Distributed acoustic sensing system environment noise monitoring and event identification method, system and medium

Abstract

The invention relates to a distributed acoustic sensing system environment noise monitoring and event recognition method and system and a medium, and the method comprises the following steps: obtaining real-time environment sound signals f (t) at different distances from a distributed acoustic sensing system; obtaining a frequency domain signal # imgabs0 # in the sound signal through Fourier transform, wherein the frequency domain signal # imgabs0 # is called S for short; based on the frequency domain signal, extracting a characteristic value corresponding to each event by adopting a double-domain self-adaptive signal peak value extraction method based on dynamic morphological coupling; classifying sound signal features through convolutional neural network machine learning, and judging corresponding events; according to the method and the device, diversified noise types can be effectively processed, and real-time monitoring of key infrastructures is realized.

Description

Technical Field

[0001] The present application relates to the technical field of optical fiber sensing, and particularly relates to a method, system and medium for environmental noise monitoring and event recognition in a distributed acoustic sensing system. Background Art

[0002] The existing monitoring technologies mainly face the following limitations: in the field of environmental pollution control, traditional noise monitoring devices mainly rely on sound level meters, wireless sensor networks or Internet of Things node networks for data collection. Although such devices can achieve basic noise intensity detection, there are technical bottlenecks such as significant response delays, high equipment layout costs, difficult power consumption management, and poor terrain adaptability. Especially in the face of a composite noise field caused by traffic noise sources (such as heavy vehicles), construction vibration sources, and industrial equipment operation, the low-frequency vibration energy will be conducted along the surface medium to the underground pipe network system, causing potential safety hazards such as fatigue damage to the cable sheath structure and deterioration of optical fiber transmission performance. The traditional monitoring system is difficult to synchronously capture the spatio-temporal correlation characteristics of acoustic signals and conducted vibrations, which restricts the risk early warning ability of underground infrastructure. Summary of the Invention

[0003] The purpose of the embodiments of the present application is to overcome the deficiencies of the prior art, and provides a method, system and medium for environmental noise monitoring and event recognition in a distributed acoustic sensing system, which does not require cumbersome signal preprocessing, reduces the computational complexity, and is particularly suitable for safety monitoring applications that require rapid response.

[0004] To achieve the above purpose, the present application provides the following technical solutions:

[0005] In a first aspect, the embodiments of the present application provide a method for environmental noise monitoring and event recognition in a distributed acoustic sensing system, including the following steps:

[0006] Obtain the real-time environmental sound signal f(t) at different distances from the distributed acoustic sensing system;

[0007] Obtain the frequency-domain signal in the sound signal through Fourier transform That is: S

[0008] Based on the frequency-domain signal, adopt a dual-domain adaptive signal peak extraction method based on dynamic morphology coupling to extract the characteristic values corresponding to each event;

[0009] Classify the sound signal features through convolutional neural network machine learning to determine the corresponding event;

[0010] Report the noise determination result.

[0011] The specific operation of adopting a dual-domain adaptive signal peak extraction method based on dynamic morphology coupling to extract the characteristic values corresponding to each event based on the frequency-domain signal is as follows,

[0012] Step 1: Dynamic variable scale morphological base separation;

[0013] Step 2: Differentiation-energy dual-domain enhancement;

[0014] Step 3: Closed-loop morphological peak location.

[0015] The dynamic variable scale morphological base separation is specifically as follows:

[0016] Step 1.1: Generate a sequence of increasing structural element exponentials

[0017] ①Define the upper limit of the number of iterations Niter;

[0018] ②Calculate the initial structural element length:

[0019] Where fs is the sampling rate and k is the iteration number;

[0020] Step 1.2: Multi-scale morphological opening operation

[0021] ① Create horizontal structural elements: SE k =strel('rectangle',[1,L se (k)])

[0022] ② Perform morphological opening operation: Opened k =imopen(S-Trend k-1 ,SE k )

[0023] Where S is the original signal, Trendk is the trend estimate of the kth iteration

[0024] Step 1.3: Savitzky-Golay Trend Forecast Correction

[0025] ①Set filter parameters:

[0026] Polynomial order d = 2 (quadratic polynomial)

[0027] Window length W k =2L se (k)+1

[0028] ② Trend correction update:

[0029] Treand k =α·Treand k-1 +(1-α)·sgolayfilt(Opened k ,d,W k )

