Chemical gas leakage safety detection method and system based on DAS

Abstract

The invention discloses a chemical gas leakage safety detection method and system based on DAS, and relates to the technical field of leakage detection. Acquiring global signal data of the target area, and performing security detection according to the global signal data to obtain an anomaly detection result; determining a sub-region according to an anomaly detection result, and obtaining local signal data of the sub-region; and detecting the local signal data to obtain a local detection result, and performing safety early warning according to the local detection result. Global signals are collected through the DAS, the pipeline abnormal area is rapidly recognized through the distributed characteristics of the global signals, the detection workload is reduced, and the response efficiency is improved; the DAS cooperates with the infrasonic wave sensor, the former provides an abnormal orientation, and the latter focuses local details and accurately captures micro-leakage features.

Description

Technical Field

[0001] The present invention belongs to the technical field of leakage detection, and in particular relates to a chemical gas leakage safety detection method and system based on DAS. Background Art

[0002] With the rapid development of the chemical industry, chemical gas detection technology has undergone a major transformation, from simple qualitative analysis to precise quantitative analysis. From early chemical reagent test strips to today's high-precision sensors and intelligent analysis systems, detection efficiency and accuracy have been significantly improved, providing a solid foundation for chemical safety and environmental protection.

[0003] Patent number: CN119915446A, discloses a gas pipeline leakage acoustic detection device and method based on CPO-VMD and multi-feature extraction; collects the acoustic wave signal of the gas pipeline leakage; combines the CPO algorithm and variational mode decomposition to adaptively decompose the original signal; reconstructs and reduces noise of key modal components through correlation coefficients; extracts the multidimensional features of the reconstructed signal, screens and selects the signal features, and detects and classifies gas pipeline leaks through a classification detector.

[0004] Existing chemical pipeline gas leak detection methods have problems such as low detection efficiency, large workload, inability to quickly locate abnormal areas, and difficulty in accurately capturing micro-leakage characteristics, making it difficult to meet the needs of efficient and accurate leak detection. Summary of the Invention

[0005] The purpose of the present invention is to solve the problems of low detection efficiency, large workload, inability to quickly locate abnormal areas, and difficulty in accurately capturing micro-leakage characteristics in existing chemical pipeline gas leakage detection methods, which are difficult to meet the needs of efficient and accurate leakage detection, and propose a chemical gas leakage safety detection method and system based on DAS.

[0006] In a first aspect of the present invention, a DAS-based chemical gas leakage safety detection method is first proposed, the method comprising: Acquiring global signal data of a target area, and performing safety detection based on the global signal data to obtain an anomaly detection result; Determine a sub-region according to the abnormality detection result, and obtain local signal data of the sub-region; the target region includes multiple sub-regions; The local signal data is detected to obtain the local detection results, and a safety warning is issued based on the local detection results.

[0007] Optionally, before performing security detection according to the global signal data to obtain an abnormality detection result, the method further includes: Acquiring historical data, preprocessing the historical data to obtain target historical data, determining the length of a sliding window, and segmenting the target historical data according to the sliding window to obtain multiple signal segments; the historical data is leakage sample signals under different pressure and leakage aperture conditions; Multiple upper envelopes are obtained by extracting each signal segment, and each upper envelope is subjected to quadratic Gaussian smoothing and Lagrangian downsampling to obtain multiple target upper envelopes. A sample library is constructed based on each target upper envelope; the signal segments and upper envelopes correspond one to one.

[0008] Optionally, performing security detection based on the global signal data to obtain an abnormality detection result includes: Step 1: Decomposing the global signal data by variational mode decomposition to obtain multiple intrinsic mode functions, determining the frequency of each intrinsic mode function, and reconstructing the intrinsic mode functions less than a preset frequency to obtain a noise reduction signal; Step 2: Segment the denoised signal according to the sliding window to obtain multiple segmented signals, and perform target operations on each segmented signal to obtain a real-time signal; the target operations include: upper envelope extraction, quadratic Gaussian smoothing, and Lagrangian downsampling operations, and the real-time signals include x1 and x2; Step 3: Match the real-time signal according to the sample library to obtain the minimum cumulative distance matrix D1 and D2, calculate the total cumulative distance matrix according to D1 and D2, and determine the matching signal according to the total cumulative distance matrix; Step 4: Determine the step size of the sliding window, repeat steps 2 and 3, and update the real-time signal and the matching signal to obtain a signal sequence; the signal sequence includes multiple matching signals; Step 5: Set a leakage warning threshold. If all matching signals in the signal sequence are smaller than the leakage warning threshold, add an abnormality label to the global signal data to obtain an abnormality detection result.

