An adaptive noise reduction filtering method, system, medium and device

Abstract

The application discloses a kind of self-adapting noise reduction filtering method, system, medium and equipment, the method includes to input dynamic signal and carry out direct current offset preprocessing;Time-frequency representation matrix is generated by time-frequency transform, and amplitude spectrum is calculated accordingly;Along time axis analysis amplitude spectrum, extract short time and long time noise amplitude and fusion generate basic noise amplitude;Signal noise amplitude ratio is calculated based on basic noise amplitude, and final noise amplitude is determined;According to final noise amplitude, calculate noise reduction spectrum amplitude, retain original phase information, reconstruct as noise reduction time domain signal by inverse transform;The system includes preprocessing module, amplitude spectrum generation module, double time scale noise estimation module, denoising module, noise reduction and signal reconstruction module;The application is accurately described by double time scale noise estimation Dynamic distribution of noise, combined with the adaptive noise reduction adjustment based on signal noise amplitude ratio, effectively suppress complex, non-stationary noise, while enhancing the relative intensity of characteristic frequency band.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of signal processing, in particular to an adaptive noise reduction filtering method, system, medium and device. BACKGROUND

[0002] Dynamic signals have wide applications in industrial, energy, medical and communication fields, such as mechanical equipment running sound, biomedical signals (such as electrocardiogram or electroencephalogram), wireless communication signals, etc. These signals usually contain rich state information, which can reflect the health condition of equipment, the running characteristics of system or biological activities. However, dynamic signals are easily disturbed by complex environmental noise, which has non-stationary, wide-spectrum and time-varying characteristics, often mixed with the spectrum of target signals, significantly reducing the accuracy of signal analysis. For example, in industrial environment, mechanical vibration, turbulent noise, transient interference and background noise are highly overlapped with target signals, making it difficult to extract features and limiting the application of signals in condition monitoring and fault diagnosis.

[0003] Traditional noise reduction methods, such as spectral subtraction, Wiener filtering and wavelet denoising, rely on static noise models or fixed parameters, which are difficult to adapt to the non-stationary characteristics of dynamic noise, resulting in signal quality degradation or distortion of key features. Spectral subtraction suppresses stationary noise by estimating the noise power spectrum, but it is easy to introduce artifacts in non-stationary scenarios; Wiener filtering is based on the least mean square error criterion, and its performance is limited when dynamic noise changes rapidly; although wavelet denoising provides time-frequency resolution through multi-scale decomposition, improper threshold selection can lead to feature distortion in complex noise environment, especially when noise and target signal spectrum overlap, the separation efficiency is low.

[0004] Signal decomposition technology is an important direction of dynamic signal processing, which supports feature extraction and noise suppression by decomposing the signal into IMF components. However, empirical mode decomposition (EMD) is affected by mode mixing and end effect, generating false components under strong noise; variational mode decomposition (VMD) is more robust, but the parameter adjustment is complex and the computational cost is high, which conflicts with the real-time application requirements; non-negative matrix factorization (NMF) and wavelet packet denoising can extract periodic features in specific scenarios, but face strong noise interference, the accuracy decreases and the computational overhead is large; local mean decomposition (LMD) and singular spectrum analysis (SSA) are effective, but they are highly sensitive to noise and difficult to handle large-scale dynamic signals.

[0005] In recent years, with the increasing demand for non-stationary signal processing, dynamic noise reduction technology has made progress, such as adaptive weighted mean square filtering (AWMF) combined with gradient descent enhancement, sparse attention denoising, and deep learning-based noise estimation, which perform well in signal enhancement or semi-stationary noise scenarios. However, these methods lack adaptability in complex industrial noise environments. Deep learning methods are particularly difficult to meet the low latency requirements of real-time monitoring due to their reliance on large amounts of training data and high computational resources. Moreover, the time-frequency variation of dynamic noise poses higher challenges to the dynamic response and robustness of algorithms.

[0006] In addition, traditional noise reduction methods are prone to introducing false signals or overfitting noise, reducing reliability. While new methods such as neural Kalman filtering, recursive denoising learning, and wavelet denoising based on Mahalanobis distance have made progress in stationary or semi-stationary scenarios, they lack sufficient response to non-stationary noise and limited signal feature protection capabilities, making it difficult to suppress noise while avoiding interference with key features. The versatility needs to be improved.

[0007] Therefore, there is an urgent need for an adaptive noise reduction method that can effectively suppress dynamic noise while strengthening the relative intensity of feature bands to meet the needs of diverse signal processing scenarios. SUMMARY

[0008] Therefore, the present application provides an adaptive noise reduction filtering method, system, medium and device to at least solve the problem of existing noise reduction methods interfering with key signal features while suppressing noise and multi-scene adaptability.

[0009] To achieve the above purpose, the present application adopts the following technical solutions:

[0010] An adaptive noise reduction filtering method, comprising the following steps:

[0011] S1. Obtain a dynamic signal from a signal receiving device , and obtain a corresponding time-domain signal after preprocessing, wherein is the signal length;

[0012] S2. Obtain a corresponding time-frequency representation matrix by time-frequency transformation, which is used to represent the complex amplitude of the th frame and the th frequency point, and calculate the amplitude spectrum according to ;

[0013] S3. Adopt a double-time-scale noise estimation method to analyze the amplitude spectrum along the time axis, extract the short-time noise amplitude and the long-time noise amplitude , and introduce a dynamic weight Will and Perform fusion to generate basic noise amplitude ;

[0014] S4. Based on Calculate the signal-to-noise ratio ,according to Calculate the minimum amplitude threshold And based on dynamic noise scaling factor as well as Calculate and estimate noise amplitude ;

[0015] S5. Based on and Calculate the noise reduction spectrum amplitude Preserving the original phase information, the inverse transform is used to... Reconstructed into a denoised time-domain signal .

[0016] Preferably, the preprocessing in S1 includes the following:

[0017] Acquire dynamic signals from signal receiving devices. ,right DC offset removal preprocessing is performed to remove low-frequency interference, resulting in the preprocessed time-domain signal. .

[0018] Preferably, the specific content of S2 includes:

[0019] Obtained through time-frequency transformation The corresponding time-frequency representation matrix The time-frequency transformation methods specifically include: Short Time Fourier Transform (STFT), Fast Fourier Transform (FFT), wavelet transform, or wavelet packet transform.

[0020] In the time-frequency representation matrix Based on this, the amplitude spectrum is calculated. :

[0021] .

[0022] Preferably, the specific content of time-frequency transformation using the Short Time Fourier Transform (STFT) method includes:

[0023] ;

[0024] ;

[0025] ;

[0026] In the formula, is the window length, representing the number of sample points per time window, is the frame shift, represents the time-domain samples of the input dynamic signal in the frame; represents the window function, represents the total number of frames, represents the floor function, is the frequency index of the time-frequency representation matrix, , , is the number of points of the time-frequency transform, is the imaginary unit; is the time frame index, ; is the number of overlapping sample points between adjacent frames.

[0027] Preferably, the amplitude spectrum along the time axis in S3 is analyzed to extract the short-time noise amplitude and the long-time noise amplitude The specific content includes: A double-time-scale noise estimation is used to calculate the short-time noise amplitude

[0028] and the long-time noise amplitude by analyzing the statistical distribution of the amplitude spectrum along the time axis:

[0029] ;

[0030] ;

[0031] In the formula, and represent the th quantile and the th quantile, respectively; and are the quantile thresholds for short-time and long-time noise estimation, respectively; reflects the local background level of the noise, captures the overall trend and possible sudden changes of the noise.

