denoised signal of the current frame;

[0016] Step 11, determine whether the current frame is the last frame of the current signal channel, if so, use the underground fluid dynamics acoustic wave time domain signal of the next signal channel as the underground fluid dynamics acoustic wave time domain signal of the current signal channel, and enter steps 2 to 10, if not, repeat steps 5 to 10 until the current frame is the last frame of the current signal channel, then use the underground fluid dynamics acoustic wave time domain signal of the next signal channel as the underground fluid dynamics acoustic wave time domain signal of the current signal channel, and enter steps 2 to 10;

[0017] Step 12, determining whether the current signal channel is the last channel among the multiple signal channels, if so, proceeding to step 13, if not, repeating steps 2 to 11 until the current signal channel is the last channel among the multiple signal channels, proceeding to step 13;

[0018] Step 13: Output the obtained denoised signals of all current frames as denoising results.

[0019] The present invention also has the following technical features:

[0020] Specifically, the logarithmic energy value described in step 6 is obtained by the following formula:

[0021]

[0022] in,

[0023] E(m) is the logarithmic energy value of the frequency domain signal of the current frame;

[0024] N is half of the sampling frequency of the acoustic sensor;

[0025] Y(m,f k ) is the amplitude value of the frequency domain signal of the current frame.

[0026] Furthermore, in step 8, it is determined whether there is an underground fluid dynamic sound wave according to the following criteria: if the logarithmic energy value>threshold value, it is determined that there is an underground fluid dynamic sound wave;

[0027] If the logarithmic energy value is less than or equal to the threshold, it is determined that there is no underground fluid dynamic sound wave.

[0028] Furthermore, the threshold value in step 8 is determined according to the following formula:

[0029]

[0030] In the formula,

[0031] η is the threshold value;

[0032] A is an empirical value, ranging from 2 to 3;

[0033] N is half of the sampling frequency of the acoustic sensor;

[0034] X(m,f k ) is the probability density function of the frequency domain signal of the mth frame in the presence of underground fluid dynamic acoustic waves;

[0035] λ μ (l,f k ) is the noise spectrum mean calculated last time.

[0036] Furthermore, the noise spectrum variance described in step 8 is calculated by the following formula:

[0037] λ μ (m,f k )=α·λ μ (l,f k )+(1-α)·P mm (m,f k )

[0038] In the formula,

[0039] λ μ (m,f k ) is the recalculated noise spectrum mean;

[0040] α is the forgetting factor, 0≤α≤1;

[0041] λ μ (l,f k ) is the noise spectrum mean calculated last time;

[0042] P mm (m,f k ) is the power of the frequency domain signal of the current frame.

[0043] Furthermore, the noise spectrum variance estimation described in step 8 is calculated by the following formula:

[0044] λ σ (m,f k )=α·λ σ (l,f k )+(1-α)·P mm (m,f k )

[0045] In the formula,

[0046] is the noise spectrum variance after the frequency domain signal of the current frame is updated;

[0047] α is the forgetting factor, 0≤α≤1;

[0048] λ σ (l,f k ) is the noise spectrum variance before the current frame frequency domain signal is updated;

[0049] P mm (m,f k ) is the power of the frequency domain signal of the current frame.

[0050] Furthermore, the a posteriori signal-to-noise ratio described in step 8 is determined by the following formula:

[0051]

[0052] γ(m,f k ) is the a posteriori signal-to-noise ratio of the frequency domain signal of the current frame;

[0053] λ σ (l,f k ) latest noise spectrum variance;

[0054] P mm (m,f k) is the power of the frequency domain signal of the current frame.

[0055] Furthermore, the updating of the priori signal-to-noise ratio in step 9 is implemented by the following formula:

[0056] ξ(m,f k )=β·γ(m-1,f k )+(1-β)max[γ(m,f k )-1,0]

[0057] in,

[0058] ξ(m,f k ) is the updated priori signal-to-noise ratio of the current frame frequency domain signal;

[0059] β is the update factor, 0≤β≤1;

[0060] γ(m,f k ) is the a posteriori signal-to-noise ratio of the frequency domain signal of the current frame;

[0061] γ(m-1,f k ) is the posterior signal-to-noise ratio of the frequency domain signal of the previous frame.

[0062] Furthermore, the gain function described in step 10 is:

[0063]

[0064] In the formula,

[0065] is the amplitude value of the frequency domain signal of the current frame after filtering and enhancement;

[0066] H(m,f k ) is the gain function;

[0067] Y(m,f k ) is the amplitude value of the frequency domain signal of the current frame.