[0009] Optionally, detecting the local signal data to obtain a local detection result includes: The local signal data is converted into a local spectrum through short-time Fourier transform, and the local spectrum is extracted through the target YOLOv5 model to obtain the local detection result; Improvements to the target YOLOv5 model include: The C3 module in the YOLOv5 model is replaced with the target C3 module, and the neck structure of the YOLOv5 model is replaced with the target neck structure to obtain the target YOLOv5 model; The working principle of the target C3 module includes: Obtain an initial feature map, input the initial feature map into a CBS module to obtain a first feature map, input the first feature map into a PConv module, a CBS module, and a CBS module in sequence to obtain a second feature map, and add the first feature map and the second feature map to obtain a third feature map; The initial feature map is input into the CBS module to obtain a fourth feature map, the third feature map and the fourth feature map are added to obtain a fifth feature map, the fifth feature map is input into the CBS module to obtain a sixth feature map, and the sixth feature map is sequentially input into four SPAB modules to obtain a seventh feature map.

[0010] Optionally, the working principle of the target neck structure includes: Obtain a feature tensor set of the backbone network, input the first feature tensor into the DBS module to obtain a fourth feature tensor, upsample the fourth tensor to obtain a target fourth tensor, and concatenate the four feature tensors, the target fourth feature tensor, and the second feature tensor to obtain a fifth feature tensor; the feature tensor set includes: the first feature tensor, the second feature tensor, and the third feature tensor; Inputting the fifth feature tensor into the target C3 module and the DBS module in sequence to obtain a sixth feature tensor, upsampling the sixth feature tensor to obtain a target sixth feature tensor, and concatenating the third feature tensor and the target sixth feature tensor to obtain a seventh feature tensor; Inputting the seventh feature tensor into the target CSPStage module to obtain an eighth feature tensor, inputting the eighth feature tensor into the CBS module to obtain a ninth feature tensor, concatenating the ninth feature tensor and the sixth feature tensor to obtain a tenth feature tensor, inputting the tenth feature tensor into the target CSPStage module to obtain a first target output, and using the eighth feature tensor as a second target output; The working process of the target CSPStage module includes: Obtain an initial target feature map, input the initial target feature map into a 1×1Conv module to obtain a first target feature map, input the first target feature map into a target Conv module to obtain a second target feature map, input the second target feature map into a Biformer module and a 3×3Conv module in sequence to obtain a third target feature map, add the first target feature map and the third target feature map to obtain a fourth target feature map, concatenate the fourth target feature map and the first target feature map to obtain a fifth target feature map, input the fifth target feature map into a 1×1Conv module to obtain a sixth target feature map, and use the sixth target feature map as the output of the target CSPStage module.

[0011] In the second aspect of the present invention, a DAS-based chemical gas leakage safety detection system is proposed, comprising: a global detection module, a local detection module, and a safety warning module: The global detection module is used to obtain global signal data of the target area and perform safety detection based on the global signal data to obtain an abnormality detection result; The local detection module is used to determine a sub-region according to the abnormality detection result and obtain local signal data of the sub-region; the target region includes multiple sub-regions; The safety warning module is used to detect local signal data to obtain local detection results, and issue a safety warning based on the local detection results.

[0012] Optionally, the system further includes: a data acquisition module and a sample library construction module: The data acquisition module is used to acquire historical data, pre-process the historical data to obtain target historical data, determine the length of the sliding window, and segment the target historical data according to the sliding window to obtain multiple signal segments; the historical data is leakage sample signals under different pressure and leakage aperture conditions; The sample library construction module is used to extract each signal segment to obtain multiple upper envelopes, perform quadratic Gaussian smoothing and Lagrangian downsampling on each upper envelope to obtain multiple target upper envelopes, and construct a sample library based on each target upper envelope; the signal segments and upper envelopes correspond one to one.