[0032] Preferably, dynamic weights are introduced in S3 to fuse and to generate the basic noise amplitude The specific content includes:

[0033] Dynamic weights are introduced to balance the contributions of the two estimates of short-time and long-time: ​

[0034] ;

[0035] ;

[0036] wherein, is the initial weight, is the time frame index, , is the frequency index of the time-frequency representation matrix, , , is the number of points of the time-frequency transform, is the imaginary unit, and is the upper and lower bounds of the range, satisfying 0≤ <1. <1.

[0037] The basic noise amplitude is:

[0038] .

[0039] Preferably, in S4, the signal-to-noise amplitude ratio is calculated based on , and the minimum amplitude threshold is calculated according to The specific content of includes:

[0040] The signal-to-noise amplitude ratio is calculated based on the basic noise amplitude , which is used to quantify the relative strength of the signal and the noise:

[0041] ;

[0042] The minimum amplitude threshold is calculated based on :

[0043] ;

[0044] wherein, is the scaling factor, which controls the range of the minimum amplitude threshold and ensures that the key features of the signal are not weakened.

[0045] Preferably, in S4, the estimated noise amplitude is calculated based on the dynamic noise scaling factor and The specific content of includes:

[0046] The dynamic noise scaling factor is introduced to dynamically adjust the noise amplitude:

[0047] ;

[0048] in, This is the scaling factor, which controls the sensitivity of the adjustment factor. These are the initial weights;

[0049] Based on dynamic noise scaling factor Calculate and estimate noise amplitude :

[0050] .

[0051] Preferably, the specific content of S5 includes:

[0052] in accordance with and Calculate the noise reduction spectrum amplitude :

[0053] ;

[0054] Preserve the original phase and recover the denoised complex spectrum. :

[0055] ;

[0056] The inverse transform is used to reconstruct the time-domain signal from the denoised spectrum. :

[0057] ;

[0058] In the formula, For overlapping windowing functions.

[0059] Preferably, it also includes: S6. Real-time monitoring of the noise reduction time-domain signal. Quality, generating performance evaluation reports.

[0060] An adaptive noise reduction filtering system, comprising:

[0061] The preprocessing module is used to acquire dynamic signals from the signal receiving device. The corresponding time-domain signal is obtained after preprocessing. ,in The signal length;

[0062] The amplitude spectrum generation module is used to obtain the amplitude spectrum through time-frequency transformation. The corresponding time-frequency representation matrix , used to indicate the first Frame, First The complex amplitude of the frequency point, and according to Calculate the amplitude spectrum ;

[0063] A double-time-scale noise estimation module is configured to analyze the amplitude spectrum along the time axis using a double-time-scale noise estimation method, extract a short-time noise amplitude , and extract a long-time noise amplitude ; Introduce a dynamic weight Fuse and to generate a basic noise amplitude ;

[0064] A denoising module is configured to calculate a signal noise amplitude ratio SNMR based on , calculate a minimum amplitude threshold based on , and calculate an estimated noise amplitude based on a dynamic noise scaling factor and ; ;

[0065] A noise reduction and signal reconstruction module is configured to calculate a noise reduction spectrum amplitude based on and , retain original phase information, and reconstruct a noise reduction time domain signal from through inverse transformation. .

[0066] A computer readable storage medium has a computer program stored thereon, and the computer program is executed by a processor to implement any of the adaptive noise reduction filtering methods described above.

[0067] An electronic device includes a processor and a memory configured to store one or more programs, and when the one or more programs are executed by the processor, any of the adaptive noise reduction filtering methods described above is implemented.

[0068] According to the technical solutions described above, compared with the prior art, the present disclosure provides an adaptive noise reduction filtering method, system, medium, and device, which has the following beneficial effects:

[0069] 1. Excellent noise suppression capability: by introducing double-time-scale noise estimation, combining dynamic fusion of a short-time noise amplitude and a long-time noise amplitude , the present disclosure accurately captures the dynamic characteristics of non-stationary noise, and further realizes efficient noise suppression through adaptive noise reduction adjustment of a signal noise amplitude ratio SNMR and a dynamic noise scaling factor .

[0070] 2. Key feature protection: the present disclosure also introduces a minimum amplitude threshold and a final noise amplitude​​ The signal noise amplitude ratio (SNMR) is dynamically controlled to adjust the noise reduction strength, effectively avoiding excessive signal suppression, and enhancing the relative strength of the characteristic frequency band.

[0071] 3. Wide adaptability: Various time-frequency transform techniques (including but not limited to STFT, FFT, wavelet transform, etc.) and flexible window function and frame shift selection (including but not limited to Hanning window, fixed frame shift or adaptive frame shift) are introduced in the application, combined with optimized multi-scale noise estimation and low-complexity SNMR adaptive noise reduction adjustment algorithm, which significantly improves the robustness to complex non-stationary noise environment. BRIEF DESCRIPTION OF DRAWINGS

[0072] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0073] Figure 1 A flowchart of an adaptive noise reduction filtering method provided by the present application is shown in the figure.

[0074] Figure 2 A time-frequency graph of an original urban environment sound signal provided by an embodiment of the present application is shown in the figure.

[0075] Figure 3 A spectrum graph of an urban environment sound signal after noise reduction by the method provided by an embodiment of the present application is shown in the figure.

[0076] Figure 4 A spectrum graph of an urban environment sound signal after noise reduction by spectral subtraction provided by an embodiment of the present application is shown in the figure.

[0077] Figure 5 A spectrum graph of an urban environment sound signal after noise reduction by Wiener filtering provided by an embodiment of the present application is shown in the figure.

[0078] Figure 6 A spectrum graph of an urban environment sound signal after noise reduction by wavelet provided by an embodiment of the present application is shown in the figure.

[0079] Figure 7 A spectrum graph of an urban environment sound signal after noise reduction by wavelet packet provided by an embodiment of the present application is shown in the figure.

[0080] Figure 8 A schematic diagram of the hardware system structure of a hydroelectric generating set provided by an embodiment of the present application is shown in the figure.

[0081] Figure 9 A time-frequency graph of an abnormal working condition sound of a hydroelectric generating set collected by an embodiment of the present application is shown in the figure.

[0082] Figure 10 The noise reduction results for abnormal operating condition sounds provided in the embodiments of the present invention;

[0083] Figure 11 The sub-band energy proportion characteristics provided in the embodiments of the present invention; wherein Figure 11 (a) ~ Figure 11 (h) are the curves showing the change of energy proportion of the first to eighth sub-bands over time, respectively;

[0084] Figure 12 A schematic diagram illustrating the verification results of the CNN model provided in this embodiment of the invention;

[0085] Figure 13 This is a schematic diagram of a wind turbine hardware system provided in an embodiment of the present invention;

[0086] Figure 14 This is a time-frequency diagram of abnormal sounds from wind turbine generators collected according to an embodiment of the present invention;

[0087] Figure 15 This is a schematic diagram illustrating the noise reduction result of the method provided in an embodiment of the present invention;

[0088] Figure 16 A schematic diagram of the extracted MFCC features provided in an embodiment of the present invention; Figure 16 (a)- Figure 16 (l) The curves showing the changes of 12 MFCC characteristics are shown in sequence;

[0089] Figure 17 A schematic diagram illustrating the verification results of the Time Delay Neural Network (TDNN) model provided in an embodiment of the present invention;

[0090] Figure 18 A schematic diagram of a bridge monitoring hardware system provided in an embodiment of the present invention;

[0091] Figure 19 The bridge vibration signal collected is provided in the embodiments of the present invention;

[0092] Figure 20 This is a schematic diagram illustrating the noise reduction result of the method provided in an embodiment of the present invention;

[0093] Figure 21 A schematic diagram of the extracted time-domain features provided in an embodiment of the present invention; Figure 21 (a)- Figure 21 (j) The curves showing the changes of 10 time-domain features are shown in turn;

[0094] Figure 22 This is a schematic diagram illustrating the verification results of the ResNet residual network model provided in an embodiment of the present invention. Detailed Implementation

[0095] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the protection scope of the present application.