[0068] Compared with the prior art, the present invention has the following technical effects:

[0069] (1) The method of the present invention adaptively adjusts the filter parameters through underground fluid dynamic sound activity detection, which can dynamically respond to signal and noise changes. The activity detection mechanism enables the filter to quickly adjust when the signal undergoes sudden changes, thereby maintaining a good noise reduction effect. The adaptive forgetting factor is adjusted according to energy changes. The introduction of the forgetting factor reduces the dependence on historical data, reduces the computational complexity, and improves the processing speed, so that the filter can quickly adapt to changes in the noise environment and enhance the ability to handle complex noise environments.

[0070] (2) The method of the present invention adopts activity detection based on Gaussian distribution, which can more accurately distinguish signals from noise, thereby improving the accuracy of noise estimation.

[0071] (3) The method of the present invention utilizes the dynamic update of the priori signal-to-noise ratio and the posterior signal-to-noise ratio to improve the noise reduction performance of the filter under different signal-to-noise ratio conditions, and can effectively suppress noise while retaining the details of the sound wave signal.

[0072] (4) The method of the present invention is adaptable to noises of different types and intensities, thereby improving the robustness and stability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 It is a schematic diagram of the process of the present invention;

[0074] Figure 2 A diagram of the single-channel filtering effect obtained in Example 1. DETAILED DESCRIPTION

[0075] The present invention is described in detail below in conjunction with the accompanying drawings and embodiments so that those skilled in the art can better understand the present invention. In the description of the embodiments of the present invention, the terms used in the following embodiments are only for the purpose of describing specific embodiments and are not intended to be limiting of the present application.

[0076] The terms "include", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized. The term "connected" includes direct and indirect connections, unless otherwise specified. "First" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features.

[0077] In the embodiments of the present invention, words such as "exemplarily" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplarily" or "for example" in the embodiments of the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplarily" or "for example" is intended to present related concepts in a specific way.

[0078] The method provided in the present application can be applied to a computer device, which can be a terminal or a server. It is understandable that the speech keyword recognition method provided in the present application can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server.

[0079] Example 1

[0080] This embodiment takes the identification of the groundwater yield layer section of the BK1 borehole in the South Ordos hydrogeological exploration in the central mining area of ​​Ordos Jungar, Inner Mongolia Autonomous Region, as an example, and uses a distributed acoustic sensing (DAS) system based on TGD-OFDR to collect distributed acoustic wave data for the BK1 borehole. The data acquisition equipment is the HiFi-DAS system of Pu Niu (Shanghai) Technology Co., Ltd., which uses optical fiber as a sensing medium. On the basis of optical reflectometer technology, by detecting the phase change of the Rayleigh scattered light signal, the amplitude and phase information of the sound wave (vibration) waveform at any position in the optical fiber is obtained. The system is based on the principle of coherent Rayleigh scattering, uses a single-mode optical cable as the front-end sensor, continuously injects laser pulse signals at one end of the optical cable, and uses the sensitive characteristics of optical fiber to sound waves (vibration). When the hydrodynamic sound waves in the hydrological exploration borehole act on the sensing optical fiber, the slight deformation of the optical fiber changes the spacing and refractive index of the scatterers inside the optical fiber, thereby causing the phase and intensity of the backscattered signal to change. By analyzing the phase and intensity of the backscattered signal, the deformation of the optical cable caused by the hydrodynamic sound waves in the hydrological exploration borehole can be detected, and all the sound wave (vibration) signals along the optical cable can be obtained at one time. The system has low noise and high fidelity characteristics throughout the entire length of the sensing optical fiber. Through data processing and interpretation, the water-producing layer section of the limestone aquifer in the borehole can be accurately identified, providing a geological basis for mine water prevention and control measures.

[0081] The drilling structure of this embodiment is shown in Table 1. Under environmental conditions, a distributed acoustic sensor is used to monitor groundwater discharge for 2 hours. The depth of the drilling test section is 785.66-875.53m, the test layer is the Ordovician Majiagou Formation (O2m), the test optical cable is 878m deep, the test data sampling frequency is 4000Hz, and one set of data is stored every 30s. 240 sets of effective differential acoustic wave data of continuous differential acoustic wave signals in the 2-hour monitoring period of the BK1 borehole are obtained, and the data size is more than 180G.