[0013] Optionally, the global detection module includes: a first operation module, a second operation module, a third operation module, a fourth operation module, and a fifth operation module: The first operating module is configured to obtain a plurality of intrinsic mode functions (IMFs) by performing variational mode decomposition on the global signal data, determine the frequency of each IMF, and reconstruct the IMFs with a frequency less than a preset frequency to obtain a noise reduction signal; The second operation module is configured to segment the noise reduction signal according to the sliding window to obtain a plurality of segmented signals, and perform a target operation on each segmented signal to obtain a real-time signal; the target operation includes upper envelope extraction, quadratic Gaussian smoothing, and Lagrangian downsampling operations, and the real-time signal includes x1 and x2; The third operation module is configured to match the real-time signal according to the sample library to obtain minimum cumulative distance matrices D1 and D2, calculate a total cumulative distance matrix according to D1 and D2, and determine a matching signal according to the total cumulative distance matrix; The fourth operation module is configured to determine a step size of the sliding window, repeatedly execute the second operation module and the third operation module, and update the real-time signal and the matching signal to obtain a signal sequence; the signal sequence includes a plurality of matching signals; The fifth operation module is configured to set a leakage warning threshold, and if all matching signals in the signal sequence are smaller than the leakage warning threshold, add an abnormality label to the global signal data to obtain an abnormality detection result.

[0014] Optionally, the local detection module is further used to convert the local signal data through short-time Fourier transform to obtain a local spectrum graph, and perform feature extraction on the local spectrum graph through the target YOLOv5 model to obtain a local detection result; Improvements to the target YOLOv5 model include: The C3 module in the YOLOv5 model is replaced with the target C3 module, and the neck structure of the YOLOv5 model is replaced with the target neck structure to obtain the target YOLOv5 model; The working principle of the target C3 module includes: Obtain an initial feature map, input the initial feature map into a CBS module to obtain a first feature map, input the first feature map into a PConv module, a CBS module, and a CBS module in sequence to obtain a second feature map, and add the first feature map and the second feature map to obtain a third feature map; The initial feature map is input into the CBS module to obtain a fourth feature map, the third feature map and the fourth feature map are added to obtain a fifth feature map, the fifth feature map is input into the CBS module to obtain a sixth feature map, and the sixth feature map is sequentially input into four SPAB modules to obtain a seventh feature map.

[0015] Optionally, the working principle of the target neck structure includes: Obtain a feature tensor set of the backbone network, input the first feature tensor into the DBS module to obtain a fourth feature tensor, upsample the fourth tensor to obtain a target fourth tensor, and concatenate the four feature tensors, the target fourth feature tensor, and the second feature tensor to obtain a fifth feature tensor; the feature tensor set includes: the first feature tensor, the second feature tensor, and the third feature tensor; Inputting the fifth feature tensor into the target C3 module and the DBS module in sequence to obtain a sixth feature tensor, upsampling the sixth feature tensor to obtain a target sixth feature tensor, and concatenating the third feature tensor and the target sixth feature tensor to obtain a seventh feature tensor; Inputting the seventh feature tensor into the target CSPStage module to obtain an eighth feature tensor, inputting the eighth feature tensor into the CBS module to obtain a ninth feature tensor, concatenating the ninth feature tensor and the sixth feature tensor to obtain a tenth feature tensor, inputting the tenth feature tensor into the target CSPStage module to obtain a first target output, and using the eighth feature tensor as a second target output; The working process of the target CSPStage module includes: Obtain an initial target feature map, input the initial target feature map into a 1×1Conv module to obtain a first target feature map, input the first target feature map into a target Conv module to obtain a second target feature map, input the second target feature map into a Biformer module and a 3×3Conv module in sequence to obtain a third target feature map, add the first target feature map and the third target feature map to obtain a fourth target feature map, concatenate the fourth target feature map and the first target feature map to obtain a fifth target feature map, input the fifth target feature map into a 1×1Conv module to obtain a sixth target feature map, and use the sixth target feature map as the output of the target CSPStage module.