[0096] The present application provides an adaptive noise reduction filtering method, as shown in the formula (1), comprising the following steps: Figure 1

[0097] S1. Obtaining a dynamic signal from a signal receiving device, and obtaining a corresponding time domain signal through preprocessing , wherein is a signal length;

[0098] S2. Obtaining a corresponding time-frequency representation matrix through time-frequency transformation, which is used to represent a complex amplitude of the i-th frame and the j-th frequency point, and calculating an amplitude spectrum according to

[0099] S3. Adopting a double-time-scale noise estimation method to analyze the amplitude spectrum along a time axis, extracting a short-time noise amplitude and a long-time noise amplitude , introducing a dynamic weight , and fusing and to generate a basic noise amplitude ;

[0100] S4. Calculating a signal-to-noise amplitude ratio based on , calculating a minimum amplitude threshold according to the signal-to-noise amplitude ratio , and calculating an estimated noise amplitude based on a dynamic noise scaling factor and ;

[0101] S5. Calculating a noise reduction spectral amplitude according to and , reserving original phase information, and reconstructing into a noise reduction time domain signal through inverse transformation.

[0102] It should be noted that:​​​​​​​

[0103] The initial stage pre-processes the signal and removes the DC offset to effectively eliminate the DC component interference to obtain a pre-processed time domain signal The DC offset removal can be implemented by various technical means, including but not limited to traditional mean subtraction, high-pass filtering to suppress low-frequency DC components, and adaptive filtering technology.

[0104] In order to further implement the above technical solutions, the specific content of S1 pre-processing includes:

[0105] Obtain a dynamic signal from a signal receiving device , remove low-frequency interference by DC offset preprocessing to obtain a pre-processed time domain signal .

[0106] In order to further implement the above technical solutions, the specific content of S2 includes:

[0107] Obtain the corresponding time-frequency representation matrix by time-frequency transformation , wherein the time-frequency transformation method specifically includes: short-time Fourier transform STFT, fast Fourier transform FFT, wavelet transform or wavelet packet transform;

[0108] On the basis of the time-frequency representation matrix , the amplitude spectrum is calculated:

[0109] .

[0110] In order to further implement the above technical solutions, the specific content of time-frequency transformation by short-time Fourier transform STFT method includes:

[0111] ;

[0112] ;

[0113] ;

[0114] In the formula, is the window length, indicating the number of sampling points of each time window, is the frame shift, represents the time domain sample of the input dynamic signal in the frame; represents the window function, represents the total number of frames, represents the floor function, is the frequency index of the time-frequency representation matrix, ,​ , The number of points in the time-frequency transformation. The imaginary unit; For time frame indexing, ; This represents the number of overlapping sampling points between adjacent frames, used to ensure smooth signal reconstruction and avoid distortion caused by windowing.

[0115] It should be noted that:

[0116] The power spectrum can be implemented using time-domain to time-frequency domain transformation techniques, including but not limited to Short-Time Fourier Transform (STFT), Fast Fourier Transform (FFT), wavelet transform, and wavelet packet transform. This embodiment uses STFT as an example.

[0117] To further implement the above technical solution, amplitude spectrum analysis is performed along the time axis in S3. Extracting short-time noise amplitude and long-term noise amplitude The specific content includes:

[0118] A dual-timescale noise estimation method is employed, and the amplitude spectrum is analyzed. Calculate the short-time noise amplitude based on the statistical distribution along the time axis. and long-term noise amplitude :

[0119] ;

[0120] ;

[0121] In the formula, and They represent the first quantiles and the quantiles; and These are the quantile thresholds for short-time and long-time noise estimation, respectively, and their ranges can be dynamically adjusted to adapt to different noise environments. Reflects the local background level of noise. Capture the overall trend and possible sudden changes in noise.

[0122] To further implement the above technical solution, dynamic weights are introduced in S3. Will and Perform fusion to generate basic noise amplitude The specific content includes: introducing dynamic weights To balance the contributions of short-time and long-time estimates:

[0123] ;

[0124] ;

[0125] wherein, is the initial weight, is the time frame index, , is the frequency index of the time-frequency representation matrix, , , is the number of points of the time-frequency transform, is the imaginary unit, and is the upper and lower bounds of the range, satisfying 0≤ <0 ≤1;

[0126] The basic noise amplitude is:

[0127] .

[0128] It should be noted that:

[0129] The noise of the dynamic signal has the characteristics of short-time burstiness and long-time stationarity, and single-scale noise estimation is difficult to accurately capture its dynamic distribution. To solve this problem, the present application uses double-time scale noise estimation. and are short-time and long-time noise estimation thresholds, respectively, which can be calculated based on statistical methods, or other methods, including but not limited to mean plus multiple of standard deviation or K-means clustering method, to optimize the noise estimation performance.

[0130] reflects the local background level of the noise, captures the overall trend and possible burstiness of the noise. To fuse the short-time and long-time noise characteristics, a dynamic weight is introduced to balance the contributions of the two estimates. The weight reflects the average intensity of the signal relative to the long-time noise, and is used to adjust the contribution ratio of the short-time and long-time estimates.

[0131] To further implement the above technical solutions, the signal noise amplitude ratio is calculated based on in S4, and the minimum amplitude threshold is calculated according to the signal noise amplitude ratio The specific content includes:

[0132] The signal noise amplitude ratio is calculated based on the basic noise amplitude , for quantifying the relative strength of signal and noise:

[0133] ;

[0134] Based on the signal-to-noise amplitude ratio Calculate the minimum amplitude threshold :

[0135] ;

[0136] In the formula, is a scaling factor that controls the range of the minimum amplitude threshold, ensuring that the key features of the signal are not weakened.

[0137] To further implement the above technical solution, the dynamic noise scaling factor and Calculate the estimated noise amplitude The specific content includes:

[0138] Introduce a dynamic noise scaling factor to dynamically adjust the noise amplitude:

[0139] ;

[0140] Where, is a scaling factor that controls the sensitivity of the adjustment factor, is the initial weight, calculated according to the formula above ;

[0141] Based on the dynamic noise scaling factor Calculate the estimated noise amplitude :

[0142] .

[0143] It should be noted that:

[0144] In order to effectively denoise, the present application calculates the signal-to-noise amplitude ratio SNMR based on the basic noise amplitude, which is used to quantify the relative strength of signal and noise. To avoid excessive suppression of the signal during denoising, the minimum amplitude threshold is further calculated based on SNMR.

[0145] To further implement the above technical solution, the specific content of S5 includes:

[0146] According to and Calculate the denoised spectrum amplitude :

[0147] ;

[0148] Preserve the original phase and recover the denoised complex spectrum. :

[0149] ;

[0150] The inverse transform is used to reconstruct the time-domain signal from the denoised spectrum. :

[0151] ;

[0152] In the formula, For overlapping windowing functions.

[0153] To further implement the above technical solution, it also includes: S6. Real-time monitoring of the noise reduction time domain signal. Quality, generating performance evaluation reports.