[0082] BK1 drilling structure

[0083]

[0084] Following the above technical solution, Figure 1 As shown, this embodiment provides a distributed acoustic wave monitoring signal noise reduction method based on adaptive Wiener filtering, comprising the following steps:

[0085] Step 1: using an acoustic wave sensor to collect underground fluid dynamic acoustic wave time domain signals from multiple signal channels distributed along the depth direction of the borehole, to obtain underground fluid dynamic acoustic wave time domain signals of the multiple signal channels; the interval between adjacent signal channels along the depth direction of the borehole is 0.1 to 3 meters;

[0086] Step 2: extracting the underground fluid dynamic acoustic wave time domain signal of the current signal channel to obtain the effective acoustic wave time domain signal of the current signal channel;

[0087] Step 3, performing frame-by-frame windowing processing on the effective acoustic wave time domain signal of the extracted current signal channel;

[0088] Step 4, initializing the effective sound wave signal after the frame segmentation and windowing process to obtain the noise spectrum mean and noise spectrum variance;

[0089] Step 5: Perform short-time Fourier transform on the current frame time domain signal, convert the current frame time domain signal into the current frame frequency domain signal, and obtain the amplitude spectrum of the current frame frequency domain signal; in the frequency domain, the amplitude spectrum is the set of amplitude values ​​of all frequency components of the signal.

[0090] Step 6: Perform square operation and frequency domain integration processing on the amplitude spectrum of the current frame frequency domain signal in sequence to obtain the energy value of the current frame frequency domain signal in the full frequency range; take the logarithm of the energy value to obtain the logarithmic energy value of the current frame frequency domain signal;

[0091] Specifically, the logarithmic energy value of the current frame (the mth frame) is determined by the following formula:

[0092]

[0093] in,

[0094] E(m) is the logarithmic energy value of the frequency domain signal of the current frame;

[0095] N is half of the sampling frequency of the acoustic sensor;

[0096] Y(m,f k ) is the amplitude value of the frequency domain signal of the current frame, that is, the amplitude value of the frequency domain signal of the current frame (the mth frame) after short-time Fourier transformation at the kth frequency.

[0097] Step 7, determine whether the current frame frequency domain signal is the initial frame frequency domain signal, if not, proceed to step 8, if yes, proceed to step 9;

[0098] Step 8, performing underground fluid dynamic acoustic wave activity detection according to the logarithmic energy value and threshold value obtained in step 6, and judging whether there is underground fluid dynamic acoustic wave, if there is underground fluid dynamic acoustic wave, proceeding to step 9, if there is no underground fluid dynamic acoustic wave, recalculating the noise spectrum mean and noise spectrum variance, and then proceeding to step 9;

[0099] By adaptively adjusting the filter parameters through groundwater sound activity detection, it is possible to dynamically respond to signal and noise changes. The activity detection mechanism enables the filter to adjust quickly when sudden changes occur in the signal, maintaining a good noise reduction effect.

[0100] Specifically, the presence of underground hydrodynamic sound waves is determined based on the following criteria:

[0101] If the logarithmic energy value is greater than the threshold, it is determined that there is an underground fluid dynamic sound wave;

[0102] If the logarithmic energy value is less than or equal to the threshold, it is determined that there is no underground fluid dynamic sound wave.

[0103] The threshold is determined according to the following formula:

[0104]

[0105] In the formula,

[0106] η is the threshold value;

[0107] A is an empirical value, ranging from 2 to 3;

[0108] N is half of the sampling frequency of the acoustic sensor;

[0109] X(m,f k ) is the probability density function of the frequency domain signal of the mth frame in the presence of underground fluid dynamic acoustic waves;

[0110] λ μ (l,f k ) is the noise spectrum mean calculated last time.

[0111] Preferably, the noise spectrum mean is calculated by the following formula:

[0112] λ μ (m,f k )=α·λ μ (l,f k )+(1-α)·P mm (m,f k )

[0113] In the formula,

[0114] λ μ (m,f k ) is the recalculated noise spectrum mean;

[0115] α is the forgetting factor, 0≤α≤1;

[0116] λ μ (l,f k ) is the noise spectrum mean calculated last time;

[0117] P mm (m,f k ) is the power of the frequency domain signal of the current frame.