[0030] Among them, α ∈ [0.6, 0.8] is the smoothing coefficient

[0031] Step 1.4: Trend convergence judgment

[0032] Calculate the relative change rate:

[0033]

[0034] When Δ k <ε (threshold ε = 0.01), stop the iteration and output the final trend base. The differential - energy double - domain enhancement is specifically as follows

[0035] Step 2.1: Second - order differential sharpening

[0036] Calculate the residual signal:

[0037] Residual = S - Trend final

[0038] Perform central difference:

[0039]

[0040] Enhanced signal construction:

[0041]

[0042] Step 2.2: Dynamic energy window statistics

[0043] ① Define the sliding window length:

[0044]

[0045] ② Calculate the moving average energy:

[0046]

[0047] Step 2.3: Dual - threshold generation

[0048] ① Calculate the background statistic:

[0049]

[0050] IQR energy = iqr(Energy)

[0051] ② Generate the adaptive threshold:

[0052] Thresh peak = Median energy + 3×IQR energy .

[0053] The specific closed-loop morphological peak positioning is as follows:

[0054] Step 3.1: Binary mask generation

[0055] ① Initial marking:

[0056]

[0057] ② Closing operation for noise reduction:

[0058]

[0059] Step 3.2: Accurate peak positioning

[0060] ① Extract candidate regions:

[0061] Candidate = Residual ⊙ Mask clean

[0062] ② Set peak recognition conditions:

[0063] Minimum prominence P min

[0064] Minimum peak spacing

[0065]

[0066] ③ Apply the peak detection algorithm:

[0067] PeakLocations = findpeaks(Candidate, 'MinPeakProminence', P min , 'MinPeakDistance', I.

[0068] Classify the sound signal features through convolutional neural network machine learning, and determine the corresponding events specifically as follows:

[0069] Data preprocessing and feature structuring;

[0070] Select a 2D CNN architecture suitable for spectrogram-like image features:

[0071] Training and optimization;

[0072] Evaluation and tuning;

[0073] Expand the optimization strategy to accurately associate feature values with target event categories.

[0074] In a second aspect, an embodiment of the present application provides a distributed acoustic sensing system environmental noise monitoring and event recognition system, including a memory and a processor. The memory includes a program for the distributed acoustic sensing system environmental noise monitoring and event recognition method. When the program for the distributed acoustic sensing system environmental noise monitoring and event recognition method is executed by the processor, the above-mentioned steps are implemented.

[0075] In a third aspect, an embodiment of the present application provides a computer-readable storage medium. The computer-readable storage medium stores program code. When the program code is executed by a processor, the steps of the above-mentioned distributed acoustic sensing system environmental noise monitoring and event recognition method are implemented.

[0076] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0077] The noise monitoring and recognition method of the present invention improves the adaptability and robustness of the distributed acoustic sensing (DAS) system in complex and dynamic environments.

[0078] A hybrid model combining 1D-CNN and 2D-CNN architectures can effectively process diverse noise types and achieve real-time monitoring of critical infrastructure.

[0079] The method of the present invention does not require cumbersome signal preprocessing, reduces computational complexity, and is particularly suitable for security monitoring applications that require rapid response. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] To more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required to be used in the embodiments of the present application. It should be understood that the following drawings only show some embodiments of the present application and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0081] Figure 1 It is a time-domain signal diagram of a single-distance sound;

[0082] Figure 2 It is a frequency-domain signal diagram of a sound at an abnormal distance point;

[0083] Figure 3 It is an effect diagram of dynamic base signal separation;

[0084] Figure 4 It is a residual signal and precise peak detection diagram;

[0085] Figure 5 It is a method flow chart of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0086] The technical solutions in the embodiments of the present application will be described below with reference to the accompanying drawings in the embodiments of the present application. It should be noted that like reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0087] The term "comprising", "including" or any other variant thereof is intended to cover a non-exclusive inclusion, such that a process, method, article or apparatus comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or apparatus. Without further limitation, an element defined by the phrase "comprising a..." does not exclude the presence of additional identical elements in the process, method, article or apparatus comprising the element.

[0088] Terms such as "first", "second", etc. are only used to distinguish one entity or operation from another entity or operation, and cannot be construed as indicating or implying relative importance, nor can it be construed as requiring or implying any actual relationship or order between these entities or operations.