[0154] An adaptive noise reduction filtering system, comprising:

[0155] The preprocessing module is used to acquire dynamic signals from the signal receiving device. The corresponding time-domain signal is obtained after preprocessing. ,in The signal length;

[0156] The amplitude spectrum generation module is used to obtain the amplitude spectrum through time-frequency transformation. The corresponding time-frequency representation matrix , used to indicate the first Frame, First The complex amplitude of the frequency point, and according to Calculate the amplitude spectrum ;

[0157] The dual-timescale noise estimation module is used to analyze the amplitude spectrum along the time axis using a dual-timescale noise estimation method. Extracting short-time noise amplitude and long-term noise amplitude Introducing dynamic weights Will and Perform fusion to generate basic noise amplitude ;

[0158] Denoising module, used for based Calculate the signal-to-noise ratio According to the signal-to-noise amplitude ratio Calculate the minimum amplitude threshold And based on dynamic noise scaling factor as well as Calculate and estimate noise amplitude ;

[0159] a noise reduction and signal reconstruction module configured to reconstruct a noise-reduced time-domain signal from the noise-reduced frequency-domain signal by inverse transform and calculating noise-reduced frequency-domain amplitudes , preserving original phase information, by inverse transform reconstruction into a noise-reduced time-domain signal .

[0160] A computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements any of the above adaptive noise reduction filtering methods.

[0161] An electronic device comprising: a processor and a memory storing one or more programs; when the one or more programs are executed by the processor, any of the above adaptive noise reduction filtering methods is implemented.

[0162] The application will be further described below through specific experiments:

[0163] This experiment is based on the actual collected urban environmental sound signal data (this embodiment only takes environmental sound signal as an example, in the actual application process, it can be used for adaptive filtering of dynamic modulation signals, such as original acoustic and vibration signals when hydroelectric generating set is running, original sound signals when wind turbine generator is running, vibration displacement original signals when bridge is running, etc.), the sampling rate is 32 kHz, and the signal time domain range is 0 to 1.8 seconds. In this experiment, the main parameter configuration is: the window length is set to 256 sampling points, the overlap length is set to 128 sampling points, the FFT point number is set to 512, the short-time noise amplitude and the long-time noise amplitude are respectively taken as the 25th percentile and the 75th percentile, 0.5 is set as the lower limit of weight calculation, 0.9 is set as the upper limit of weight calculation, 0.005 is set as the base value of the minimum amplitude threshold, 1.5 is set as the dynamic noise scaling factor parameter.

[0164] As shown in Figures 2-7 , the first 1.1 seconds of the signal is a characteristic frequency band, and 1.1 seconds to 1.8 seconds is a noise frequency band, wherein Figure 2 represents the original frequency spectrum of the signal, Figure 3 is the signal frequency spectrum after noise reduction by applying the method, Figure 4 is the signal frequency spectrum after noise reduction by applying the spectral subtraction method, and the spectral subtraction method uses a subtraction factor =1.0, a floor factor =0.1, and noise estimation is based on the 20th percentile of the amplitude spectrum, so as to moderately suppress noise and reduce artifacts;Figure 5 The signal spectrum graph after applying the Wiener filter for noise reduction, the filter kernel size of the Wiener filter is 5, and the balance between noise suppression and signal fidelity is achieved; Figure 6 The signal spectrum graph after applying the wavelet for noise reduction, the wavelet uses db4 wavelet, the decomposition layer is 4, the soft threshold method is used, and the threshold is calculated based on the noise variance to achieve adaptive noise reduction and preserve signal details, Figure 7 The signal spectrum graph after applying the wavelet packet for noise reduction, the wavelet packet uses db4 wavelet, the decomposition layer is 4, the soft threshold method is used to reduce artifacts, and the effectiveness of the noise reduction method is evaluated.

[0165] The sampling rate of the dynamic signal is 32000Hz, and the total duration is about 1.8 seconds (57498 sampling points in total). After normalization and removal of DC offset, the time-frequency representation matrix is generated, and the amplitude spectrum is further calculated, resulting in an amplitude spectrum size of (257, 450), corresponding to 257 frequency components and 450 time frames, reflecting the time-frequency characteristics of about 1.8 seconds of signal, wherein the amplitude spectrum statistical values are shown in Table 1.

[0166]

[0167] By analyzing the 1.8 second amplitude spectrum, a multi-scale method is used to calculate the short-time and long-time noise amplitudes, and based on the 25% and 75% quantiles, short-time noise and long-time noise matrices are generated, with a matrix size of (257, 1). Then, based on the formula , the calculation is 0.1480, and the output is shown in Tables 2 and 3.

[0168]

[0169]

[0170] Using the basic noise matrix and the amplitude spectrum , the SNMR is calculated, the minimum amplitude threshold is calculated, the dynamic noise scaling factor is calculated, and the final noise amplitude is calculated, and the output is shown in Tables 4 and 5.

[0171]

[0172]

[0173] Using the final noise amplitude, the final amplitude spectrum matrix is calculated, the complex spectrum matrix is reconstructed, and the noise-reduced signal is reconstructed, and the output is shown in Table 6.

[0174]

[0175] Based on spectral statistics and generalized index values, a comprehensive comparative analysis is conducted on six methods (no denoising, our method, spectral subtraction, Wiener filtering, wavelet denoising, and wavelet packet denoising).

[0176] 1. Comparison of statistical indicators of the time-frequency matrix after noise reduction

[0177] like Figures 2-7 As shown, these spectrograms uniformly demonstrate the signal's distribution characteristics in the time domain: the first 1.1 seconds correspond to the characteristic frequency band, which mainly contains the periodic or modulated components of the target signal, typically exhibiting high energy concentration, spectral clarity, and structured characteristics (mainly in the high-amplitude green to yellow region); from 1.1 seconds to 1.8 seconds is the noise frequency band, presenting as randomly distributed broadband noise, with relatively dispersed energy and lacking obvious patterns (mainly in the low-amplitude blue to purple region).

[0178] Specifically, Figure 2 The original spectrum of the signal directly reflects the time-frequency characteristics of the unprocessed signal and serves as a benchmark for subsequent method evaluation. The characteristic frequency bands have prominent energy, while the noise frequency bands, although lower, still have significant interference. Figure 3 The signal spectrum after applying this method shows that the noise band amplitude is significantly reduced (the color tends to be dark purple, close to -100 dB), while the characteristic band maintains high fidelity and energy concentration, and the spectral lines are sharper. Figures 4-7 The graphs show the signal spectra after applying different contrast methods (such as spectral subtraction, Wiener filtering, wavelet denoising, and wavelet packet denoising). As can be seen from the graphs, although these contrast methods can partially suppress noise (the noise area becomes darker), there is still a lot of residual interference, the fidelity of the characteristic frequency band is low, and the overall amplitude reduction is not as thorough as that of this method. In contrast, this method shows stronger noise suppression capability and better feature protection and enhancement, achieving a more uniform low amplitude distribution in the noise frequency band, while enhancing the relative intensity and clarity of the characteristic frequency band.

[0179] Table 7 provides a visual comparison of the final output spectrum statistics, the spectrum statistics for the first 1.1 seconds, and the spectrum statistics for the last 1.1 seconds of the six methods, with a focus on the mean and the maximum value.

[0180]

[0181] 1) Compared with no noise reduction, this method

[0182] Outstanding noise suppression: The proposed method significantly outperforms the no-denoising in noise suppression. The final output spectrum statistics mean drops from -20.0055 dB of no-denoising to -75.4310 dB, a decrease of about 55 dB, indicating strong noise suppression. The mean of the first 1.1 seconds drops from -9.4297 dB to -59.8047 dB (a decrease of about 50 dB), and the mean of the rest of the 1.1 seconds drops from -36.6246 dB to -99.9867 dB (a decrease of about 63 dB). This significant noise reduction is due to the double-time-scale noise estimation and the adaptive adjustment of the SNMR and dynamic noise scaling factor in S3, which accurately captures and eliminates non-stationary noise.