[0118] Preferably, the noise spectrum variance estimate is calculated by the following formula:

[0119] λ σ (m,f k )=α·λ σ (l,f k )+(1-α)·P mm (m,f k )

[0120] In the formula,

[0121] λ σ (m,f k ) is the recalculated noise spectrum variance;

[0122] α is the forgetting factor, 0≤α≤1;

[0123] λ σ (l,f k ) is the noise spectrum variance calculated last time;

[0124] P mm (m,f k ) is the power of the frequency domain signal of the current frame.

[0126] The introduction of the forgetting factor reduces the dependence of the calculation on historical data, reduces the computational complexity, improves the processing speed, enables the filter to quickly adapt to changes in the noise environment, and enhances the ability to handle complex noise environments. 0≤α≤1, and the closer α is to 1, the greater the dependence of the noise spectrum mean of the current frame on the noise spectrum mean of the previous frame of the current frame, thereby achieving a smoother estimation. On the contrary, the closer α is to 0, the faster the noise spectrum mean reacts to new data and the worse the smoothing effect. In this embodiment, α is taken as 0.99.

[0127] Liveness detection based on Gaussian distribution can more accurately distinguish signals from noise, thereby improving the accuracy of noise estimation.

[0128] Step 9: Calculate the a posteriori signal-to-noise ratio of the frequency domain signal of the current frame according to the noise spectrum variance, and update the a priori signal-to-noise ratio of the frequency domain signal of the current frame according to the calculated a posteriori signal-to-noise ratio of the frequency domain signal of the current frame to obtain an updated a priori signal-to-noise ratio of the current frame signal;

[0129] Specifically, the posterior signal-to-noise ratio is determined by the following formula:

[0130]

[0131] γ(m,f k ) is the a posteriori signal-to-noise ratio of the frequency domain signal of the current frame;

[0132] λ σ (l,f k)The noise spectrum variance calculated last time;

[0133] P mm (m,f k ) is the power of the frequency domain signal of the current frame.

[0135] The update of the prior signal-to-noise ratio is achieved through the following formula:

[0136] ξ(m,f k )=β·γ(m-1,f k )+(1-β)max[γ(m,f k )-1,0]

[0137] In the formula,

[0138] ξ(m,f k ) is the updated priori signal-to-noise ratio of the current frame frequency domain signal;

[0139] β is the update factor, 0≤β≤1;

[0140] γ(m,f k ) is the a posteriori signal-to-noise ratio of the frequency domain signal of the current frame;

[0141] γ(m-1,f k ) is the posterior signal-to-noise ratio of the frequency domain signal of the previous frame.

[0142] The closer β is to 1, the more the updated noise spectrum mean depends on the noise spectrum mean of the previous frame of the current frame, thereby achieving a smoother estimation. On the contrary, the closer β is to 0, the more the updated noise spectrum mean depends on the noise power estimation of the current frame, and the response to new information is faster, but the smoothing effect is weakened. Therefore, in this embodiment, β is taken as 0.98.

[0143] Step 10: determine the gain function according to the updated prior signal-to-noise ratio of the current frame frequency domain signal, and use the gain function to filter and enhance the amplitude spectrum of the current frame frequency domain signal to obtain the amplitude spectrum of the current frame frequency domain signal after filtering and enhancement; convert the current frame frequency domain signal after filtering and enhancement into a time domain signal through an inverse short-time Fourier transform, and use the obtained time domain signal as the denoised signal of the current frame;

[0144] The gain function is:

[0145]

[0146] In the formula,

[0147] is the amplitude value of the frequency domain signal of the current frame after filtering and enhancement;

[0148] H(m,f k ) is the gain function;

[0149] Y(m,f k ) is the amplitude value of the frequency domain signal of the current frame.

[0150] Step 11, determine whether the current frame is the last frame of the current signal channel, if so, use the underground fluid dynamics acoustic wave time domain signal of the next signal channel as the underground fluid dynamics acoustic wave time domain signal of the current signal channel, and enter steps 2 to 10, if not, repeat steps 5 to 10 until the current frame is the last frame of the current signal channel, then use the underground fluid dynamics acoustic wave time domain signal of the next signal channel as the underground fluid dynamics acoustic wave time domain signal of the current signal channel, and enter steps 2 to 10;

[0151] Step 12, determining whether the current signal channel is the last channel among the multiple signal channels, if so, proceeding to step 13, if not, repeating steps 2 to 11 until the current signal channel is the last channel among the multiple signal channels, proceeding to step 13;

[0152] Step 13: Output the obtained denoised signals of all current frames as denoising results.