[0089] Please refer to Figure 5 , a method for environmental noise monitoring and event recognition in a distributed acoustic sensing system, comprising the following steps:

[0090] Obtain real-time environmental sound signals at different distances from the distributed acoustic sensing system;

[0091] Obtain the frequency-domain signal in the sound signal through Fourier transform;

[0092] Based on the frequency-domain signal, adopt a dual-domain adaptive signal peak extraction method based on dynamic morphology coupling to extract the characteristic values corresponding to each event;

[0093] Classify the sound signal features through convolutional neural network machine learning to determine the corresponding events;

[0094] Report the noise determination result.

[0095] The specific process of adopting the dual-domain adaptive signal peak extraction method based on dynamic morphology coupling to extract the characteristic values corresponding to each event based on the frequency-domain signal is as follows:

[0096] Step 1: Dynamic variable-scale morphological basis separation;

[0097] Step 2: Differential-energy dual-domain enhancement;

[0098] Step 3: Closed-loop morphological peak positioning.

[0099] The specific dynamic variable-scale morphological basis separation is as follows:

[0100] Step 1.1: Generation of the structural element index increasing sequence

[0101] ① Define the upper limit of the iteration times Niter;

[0102] ② Calculate the initial length of the structural element:

[0103] where fs is the sampling rate and k represents the iteration sequence number;

[0104] Step 1.2: Multi-scale morphological opening operation

[0105] ① Create a horizontal structural element: SE k = strel('rectangle',[1,L se (k)])

[0106] ② Perform the morphological opening operation: Opened k = imopen(S-Trend k-1 ,SE k )

[0107] where S is the original signal and Trendk is the trend estimate of the k-th iteration

[0108] Step 1.3: Savitzky-Golay trend prediction and correction

[0109] ① Set the filter parameters:

[0110] The polynomial order d = 2 (quadratic polynomial)

[0111] The window length W k = 2L se (k)+1

[0112] ② Trend correction and update:

[0113] Treand k = α·Treand k-1 +(1-α)·sgolayfilt(Opened k ,d,W k )

[0114] where α∈[0.6,0.8] is the smoothing coefficient

[0115] Step 1.4: Trend convergence judgment

[0116] Calculate the relative change rate:

[0117]

[0118] When Δ k <ε (threshold ε = 0.01), stop the iteration and output the final trend basis. The differential-energy dual-domain enhancement is specifically as follows:

[0119] Step 2.1: Second-order differential sharpening

[0120] Calculate the residual signal:

[0121] Residual = S - Trend final

[0122] Perform central difference:

[0123]

[0124] Enhanced signal construction:

[0125]

[0126] Step 2.2: Dynamic energy window statistics

[0127] ① Define the sliding window length:

[0128]

[0129] ② Calculate the moving average energy:

[0130]

[0131] Step 2.3: Dual-threshold generation

[0132] ① Calculate the background statistic:

[0133]

[0134] IQR energy = iqr(Energy) ② Generate the adaptive threshold:

[0135] Thresh peak = Median energy + 3 × IQR energy .

[0136] The closed-loop morphological peak positioning is specifically as follows:

[0137] Step 3.1: Binary mask generation

[0138] ① Initial marking:

[0139]

[0140] ② Closing operation for noise reduction:

[0141]

[0142] Step 3.2: Accurate peak location

[0143] ① Extract candidate regions:

[0144] Candidate = Residual ⊙ Mask clean

[0145] ② Set peak recognition conditions:

[0146] Minimum prominence Pmin

[0147] Minimum peak distance

[0148]

[0149] ③ Apply peak detection algorithm:

[0150] PeakLocations = findpeaks(Candidate, 'MinPeakProminence', P min , 'MinPeakDistance', I.

[0151] Classify the features of the sound signal through convolutional neural network machine learning, and determine the corresponding event specifically as

[0152] (1) Data preprocessing and feature structuring

[0153] Input format:

[0154] The feature is the time-frequency spectrum, regarded as a two-dimensional input (similar to a single-channel grayscale image of an RGB image), with dimensions (height, width, 1).

[0155] Normalization:

[0156] Normalize the features (such as scaling to [0, 1] or normalizing to a mean of 0 and a variance of 1).

[0157] Annotation matching:

[0158] Ensure that each feature sample corresponds to the correct event category label (one-hot encoding is commonly used in classification tasks).