[0183] Relative intensity enhancement of key features: Despite the strong noise suppression, the maximum of the first 1.1 seconds of the proposed method drops slightly from 19.4851 dB of no-denoising to 18.9729 dB (only a decrease of about 0.5 dB), preserving the peak intensity close to the original signal. The maximum of the rest of the 1.1 seconds drops from -5.4981 dB to -39.5729 dB, but this is consistent with the target frequency band focusing on the first 1.1 seconds, and the minimum amplitude threshold and dynamic adjustment in S3 ensure the relative intensity of key features is enhanced.

[0184] 2) Comparison of the proposed method with other denoising methods

[0185] Stronger noise suppression: Compared with other denoising methods (spectral subtraction -22.2795 dB, Wiener filtering -24.4751 dB, wavelet denoising -27.6328 dB, and wavelet packet denoising -43.2331 dB), the final output spectrum statistics mean of the proposed method -75.4310 dB is the lowest, with a suppression intensity of about 32 dB -53 dB higher. The mean of the first 1.1 seconds -59.8047 dB is about 47 dB -50 dB lower than that of other methods (-9.6612 dB to -12.4225 dB), and the mean of the rest of the 1.1 seconds -99.9867 dB is about 46 dB -58 dB lower than that of other methods (-42.1082 dB to -53.8296 dB), showing that the proposed method has more complete suppression of noise in the entire 1.8 second signal.

[0186] Better feature protection and enhancement: The maximum value of the first 1.1 seconds of the method is 18.9729 dB, close to spectral subtraction (19.4788 dB), Wiener filtering (19.5175 dB), wavelet denoising (19.5015 dB) and wavelet packet denoising (19.5027 dB), but the mean value -59.8047 dB is much lower than these methods (-9.6612 dB to -12.4225 dB), indicating that the relative peak intensity is still high under strong denoising. Combined with the feature concentration of the first 1.1 seconds, the dynamic adjustment of S3 effectively enhances the relative strength of the feature band.

[0187] 2. Comparative analysis combined with generalized indicators

[0188] Table 8 provides indicator data (SNR after denoising, spectral contrast, HNR, SDR, FSIM, GMSD) for comparative analysis of six methods. Among them, SNR, Signal-to-Noise Ratio: The higher the value, the better the signal quality; Spectral Contrast: The average value based on the difference between the peak and valley of the frequency band, the higher the value, the more prominent the separation degree of harmonics and noise; HNR, Harmonics-to-Noise Ratio: The higher the value, the more dominant the harmonic component; SDR, Source-to-Distortion Ratio: The higher the value, the smaller the distortion; FSIM, Feature Similarity Index: The range is [0, 1], the closer to 1, the more similar to the original signal characteristics; GMSD, Gradient Magnitude Similarity Deviation: The smaller the value, the better the image matrix quality.

[0189]

[0190] 1) The method compared with no denoising

[0191] Excellent noise suppression capability: The method improves the SNR from 10.61 without denoising to 11.53 (an increase of about 0.92 dB) through the double-time-scale noise estimation of Step2 and the SNMR dynamic adjustment of S3, indicating that the signal-to-noise ratio is improved (the noise is reduced by MCRA2 estimation). Spectral Contrast is significantly increased from 12.96 to 36.56 (an increase of about 23.6), reflecting the enhancement of spectral feature separation (peak-valley difference increase). SDR is increased from 0 to 9.83, showing that the noise power is significantly reduced, verifying the superiority of the method in noise suppression.

[0192] The relative intensity of the key features is enhanced: after noise reduction, the spectral contrast is increased to 36.56 (vs 12.96), indicating that the target frequency band features are prominent (segment contrast optimization). The GMSD is increased from 1 to 1.05, close to 1, indicating stable quality. The SDR is 9.83, indicating that the signal reconstruction quality is better than that without noise reduction, and the relative intensity of the features is enhanced.

[0193] 2) The method is compared with other noise reduction methods

[0194] Stronger noise suppression capability: the SNR (11.53) of the method is higher than that of spectral subtraction (10.59), wavelet denoising (10.73), and wavelet packet denoising (10.72), and is only lower than that of Wiener filtering (11.16), but the spectral contrast (36.56) is much higher than that of other methods (8.73 to 19.54), showing stronger spectral feature separation (peak-valley difference maximum). The SDR (9.83) is lower than that of spectral subtraction (21.4), Wiener filtering (23.24), wavelet denoising (35.36), and wavelet packet denoising (34.19), but reflects reasonable distortion control, proving the unique advantage of the method in noise suppression.

[0195] Better feature protection and enhancement: the spectral contrast (36.56) of the method is much higher than that of other methods, indicating that the target frequency band is significantly enhanced (segment contrast optimization). The FSIM (0.17) is lower than that of other methods (0.35 to 0.56), but the GMSD (1.05) is close to that of other methods (1.01-1.03), showing controllable feature distortion. Although the SDR (9.83) is lower than that of other methods, combined with the peak value retention (18.9729dB) in Table 7, it proves that the relative intensity of the features is enhanced under strong noise reduction, which is better than the balance of other methods.

[0196] The application will be further described below through specific application examples.

[0197] Example 1

[0198] As shown in Figure 8 , the multi-channel distributed high-resolution sound and vibration signal acquisition scheme of the hydroelectric generator mainly includes audio sensors, vibration sensors, signal acquisition cards, and host computers. Analog signals are collected through sensors, then output digital signals through conditioning circuits and acquisition cards, and transmitted to servers for processing and analysis using Ethernet.

[0199] In this embodiment, the main parameter configuration is: the window length is set to 256 sampling points, the overlap length is set to 128 sampling points, the FFT point number is set to 512, and the short-time noise amplitude and long-time noise amplitudes The 25th and 75th percentiles are taken as 0.5 and 0.9, respectively, 0.5 is set as the lower limit of the weight calculation, 0.9 is set as the upper limit of the weight calculation, 0.005 is set as the base value of the minimum amplitude threshold, 1.5 is set as the dynamic noise scaling factor parameter.

[0200] Figure 9 The frequency spectrum analysis of the collected abnormal working condition sound of the hydroelectric unit is shown. The horizontal axis represents time (unit: seconds), ranging from 0 to 10 seconds; the vertical axis represents frequency (unit: Hz), ranging from 0 to 16000 Hz. The amplitude (unit: dB) gradually changes from a higher amplitude (about -10 dB) to a lower amplitude (about -40 dB) to show the intensity distribution of the sound signal. This abnormal sound is caused by the entry of foreign matter into the interior of the hydraulic turbine, resulting in abnormal frequency bands and sound characteristics, among which the abnormal frequency bands are marked in the figure, in sharp contrast to the background noise.

[0201] The sampling rate of the acoustic and vibration signals is 32000 Hz, and the total duration is about 10 seconds (320000 sampling points in total). After normalization and removal of the direct current offset, a time-frequency representation matrix is generated, and the amplitude spectrum is further calculated, resulting in an amplitude spectrum size of (257, 2501), corresponding to 257 frequency components and 2501 time frames, reflecting the time-frequency characteristics of about 10 seconds of signal, wherein the amplitude spectrum statistical values are shown in Table 9.

[0202]

[0203] By analyzing the 10-second amplitude spectrum, the short-time and long-time noise amplitudes are calculated using a multi-scale method, and the short-time noise and long-time noise matrices are generated based on the 25th and 75th percentiles, respectively, with a matrix size of (257, 1). Then, based on the formula , the is calculated as 0.1374, and the output is shown in Tables 10 and 11.