[0153] In this embodiment, the filtering effect of a single signal channel in the noise reduction process is as follows: Figure 2 shown.

[0154] from Figure 2 It can be seen from the original time series graph in that the original signal may contain obvious noise, which may interfere with the accurate detection of hydrodynamic sound waves. The hydrodynamic sound waves may be masked by background noise and difficult to observe directly. After noise reduction processing, the background noise level of the signal is significantly reduced, the characteristics of the hydrodynamic sound waves are more clearly visible, and the processed signal is clearer, which is helpful for subsequent signal analysis and feature extraction. By comparing the original signal and the signal after noise reduction processing, it can be seen that the method of the present invention can effectively suppress background noise while retaining the main characteristics of hydrodynamic sound waves. Applying the method of the present invention to the real-time monitoring and analysis of groundwater flow can effectively improve the accuracy and reliability of the monitoring system.

[0155] The above implementation process is only an example for clearly explaining the present application, and is not intended to limit the implementation methods. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from this are still within the scope of protection of this application.

Claims

1. A distributed acoustic wave monitoring signal denoising method based on adaptive Wiener filtering, characterized in that: The following steps are involved: Step 1: using an acoustic wave sensor to collect underground fluid dynamic acoustic wave time domain signals from multiple signal channels distributed along the depth direction of the borehole, to obtain underground fluid dynamic acoustic wave time domain signals of the multiple signal channels; the interval between adjacent signal channels along the depth direction of the borehole is 0.1 to 3 meters; Step 2: extracting the underground fluid dynamic acoustic wave time domain signal of the current signal channel to obtain the effective acoustic wave time domain signal of the current signal channel; Step 3, performing frame-by-frame windowing processing on the effective acoustic wave time domain signal of the extracted current signal channel; Step 4, initializing the effective acoustic wave time domain signal after the frame division and windowing process to obtain the noise spectrum mean and noise spectrum variance; Step 5: Perform short-time Fourier transform on the current frame time domain signal, convert the current frame time domain signal into the current frame frequency domain signal, and obtain the amplitude spectrum of the current frame frequency domain signal; Step 6: Perform square operation and frequency domain integration processing on the amplitude spectrum of the current frame frequency domain signal in sequence to obtain the energy value of the current frame frequency domain signal in the full frequency range; take the logarithm of the energy value to obtain the logarithmic energy value of the current frame frequency domain signal; Step 7, determine whether the current frame frequency domain signal is the initial frame frequency domain signal, if not, proceed to step 8, if yes, proceed to step 9; Step 8, performing underground fluid dynamic acoustic wave activity detection according to the logarithmic energy value and threshold value obtained in step 6, and judging whether there is underground fluid dynamic acoustic wave, if there is underground fluid dynamic acoustic wave, proceeding to step 9, if there is no underground fluid dynamic acoustic wave, recalculating the noise spectrum mean and noise spectrum variance, and then proceeding to step 9; Step 9: Calculate the a posteriori signal-to-noise ratio of the frequency domain signal of the current frame according to the noise spectrum variance, and update the a priori signal-to-noise ratio of the frequency domain signal of the current frame according to the calculated a posteriori signal-to-noise ratio of the frequency domain signal of the current frame to obtain an updated a priori signal-to-noise ratio of the current frame signal; Step 10: determining a gain function according to the updated priori signal-to-noise ratio of the current frame frequency domain signal, and using the gain function to filter and enhance the amplitude spectrum of the current frame frequency domain signal to obtain the amplitude spectrum of the current frame frequency domain signal after filtering and enhancement; The frequency domain signal of the current frame after filtering and enhancement is converted into a time domain signal by inverse short-time Fourier transform, and the obtained time domain signal is used as the denoised signal of the current frame; Step 11, determine whether the current frame is the last frame of the current signal channel, if so, use the underground fluid dynamics acoustic wave time domain signal of the next signal channel as the underground fluid dynamics acoustic wave time domain signal of the current signal channel, and enter steps 2 to 10, if not, repeat steps 5 to 10 until the current frame is the last frame of the current signal channel, then use the underground fluid dynamics acoustic wave time domain signal of the next signal channel as the underground fluid dynamics acoustic wave time domain signal of the current signal channel, and enter steps 2 to 10; Step 12, determining whether the current signal channel is the last channel among the multiple signal channels, if so, proceeding to step 13, if not, repeating steps 2 to 11 until the current signal channel is the last channel among the multiple signal channels, proceeding to step 13; Step 13: Output the obtained denoised signals of all current frames as denoising results.