[0159] Select the type of CNN architecture

[0160] 2D CNN (suitable for image features such as spectrograms):

[0161] Input shape: (number of frequency bands, number of time frames, 1) (single channel) or (number of frequency bands, number of time frames, number of channels) (multi-feature fusion).

[0162] Hierarchical design: The Conv2D layer extracts local frequency and time domain patterns; MaxPooling2D reduces the dimension.

[0163] 1D CNN (suitable for time series features):

[0164] Input shape: (number of time steps, number of features).

[0165] Hierarchical design: The Conv1D layer captures local time patterns (such as the time correlation of pronunciation segments). MaxPooling1D compresses the time dimension.

[0166] (3) Training and optimization

[0167] Loss function: For multi-classification: CategoricalCrossentropy (labels need to be one-hot). For integer labels: SparseCategoricalCrossentropy.

[0168] Optimizer: Use Adam (default learning rate 1e-3) or adjust the learning rate (such as 1e-4).

[0169] Regularization: Add a Dropout layer (such as a dropout rate of 0.5) or L2 regularization to prevent overfitting.

[0170] Data augmentation: Add noise, time shift, and speed change at the original audio signal stage; if the features are in image form, slight deformation, occlusion, etc. can be done to the spectrogram for enhancement.

[0171] (4) Evaluation and tuning

[0172] Validation metrics: Accuracy, F1-score (especially when the classes are imbalanced). Confusion matrix analysis of error patterns.

[0173] Early Stopping: Monitor the validation set loss, and terminate training if it does not decrease for several consecutive rounds.

[0174] Hyperparameter tuning: Adjust the convolutional kernel size, number of layers, number of filters, learning rate, etc.

[0175] (5) Extended optimization strategies

[0176] Transfer of pre-trained models: If the data volume is small, the feature extraction layers of pre-trained audio models (such as VGGish, pre-trained ResNet) can be reused.

[0177] Hybrid models: Combine RNN (such as LSTM) to handle long-term time dependencies, for example, connect an LSTM layer after the CNN. Through the above process, the CNN can automatically learn discriminative patterns from sound features and accurately associate feature values with target event classes.

[0178] In the DAS system, the time-domain signals of sounds at different distances can be obtained. For a single distance condition, the time-domain signal of the sound can be obtained. As Figure 1 shown.

[0179] Once the monitoring end detects that the sound signal at a certain distance is relatively abnormal, it can judge that a special event has occurred at that distance position. And through the time-domain signal of the sound at this distance point, the type of the event will be qualitatively determined by feature extraction below.

[0180] First, the time-domain signal is transformed into a frequency-domain signal through Fourier transform. As Figure 2 shown.

[0181] It can be seen that the signal is a function with a decreasing trend similar to a negative exponential function. Next, it is necessary to suppress the noise of the signal and remove the base signal of the signal.

[0182] As Figure 3 shown, it is the base signal after reducing the noise of the frequency-domain signal. Subtracting this base signal from the original frequency-domain signal, a residual signal is obtained. The residual signal is as Figure 4 shown.

[0183] As Figure 4 shown, it is the signal after the spectral sound signal is processed by the algorithm. Each peak of the signal is the characteristic value for subsequent judgment of the event category. The position is marked by the Figure 4 triangle in.

[0184] Collect the data of various known event categories, and divide the regions according to the position and size of the characteristic peaks. The type of the abnormal event can be judged through the 2D CNN network. Judge the sound event that occurs at a specific time and a specific distance. Realize the monitoring of environmental noise anomalies.

[0185] The embodiment of the present application provides a distributed acoustic sensing system environmental noise monitoring and event recognition system, including a memory and a processor. The memory includes a program of the distributed acoustic sensing system environmental noise monitoring and event recognition method. When the program of the distributed acoustic sensing system environmental noise monitoring and event recognition method is executed by the processor, the above-mentioned steps are implemented.

[0186] The embodiment of the present application provides a computer-readable storage medium. The computer-readable storage medium stores program codes. When the program codes are executed by a processor, the steps of the above-mentioned distributed acoustic sensing system environmental noise monitoring and event recognition method are implemented.