[0016] Beneficial effects of the present invention: This paper proposes a chemical gas leak safety detection method based on DAS. The method acquires global signal data from a target area, performs safety checks based on this data, and obtains anomaly detection results. Based on the anomaly detection results, a sub-area is determined and local signal data from that sub-area is obtained. Local detection results are then tested on the local signal data to obtain local detection results, and safety warnings are issued based on these local detection results. By collecting global signals through DAS and leveraging its distributed nature, abnormal pipeline areas can be quickly identified, reducing detection workload and improving response efficiency. The DAS works in conjunction with an infrasonic sensor, with the former providing the location of anomalies and the latter focusing on local details to accurately capture micro-leak characteristics. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The present invention will be further described below with reference to the accompanying drawings.

[0018] Figure 1 A flow chart of a DAS-based chemical gas leakage safety detection method provided in an embodiment of the present invention; Figure 2 A schematic diagram of the target C3 module structure of another DAS-based chemical gas leakage safety detection method provided by an embodiment of the present invention; Figure 3 A schematic diagram of the target neck structure of another DAS-based chemical gas leakage safety detection method provided by an embodiment of the present invention; Figure 4 A framework diagram of a DAS-based chemical gas leak safety detection system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention, in conjunction with the accompanying drawings. Obviously, the described embodiments represent only a portion of the embodiments of the present invention, not all of them. The term "and / or" herein simply describes an association relationship between associated objects, indicating that three possible relationships exist. For example, "A" and "B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, references to "first," "second," and so on in the present invention are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined as "first" or "second" may explicitly or implicitly include at least one of these features. Furthermore, the technical solutions of the various embodiments may be combined, but only if they are achievable by a person of ordinary skill in the art. If a combination of technical solutions contradicts or is unachievable, such combination shall be deemed non-existent and outside the scope of protection claimed by the present invention.

[0020] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0021] The embodiment of the present invention provides a chemical gas leakage safety detection method based on DAS. Figure 1 , Figure 1 A flow chart of a DAS-based chemical gas leak safety detection method provided in an embodiment of the present invention. The method includes the following steps: S101, obtaining global signal data of a target area, performing safety detection based on the global signal data to obtain an anomaly detection result; S102, determining a sub-region according to the anomaly detection result, and obtaining local signal data of the sub-region; S103: Detect local signal data to obtain local detection results, and issue a safety warning based on the local detection results.

[0022] The target area contains multiple sub-areas; According to an embodiment of the present invention, a DAS-based chemical gas leak safety detection method is provided. The DAS collects global signals and uses its distributed characteristics to quickly identify abnormal areas in the pipeline, reducing the detection workload and improving response efficiency. The DAS works in conjunction with the infrasonic sensor. The former provides the abnormal direction, while the latter focuses on local details to accurately capture micro-leak characteristics.

[0023] In one implementation, global signals are collected by DAS, and local signals are collected by infrasound sensors; DAS is deployed on all pipelines, and infrasound sensors are deployed on the pipelines at preset distances (for example, five-meter intervals).

[0024] In one implementation, DAS collects global signal data from a target area. Leveraging its distributed fiber optic sensing capabilities (full pipeline coverage, immunity to electromagnetic interference, and long-distance real-time monitoring), it can perform a comprehensive, blind-spot-free scan of the pipeline's overall operating status. It quickly identifies areas with abnormal signals (such as vibration or temperature anomalies caused by leaks), narrowing the monitoring scope from a "full-line investigation" to "suspicious sub-areas," significantly reducing the workload for subsequent localized inspections and improving system response efficiency. For example, DAS can locate abnormal areas as small as 10 meters within a multi-kilometer pipeline, eliminating the blind spots of manual inspections.

[0025] In one implementation, global detection filters noise: the distributed nature of DAS can eliminate environmental interference (such as occasional vibrations) through the temporal and spatial distribution patterns of signals, and only trigger sub-area detection for persistent abnormal signals; local detection focuses on details: infrasound sensors are deployed in high-density sub-areas (such as 5-meter intervals) to compensate for the insufficient resolution of DAS in detecting local high-frequency signals (such as low-frequency vibrations of micro-leaks), avoiding missed detections caused by signal attenuation or noise with a single technology.