2. The distributed acoustic wave monitoring signal denoising method of adaptive Wiener filtering according to claim 1, characterized in that: The logarithmic energy value described in step 6 is obtained by the following formula: in, E(m) is the logarithmic energy value of the frequency domain signal of the current frame; N is half of the sampling frequency of the acoustic sensor; Y(m,f k ) is the amplitude value of the frequency domain signal of the current frame.

3. The distributed acoustic wave monitoring signal denoising method of adaptive Wiener filtering as claimed in claim 1, characterized in that: In step 8, the presence of underground fluid dynamic sound waves is determined according to the following criteria: If the logarithmic energy value is greater than the threshold, it is determined that there is an underground fluid dynamic sound wave; If the logarithmic energy value is less than or equal to the threshold, it is determined that there is no underground fluid dynamic sound wave.

4. The distributed acoustic wave monitoring signal denoising method of adaptive Wiener filtering according to claim 1, characterized in that: The threshold described in step 8 is determined according to the following formula: In the formula, η is the threshold value; A is an empirical value, ranging from 2 to 3; N is half of the sampling frequency of the acoustic sensor; X(m,f k ) is the probability density function of the frequency domain signal of the mth frame in the presence of underground fluid dynamic acoustic waves; λ μ (l,f k ) is the noise spectrum mean calculated last time.

5. The distributed acoustic wave monitoring signal denoising method of adaptive Wiener filtering according to claim 1, characterized in that: The noise spectrum variance of the current frame described in step 8 is calculated by the following formula: l μ (m,f k )=a·l μ (l,f k )+(1-a)·P mm (m,f k ) In the formula, λ μ (m,f k ) is the recalculated noise spectrum mean; α is the forgetting factor, 0≤α≤1; λ μ (l,f k ) is the noise spectrum mean calculated last time; P mm (m,f k ) is the power of the frequency domain signal of the current frame.

6. The distributed acoustic wave monitoring signal denoising method of adaptive Wiener filtering according to claim 1, characterized in that: The noise spectrum variance estimate described in step 8 is calculated by the following formula: l σ (m,f k )=a·l σ (l,f k )+(1-a)·P mm (m,f k ) In the formula, λ σ (m,f k ) is the recalculated noise spectrum variance; α is the forgetting factor, 0≤α≤1; λ σ (l,f k ) is the noise spectrum variance calculated last time; P mm (m,f k ) is the power of the frequency domain signal of the current frame.

7. The distributed acoustic wave monitoring signal denoising method of adaptive Wiener filtering according to claim 1, characterized in that: The a posteriori signal-to-noise ratio described in step 8 is determined by the following formula: in, γ(m,f k ) is the a posteriori signal-to-noise ratio of the frequency domain signal of the current frame; λ σ (l,f k )The noise spectrum variance calculated last time; P mm (m,f k ) is the power of the frequency domain signal of the current frame.

8. The distributed acoustic wave monitoring signal denoising method of adaptive Wiener filtering according to claim 1, characterized in that: The updating of the priori signal-to-noise ratio described in step 9 is achieved by the following formula: ξ(m,f k )=β·γ(m-1,f k )+(1-β)max[γ(m,f k )-1.0] In the formula, ξ(m,f k ) is the updated priori signal-to-noise ratio of the current frame frequency domain signal; β is the update factor, 0≤β≤1; γ(m,f k ) is the a posteriori signal-to-noise ratio of the frequency domain signal of the current frame; γ(m-1,f k ) is the posterior signal-to-noise ratio of the frequency domain signal of the previous frame.

9. The distributed acoustic wave monitoring signal denoising method of adaptive Wiener filtering according to claim 1, characterized in that: The gain function described in step 9 is: In the formula, H(m,f k ) is the gain function; ξ(m,f k ) is the updated prior signal-to-noise ratio of the frequency domain signal of the current frame.

10. The distributed acoustic wave monitoring signal denoising method of adaptive Wiener filtering according to claim 1, characterized in that: The filtering and enhancement of the current frame frequency domain signal by using the gain function in step 10 is achieved by the following formula: In the formula, is the amplitude value of the frequency domain signal of the current frame after filtering and enhancement; H(m,f k ) is the gain function; Y(m,f k ) is the amplitude value of the frequency domain signal of the current frame.

Images