[0204]

[0205]

[0206] The signal noise amplitude ratio SNMR is calculated using the basic noise matrix and the amplitude spectrum , the minimum amplitude threshold is calculated, the dynamic noise scaling factor is calculated, and the final noise amplitude is calculated, and the output is shown in Tables 12 and 13.

[0207]

[0208]

[0209] Using the final noise amplitude, the final amplitude spectrum matrix is ​​calculated, the complex spectrum matrix is ​​reconstructed, and the denoised signal is reconstructed. The output is shown in Table 14.

[0210]

[0211] Figure 10 The audio spectrum analysis after applying this noise reduction method is shown. The horizontal axis represents time (in seconds), ranging from 0 to 10 seconds; the vertical axis represents frequency (in Hz), ranging from 0 to 16000 Hz. Amplitude range and Figure 9 Consistent. Through with Figure 9 As can be seen from the comparison, the method of the present invention significantly suppresses the background noise frequency band (manifested as an increase in large areas of low amplitude in the figure), while effectively preserving the abnormal characteristic frequency band caused by foreign objects inside the turbine (manifested as a more prominent high amplitude area). This indicates that the noise reduction method effectively maintains the relative intensity of key abnormal features while removing interference noise.

[0212] like Figure 11 As shown, by performing a three-level decomposition of the reconstructed signal using the db3 wavelet, the energy proportion characteristics of each sub-band were extracted. Figure 11 (a) ~ Figure 11 (h) are the energy percentage curves of the first to eighth sub-bands over time, respectively. These time-series plots show the energy percentage curves of the eight sub-bands over time. Among them, the high-frequency sub-bands show obvious peak fluctuations, reflecting the periodic characteristics of abnormal operating condition sound signals, such as transient energy concentration caused by foreign object impact.

[0213] To detect abnormal operating conditions in the acoustic and vibration signals of hydropower units, a two-dimensional convolutional neural network (CNN) model is constructed. The energy features extracted by wavelet packet decomposition are used as input for abnormal operating condition identification. The model architecture is as follows:

[0214] Input layer: Receives wavelet packet energy features, and the data is formatted as a two-dimensional tensor (with 1 channel).

[0215] The feature extraction module includes: a first convolutional layer with 16 kernels, stride 1, padding 1, batch normalization, and ReLU activation; a first max pooling layer with a window and stride 2; a second convolutional layer with 32 kernels, stride 1, padding 1, batch normalization, and ReLU activation; a second max pooling layer with a window and stride 2; a third convolutional layer with 64 kernels, stride 1, padding 1, batch normalization, and ReLU activation; and an adaptive average pooling layer with a fixed output size of 2.

[0216] The classifier module includes: fully connected layer 1: input 64x2x2, output 64, ReLU activation, Dropout (probability 0.5); fully connected layer 2: output 2 neurons (normal / abnormal), no explicit Softmax (handled by loss function);

[0217] Loss function: cross-entropy loss, combined with the Softmax function to calculate classification probabilities. The model is implemented under the PyTorch framework and runs on CPU, with a compact and efficient structure suitable for real-time condition monitoring.

[0218] In this embodiment, to verify the effectiveness of the CNN model in water turbine unit condition anomaly detection, 1730 data samples are used. Among them, 1300 are normal samples and 430 are abnormal sound samples. These samples are based on the extracted wavelet packet energy features as input. The dataset is divided into training set (1384 samples), validation set (173 samples) and test set (173 samples) in the ratio of 8:1:1 to ensure the model's generalization ability. The optimizer is Adam with an initial learning rate of 0.001.

[0219] The model test set accuracy reaches 98.84%, showing high efficiency in distinguishing normal and abnormal conditions. Lower false positives and false negatives further verify the robustness of the model. Batch normalization and Dropout effectively prevent overfitting, and learning rate scheduling optimizes the convergence speed.

[0220] This model realizes accurate anomaly detection of hydroelectric generator vibration signals through a two-dimensional CNN architecture combined with wavelet packet energy features. Its high accuracy (98.84%) and stable classification performance prove its reliability in complex condition monitoring, providing an efficient solution for intelligent diagnosis of hydroelectric generators.

[0221] Embodiment Two:

[0222] As shown in Figure 13 , the multi-channel distributed high-resolution acoustic signal acquisition and analysis scheme of the wind turbine mainly includes audio sensors, signal acquisition cards, and host computers. Analog signals are collected by sensors, then output digital signals through conditioning circuits and acquisition cards, and transmitted to the host computer in the booster station for processing and analysis. In this embodiment, the main parameter configurations are: window length set to 256 sampling points, overlap length set to 128 sampling points, FFT point number set to 512, short-time noise amplitude and long-time noise amplitude number are taken as the 25th and 75th percentiles, set to 0.5 as the lower limit of weight calculation, set to 0.9 as the upper limit of weight calculation, a base value of 0.005 is set as the minimum amplitude threshold, a value of 1.5 is set as the dynamic noise scaling factor parameter.

[0223] Figure 14 The frequency spectrum analysis of the collected abnormal sound of the wind turbine is shown. The horizontal axis represents time (unit: seconds), ranging from 0 to 10 seconds; the vertical axis represents frequency (unit: Hz), ranging from 0 to 16000 Hz. The amplitude (unit: dB) gradually changes from a higher amplitude (about -10 dB) to a lower amplitude (about -40 dB) to show the intensity distribution of the sound signal. This abnormal sound is caused by abnormality of the turbine blade bearing, resulting in abnormal frequency bands and sound characteristics, where the abnormal frequency bands are marked in the figure, which are in sharp contrast to the background noise.

[0224] The sampling rate of the dynamic signal is 32000 Hz, and the total duration is about 10 seconds (320000 sampling points in total). After normalization and removal of the direct current offset, the time-frequency representation matrix is generated, and the amplitude spectrum is further calculated, and the amplitude spectrum size is (257, 2501), corresponding to 257 frequency components and 2501 time frames, reflecting the time-frequency characteristics of about 10 seconds of signal, where the amplitude spectrum statistics are shown in Table 15.

[0225]

[0226] By analyzing the 10-second amplitude spectrum, the short-time and long-time noise amplitudes are calculated using a multi-scale method, and the short-time noise and long-time noise matrices are generated based on the 25% and 75% quantiles, respectively, with a matrix size of (257, 1). Then, based on the formula , the is calculated as 0.1620, and the output is shown in Tables 16 and 17.

[0227]

[0228]

[0229] The signal noise amplitude ratio SNMR is calculated using the base noise matrix and the amplitude spectrum , the minimum amplitude threshold is calculated, the dynamic noise scaling factor is calculated, and the final noise amplitude is calculated, and the output is shown in Tables 18 and 19.

[0230]

[0231]

[0232] Using the final noise amplitude, the final amplitude spectrum matrix is calculated, the complex spectrum matrix is reconstructed, and the denoised signal is reconstructed, and the output is shown in Table 20.

[0233]

[0234] Figure 15 The audio spectrum analysis after applying this noise reduction method is shown. The horizontal axis represents time (in seconds), ranging from 0 to 10 seconds; the vertical axis represents frequency (in Hz), ranging from 0 to 16000 Hz. The amplitude (in dB) range is... Figure 14 Consistent. Through with Figure 14 As can be seen from the comparison, the method of the present invention significantly suppresses the background noise frequency band (manifested as an increase in large areas of low amplitude in the figure), while effectively preserving the abnormal characteristic frequency band caused by the blade bearing (manifested as a more prominent high amplitude area). This indicates that the noise reduction method effectively maintains the relative intensity of key abnormal features while removing interference noise.