[0026] In one implementation, the advantages of DAS are: long-distance, non-contact monitoring, suitable for identifying high-frequency vibrations or temperature mutations caused by high-pressure leaks, and quickly locking abnormal sections; the advantages of infrasound sensors are: single-point high sensitivity (especially for infrasound waves <20Hz), and high-density sampling of local signals under dense deployment, suitable for capturing the low-frequency characteristics of low-pressure micro-leaks or airborne infrasound signals; synergistic value: global signals provide the "azimuth" of the anomaly, and local signals provide "precise coordinates."

[0027] In one implementation, a safety warning is triggered based on local detection results (high-precision signal analysis of infrasonic sensors), ensuring that the warning information contains "the specific sub-area where the leak occurs + detailed signal characteristics" (such as leak aperture and pressure level); according to the severity of the local detection results (such as signal amplitude and duration), the warning level is dynamically adjusted (such as level one warning for micro leaks and level two warning for high-pressure jet leaks).

[0028] In one embodiment, before performing security detection based on the global signal data to obtain an abnormality detection result, the method further includes: Acquire historical data, pre-process the historical data to obtain target historical data, determine the length of the sliding window, and segment the target historical data according to the sliding window to obtain multiple signal segments; the historical data is leakage sample signals under different pressure and leakage aperture conditions; Multiple upper envelopes are obtained by extracting each signal segment, and each upper envelope is subjected to quadratic Gaussian smoothing and Lagrangian downsampling to obtain multiple target upper envelopes. A sample library is constructed based on each target upper envelope; the signal segments and the upper envelopes correspond one to one.

[0029] In one implementation, a sliding window is used: W = L*f / v, where L represents the total length of the pipeline, v represents the propagation speed of the leakage signal, and f represents the sampling frequency of the signal; the leakage signals of all pipelines are obtained periodically (that is, the sampling frequency) through the DAS.

[0030] In one implementation, preprocessing historical data (e.g., noise reduction and filtering) removes noise and interfering components (e.g., environmental vibration and equipment operating noise) from the original signal, ensuring that only valid leak signatures are retained in the target historical data. By focusing on samples under different pressures and leak apertures, a wide range of leak scenarios can be covered, enabling the subsequently constructed sample library to have stronger generalization capabilities and improving the leak detection model's adaptability to complex operating conditions.

[0031] In one implementation, a sample library is constructed based on leak signals under different pressures, leak apertures, and other conditions, encompassing waveform characteristics from a variety of typical leak scenarios. This enables the system to identify leak signals of varying severity and in different environments during real-time monitoring by matching the diverse templates in the sample library. This improves the generalization capabilities of the detection model and avoids missed or false detections due to a single operating condition. For example, the signal signature differences between a high-pressure, small-aperture leak and a low-pressure, large-aperture leak can be effectively distinguished using the corresponding templates in the sample library.

[0032] In one embodiment, performing security detection based on global signal data to obtain an abnormality detection result includes: Step 1: Decompose the global signal data into multiple intrinsic mode functions through variational mode decomposition, determine the frequency of each intrinsic mode function, and reconstruct the intrinsic mode functions less than a preset frequency to obtain a noise reduction signal; Step 2: Segment the denoised signal according to the sliding window to obtain multiple segmented signals, and perform target operations on each segmented signal to obtain a real-time signal; the target operations include: upper envelope extraction, quadratic Gaussian smoothing, and Lagrangian downsampling operations, and the real-time signals include x1 and x2; Step 3: Match the real-time signal according to the sample library to obtain the minimum cumulative distance matrix D1 and D2, calculate the total cumulative distance matrix based on D1 and D2, and determine the matching signal based on the total cumulative distance matrix; Step 4: Determine the step size of the sliding window, repeat steps 2 and 3, and update the real-time signal and the matching signal to obtain a signal sequence; the signal sequence includes multiple matching signals; Step 5: Set the leakage warning threshold. If the matching signals in the signal sequence are all smaller than the leakage warning threshold, add an abnormal label to the global signal data to obtain the abnormal detection result.