[0187] The above are only embodiments of the present application and are not intended to limit the protection scope of the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

Claims

1. A method for environmental noise monitoring and event recognition in a distributed acoustic sensing system, characterized in that, It includes the following steps: Obtain the real-time environmental sound signal f(t) at different distances from the distributed acoustic sensing system; Obtain the frequency-domain signal in the sound signal through Fourier transform Abbreviated as S; Based on the frequency-domain signal, adopt a dual-domain adaptive signal peak extraction method based on the coupling of dynamic morphology to extract the eigenvalue corresponding to each event; Classify the sound signal features through convolutional neural network machine learning to determine the corresponding event; Report the noise determination result.

2. The method for environmental noise monitoring and event recognition of a distributed acoustic sensing system according to claim 1, wherein The specific process of adopting a dual-domain adaptive signal peak extraction method based on the coupling of dynamic morphology to extract the eigenvalue corresponding to each event based on the frequency-domain signal is as follows: Step 1: Dynamic variable-scale morphological basis separation; Step 2: Differential-energy dual-domain enhancement; Step 3: Closed-loop morphological peak positioning.

3. A method for environmental noise monitoring and event recognition in a distributed acoustic sensing system according to claim 2, characterized in that, The specific process of the dynamic variable-scale morphological basis separation is as follows: Step 1.1: Generation of structural element exponential increasing sequence ① Define the upper limit of the iteration number Niter; ② Calculate the length of the initial structural element: where fs is the sampling rate and k represents the iteration number; Step 1.2: Multi-scale morphological opening operation ①Create a horizontal structure element: SE k = strel('rectangle', [1, L se (k)]) ② Perform morphological opening operation: Opened k = imopen(S-Trend k-1 , SE k ) In the formula, S is the original signal and Trendk is the trend estimation of the k-th iteration Step 1.3: Savitzky-Golay trend prediction and correction ① Set the filter parameters: The polynomial order d = 2 (quadratic polynomial) Window length W k = 2L se (k) + 1 ② Trend correction and update: Trend k = α·Trend k-1 + (1 - α)·sgolayfilt(Opened k , d, W k ) where α ∈ [0.6, 0.8] is the smoothing coefficient Step 1.4: Trend convergence judgment Calculate the relative change rate: When Δ k <ε (threshold ε = 0.01), stop the iteration and output the final trend basis.

4. A method for environmental noise monitoring and event recognition in a distributed acoustic sensing system according to claim 3, characterized in that, The specific process of the differential-energy dual-domain enhancement is as follows: Step 2.1: Second-order differential sharpening Calculate the residual signal: Residual=S-Trend final Perform central difference: Enhanced signal construction: Step 2.2: Dynamic energy window statistics ① Define the sliding window length: ② Calculate the moving average energy: Step 2.3: Dual-threshold generation ① Calculate the background statistic: IQR energy = iqr(Energy) ② Generate the adaptive threshold: Thresh peak = Median energy + 3 × IQR energy .

5. A method for environmental noise monitoring and event recognition in a distributed acoustic sensing system according to claim 4, characterized in that, The specific process of the closed-loop morphological peak positioning is as follows: Step 3.1: Binary mask generation ① Initial marking: ② Closing operation for noise reduction: Step 3.2: Accurate peak positioning ① Extract the candidate region: Candidate=Residual☉Mask clean ② Set the peak recognition conditions: Minimum prominence Pmin Minimum peak spacing ③ Apply the peak detection algorithm: PeakLocations = findpeaks(Candidate, 'MinPeakProminence', P min , 'MinPeakDistance',, I。 6. A method for environmental noise monitoring and event recognition in a distributed acoustic sensing system according to claim 1, characterized in that, The specific process of classifying the sound signal features through convolutional neural network machine learning to determine the corresponding event is as follows: Data preprocessing and feature structuring; Select a 2D CNN architecture suitable for the image features of spectrograms; Training and optimization; Evaluation and tuning; Expand the optimization strategy to accurately associate the eigenvalue with the target event category.

7. A distributed acoustic sensing system for environmental noise monitoring and event recognition, characterized in that, It includes a memory and a processor. The memory includes a program for the method of distributed acoustic sensing system environmental noise monitoring and event recognition. When the program for the method of distributed acoustic sensing system environmental noise monitoring and event recognition is executed by the processor, it realizes the steps described in any one of claims 1-6.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program codes. When the program codes are executed by the processor, it realizes the steps of the method for distributed acoustic sensing system environmental noise monitoring and event recognition described in any one of claims 1 to 6.

Images