[0235] like Figure 16 As shown, MFCC features are extracted from the denoised signal. These time-series plots illustrate the changes in 12 MFCC features.

[0236] To achieve abnormal state detection of acoustic signals from wind turbine generators, this method constructs a Time Delay Neural Network (TDNN) model, using MFCC features as input, which is suitable for pattern recognition of non-stationary time-series signals. This model introduces a time delay mechanism to efficiently capture dynamic context information, and its architecture is as follows:

[0237] Input layer: Receives MFCC feature sequences and transposes them to adapt to one-dimensional convolution operations.

[0238] The temporal feature extraction module includes: TDNN layer 1: a one-dimensional convolution (kernel size 5, padding 2), with input channels equal to the feature dimension and output channels of 128; followed by batch normalization, ReLU activation, and Dropout (probability 0.5). TDNN layer 2: a one-dimensional convolution (kernel size 5, padding 2), with input channels of 128 and output channels of 128; followed by batch normalization and ReLU activation. Adaptive pooling layer: compresses the temporal dimension to a fixed length of 1.

[0239] The classifier module includes: a fully connected layer with 128-dimensional input and two-class output (normal / abnormal), where the classification probability is implicitly handled by the cross-entropy loss function.

[0240] The time delay neural network (TDNN) model is implemented under the Python framework and runs on the CPU, making it suitable for real-time anomaly monitoring. In this embodiment, to verify the effectiveness of the TDNN model, 2450 wind turbine acoustic data samples (1800 normal samples and 650 abnormal samples) are used as input based on MFCC features. The dataset is divided into a training set (1960 samples), a validation set (245 samples), and a test set (245 samples) in a ratio of 8:1:1 to ensure the model's generalization ability. The optimizer uses Adam with an initial learning rate of 0.001, and introduces weight decay and learning rate step scheduling (decreased to 0.9 times every 5 cycles).

[0241] As shown in Figure 17 , the model achieves an accuracy of 97.96% on the test set, demonstrating high efficiency in distinguishing between normal and abnormal states. The low false positive and false negative rates further verify the model's robustness. Batch normalization and Dropout effectively prevent overfitting, and learning rate scheduling optimizes the convergence speed.

[0242] This model, through a one-dimensional TDNN architecture combined with MFCC features, achieves precise anomaly detection of wind turbine acoustic signals. Its high accuracy (97.96%) and stable classification performance prove its reliability in complex working condition monitoring, providing an efficient solution for intelligent operation and maintenance of wind turbines.

[0243] Embodiment Three:

[0244] The multi-channel distributed vibration displacement signal acquisition scheme for bridge monitoring mainly includes displacement sensors, vibration sensors, data acquisition and processing terminals, etc. The analog signals are collected by the sensors, then output digital signals through the conditioning circuit and acquisition card, transmitted to the data acquisition and processing terminal for processing and analysis, and then the wireless communication is used to alarm the abnormality.

[0245] In this embodiment, the main parameter configurations are: the window length is set to 256 sampling points, the overlap length is set to 128 sampling points, the FFT point number is set to 512, the short-time noise amplitude and the long-time noise amplitude are taken as the 25th and 75th percentiles, 0.5 is set as the lower limit of weight calculation, 0.9 is set as the upper limit of weight calculation, 0.005 is set as the base value of the minimum amplitude threshold, 1.5 is set as the dynamic noise scaling factor parameter.

[0246] Figure 19The bridge vibration signal is shown, with time axis from 0 to 5 seconds, and amplitude range from about -4 to 4. The signal is recorded on a real bridge structure, and represents the vibration response induced by vehicle passing or wind load. The signal is dominated by high frequency random oscillation, forming dense band-like fluctuation. The noise component occupies the dominant position. This noise causes the signal details to be blurred, and the inherent dynamic characteristics of the bridge are difficult to identify. Observing the details, the signal is relatively stable in the first 2 seconds, with small amplitude fluctuations. At about 2 seconds, the amplitude increases, and peak and valley alternation appears, indicating the starting time of external impact or load, such as vehicle passing or wind load induced. Corresponding to the free vibration response of the bridge, the potential damping oscillation mode is hidden in it.

[0247] The sampling rate of the dynamic signal is 32000 Hz, and the total length is about 5 seconds (total 160000 sampling points). After normalization and removal of DC offset, the time-frequency representation matrix is generated, and the amplitude spectrum is further calculated, and the amplitude spectrum size is (257, 1251), corresponding to 257 frequency components and 1251 time frames, reflecting the time-frequency characteristics of about 5 seconds signal, and the amplitude spectrum statistical values are shown in Table 21.

[0248]

[0249] By analyzing the 1.8 second amplitude spectrum, the short-time and long-time noise amplitudes are calculated using a multi-scale method, and the short-time noise and long-time noise matrices are generated based on the 25% and 75% quantiles, respectively, with a matrix size of (257, 1). Then, based on the formula , the is calculated as 0.1504, and the output is shown in Tables 22 and 23.

[0250]

[0251]

[0252] The signal noise amplitude ratio SNMR is calculated using the basic noise matrix and the amplitude spectrum , the minimum amplitude threshold is calculated, the dynamic noise scaling factor is calculated, and the final noise amplitude is calculated, and the output is shown in Tables 24 and 25.

[0253]

[0254]

[0255] Using the final noise amplitude, the final amplitude spectrum matrix is calculated, the complex spectrum matrix is reconstructed, and the denoised signal is reconstructed, and the output is shown in Table 26.

[0256]

[0257] Figure 20 The denoised bridge vibration signal is shown, with a time axis from 0 to 5 seconds and an amplitude range of about -1 to 1, significantly better than the -4 to 4 range of the pre-denoising signal, indicating that the denoising effectively reduces the noise amplitude. The signal still shows high-frequency random oscillation, but the band fluctuation is less sparse than before denoising, the noise influence is significantly weakened, and a relatively clear periodic pattern is revealed, reflecting the inherent dynamic characteristics of the bridge. Observing the details, the signal is relatively stable in the first 2 seconds, with small amplitude fluctuations and a low noise baseline, reflecting the suppression of environmental interference after denoising. At about 2 seconds, the amplitude increases significantly, with regular peak and valley alternation, indicating the start of external impact (such as vehicle load), which is consistent with the pre-denoising, but the waveform is smoother and the damped oscillation pattern is more easily identifiable. This shows that the denoising method not only removes the interference noise, but also better maintains the original key feature information.

[0258] As shown in Figure 21 , 10 kinds of time domain features are extracted from the denoised bridge vibration signal, and these time series diagrams show the change trend of mean, root mean square, standard deviation, skewness, kurtosis, peak value, wave crest factor, shape factor, pulse factor and gap factor. Extract a set of features every 0.1 seconds and draw it as a 2x5 subgraph to intuitively present the change of features over time. In the feature curve, the mean and root mean square values show obvious peak fluctuations, reflecting the periodic response caused by impact load (such as vehicle passing) in the bridge vibration signal.

[0259] To verify the effectiveness of the ResNet model, a total of 2160 bridge vibration displacement data samples (1600 normal samples and 560 abnormal samples) are used as input (feature dimension 10) based on 10 kinds of time domain features. The dataset is divided into training set (1728 samples), validation set (216 samples) and test set (216 samples) in the ratio of 8:1:1 to ensure the model's generalization ability. The model is implemented under the PyTorch framework and runs on CPU.

[0260] The ResNet model uses a one-dimensional convolutional architecture to process the time domain feature sequence of the bridge vibration displacement signal. The input layer receives a 10-dimensional feature vector and expands it to 64 channels through an initial convolutional layer. The model contains multiple residual blocks (layer1 and layer2), each composed of two convolutional layers, batch normalization and ReLU activation. The residual connection ensures smooth information transmission and avoids deep network degradation. The adaptive average pooling layer compresses the feature dimension, and the fully connected layer outputs a 2-class probability (normal / abnormal). The Dropout layer (probability 0.5) prevents overfitting. This design fully utilizes residual learning to improve the generalization ability of non-stationary signals and supports accurate identification of bridge structure anomalies.

[0261] As shown in Figure 21As shown, the accuracy of the model on the test set reached 96.76%, showing high efficiency in distinguishing normal and abnormal structural states. The low false positive and false negative further verified the robustness of the model. The residual connection and batch normalization effectively prevent overfitting, and the learning rate scheduling effectively optimizes the convergence speed. The model realizes precise abnormal monitoring of bridge building vibration displacement signals through the ResNet architecture combined with time domain features. Its high accuracy (96.76%) and stable classification performance prove its reliability in structural health monitoring, providing an efficient solution for intelligent state evaluation of bridge buildings.

[0262] The above examples are only used to illustrate the technical solutions of the present application, but not limit them; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. An adaptive noise reduction filtering method, characterized in that, Includes the following steps: S1. Acquire dynamic signals from signal receiving equipment. The corresponding time-domain signal is obtained after preprocessing. ,in The signal length; S2. Obtained through time-frequency transformation The corresponding time-frequency representation matrix , used to indicate the first Frame, First The complex amplitude of the frequency point, and according to Calculate the amplitude spectrum ; S3. Employ a dual-time-scale noise estimation method to analyze the amplitude spectrum along the time axis. Extracting short-time noise amplitude and long-term noise amplitude Introducing dynamic weights Will and Perform fusion to generate basic noise amplitude ; Dynamic weights are introduced. Will and Perform fusion to generate basic noise amplitude The specific content includes: Introducing dynamic weights To balance the contributions of short-time and long-time estimates: ； ； In the formula, As the initial weights, For time frame indexing, , The frequency index is the frequency index of the time-frequency representation matrix. , , The number of points in the time-frequency transformation. The imaginary unit, and for The upper and lower bounds of the range satisfy 0 ≤ < ≤1; Then the basic noise amplitude for: ； S4. Based on Calculate the signal-to-noise ratio ,according to Calculate the minimum amplitude threshold And based on dynamic noise scaling factor as well as Calculate and estimate noise amplitude ; S5. Based on and Calculate the noise reduction spectrum amplitude Preserving the original phase information, the inverse transform is used to... Reconstructed into a denoised time-domain signal .

2. The adaptive noise reduction filtering method according to claim 1, characterized in that, The specific preprocessing steps in S1 include: Acquire dynamic signals from signal receiving devices. ,right DC offset removal preprocessing is performed to remove low-frequency interference, resulting in the preprocessed time-domain signal. .

3. The adaptive noise reduction filtering method according to claim 1, characterized in that, The specific content of S2 includes: Obtained through time-frequency transformation The corresponding time-frequency representation matrix The time-frequency transformation methods specifically include: Short Time Fourier Transform (STFT), Fast Fourier Transform (FFT), wavelet transform, or wavelet packet transform. In the time-frequency representation matrix Based on this, the amplitude spectrum is calculated. : 。 4. The adaptive noise reduction filtering method according to claim 2, characterized in that, The specific content of time-frequency transformation using the Short Time Fourier Transform (STFT) method includes: ； ； ； In the formula, The window length represents the number of sampling points in each time window. For frame shift, This indicates that the input dynamic signal is at the 1st... Temporal samples of the frame; Represents the window function. Indicates the total number of frames. This represents the floor function. The frequency index is the frequency index of the time-frequency representation matrix. , , The number of points in the time-frequency transformation. The imaginary unit; For time frame indexing, ; This represents the number of overlapping sampling points between adjacent frames.

5. The adaptive noise reduction filtering method according to claim 1, characterized in that, S3 Analysis of amplitude spectrum along the time axis Extracting short-time noise amplitude and long-term noise amplitude The specific content includes: A dual-timescale noise estimation method is employed, and the amplitude spectrum is analyzed. Calculate the short-time noise amplitude based on the statistical distribution along the time axis. and long-term noise amplitude : ； ； In the formula, and They represent the first quantiles and the quantiles; and These are the quantile thresholds for short-time and long-time noise estimation, respectively; Reflects the local background level of noise. Capture the overall trend and possible sudden changes in noise.

6. The adaptive noise reduction filtering method according to claim 1, characterized in that, S4 based on Calculate the signal-to-noise ratio ,according to Calculate the minimum amplitude threshold The specific content includes: Based on the fundamental noise amplitude Calculate the signal-to-noise ratio Used to quantize the relative strength of signal and noise: ； based on Calculate the minimum amplitude threshold : ； In the formula, This is a scaling factor that controls the range of the minimum amplitude threshold, ensuring that key signal features are not weakened.

7. The adaptive noise reduction filtering method according to claim 6, characterized in that, S4 based on dynamic noise scaling factor as well as Calculate and estimate noise amplitude The specific content includes: Introducing a dynamic noise scaling factor To dynamically adjust the noise amplitude: ； in, This is the scaling factor, which controls the sensitivity of the adjustment factor. These are the initial weights; Based on dynamic noise scaling factor Calculate and estimate noise amplitude : 。 8. The adaptive noise reduction filtering method according to claim 1, characterized in that, The specific content of S5 includes: in accordance with and Calculate the noise reduction spectrum amplitude : ； Preserve the original phase and recover the denoised complex spectrum. : ； The inverse transform is used to reconstruct the time-domain signal from the denoised spectrum. : ； In the formula, For overlapping windowing functions.

9. The adaptive noise reduction filtering method according to claim 1, characterized in that, Also includes: S6. By monitoring the noise reduction time-domain signal in real time. Quality, generating performance evaluation reports.

10. An adaptive noise reduction filtering system, based on the adaptive noise reduction filtering method according to any one of claims 1-9, characterized in that, include: The preprocessing module is used to acquire dynamic signals from the signal receiving device. The corresponding time-domain signal is obtained after preprocessing. ,in The signal length; The amplitude spectrum generation module is used to obtain the amplitude spectrum through time-frequency transformation. The corresponding time-frequency representation matrix , used to indicate the first Frame, First The complex amplitude of the frequency point, and according to Calculate the amplitude spectrum ; The dual-timescale noise estimation module is used to analyze the amplitude spectrum along the time axis using a dual-timescale noise estimation method. Extracting short-time noise amplitude and long-term noise amplitude Introducing dynamic weights Will and Perform fusion to generate basic noise amplitude ; Denoising module, used for based Calculate the signal-to-noise ratio ,according to Calculate the minimum amplitude threshold And based on dynamic noise scaling factor as well as Calculate and estimate noise amplitude ; The noise reduction and signal reconstruction module is used to determine the signal based on the noise reduction and signal reconstruction module. and Calculate the noise reduction spectrum amplitude Preserving the original phase information, the inverse transform is used to... Reconstructed into a denoised time-domain signal .

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements an adaptive noise reduction filtering method as described in any one of claims 1-9.

12. An electronic device, comprising: A processor and a memory, the memory being used to store one or more programs; characterized in that, when the one or more programs are executed by the processor, an adaptive noise reduction filtering method as described in any one of claims 1-9 is implemented.

Images