[0033] In one implementation, variational mode decomposition (VMD) adaptively decomposes the global signal into multiple intrinsic mode functions (IMFs) by constructing and solving a variational model, avoiding the modal aliasing problem of traditional empirical mode decomposition (EMD). By reconstructing the infrasonic component of the leakage signal (typically below 20 Hz), IMFs with a dominant frequency below a preset frequency (e.g., 15 Hz) are selected. This effectively separates low-frequency leakage signatures from high-frequency noise (such as pipeline vibration and environmental interference), improving the signal-to-noise ratio. The infrasonic frequency band directly related to the leak is retained, while irrelevant high-frequency noise is suppressed, providing a clean feature input for subsequent analysis.

[0034] In one implementation, a sliding window is used to segment the denoised signal, ensuring that each segment contains complete leakage characteristics (such as bidirectional stress waves). Upper envelope extraction highlights the amplitude variation trend of the signal, quadratic Gaussian smoothing further filters out local noise, and Lagrangian downsampling reduces the data dimension while preserving the waveform contour (for example, reducing high-frequency sampled signals to characteristic frequency bands). The original time-domain signal is converted into noise-resistant feature vectors (such as x1 and x2), reducing the subsequent matching computational effort and improving real-time performance.

[0035] In one implementation, leakage signals from different operating conditions (pressure, aperture) in a sample library are used as templates. DTW is used to calculate the minimum cumulative distance between the real-time signal and the sample. This solves the time axis scaling problem caused by propagation distance and medium differences in leakage signals (such as the difference in signal arrival time at different sensors). This enables precise matching of non-aligned time series, improving the accuracy of leak signal recognition under complex operating conditions.

[0036] In one implementation, the window sliding interval is controlled by setting a step size ΔW, achieving real-time scanning of the entire pipeline while ensuring signal continuity (e.g., overlapping segments to avoid missed detections). The iterative process continuously updates the real-time signal and matching results, capturing dynamic changes in the leak signal (e.g., waveform evolution caused by pressure decay). This balances real-time monitoring with computational efficiency, ensuring coverage and signal integrity for long-distance pipeline monitoring.

[0037] In one embodiment, detecting local signal data to obtain local detection results includes: The local signal data is converted into a local spectrum through short-time Fourier transform, and the local spectrum is extracted through the target YOLOv5 model to obtain the local detection result; Improvements to the target YOLOv5 model include: Replace the C3 module in the YOLOv5 model with the target C3 module, and replace the neck structure of the YOLOv5 model with the target neck structure to obtain the target YOLOv5 model; The Objective C3 module works by: Obtain an initial feature map, input the initial feature map into the CBS module to obtain a first feature map, input the first feature map into the PConv module, the CBS module, and the CBS module in sequence to obtain a second feature map, and add the first feature map and the second feature map to obtain a third feature map; The initial feature map is input into the CBS module to obtain the fourth feature map, the third feature map and the fourth feature map are added to obtain the fifth feature map, the fifth feature map is input into the CBS module to obtain the sixth feature map, and the sixth feature map is input into four SPAB modules in sequence to obtain the seventh feature map; see Figure 2 , Figure 2 A schematic diagram of the target C3 module structure of another DAS-based chemical gas leakage safety detection method provided by an embodiment of the present invention, wherein “⊕” represents addition.

[0038] In one implementation, the original signal is converted into a spectrogram through a short-time Fourier transform, and then the YOLOv5 model is used to extract features from the spectrogram. The YOLOv5 model is used to capture features of the original signal. YOLOv7 and YOLOv8 have higher accuracy than YOLOv5, but their larger model size, higher computational complexity, and higher resource utilization make them difficult to meet real-time requirements. In one implementation, local signal data is converted into a local spectrogram using a short-time Fourier transform (STFT). The target YOLOv5 model then extracts features from the spectrogram to obtain local detection results. This process improves the accuracy and robustness of gas leak detection. The short-time Fourier transform (STFT) is a time-frequency analysis method that decomposes a signal into multiple short-term spectral components, simultaneously capturing the signal's dynamic changes in both time and frequency. This conversion method allows the originally one-dimensional signal data to appear rich in two-dimensional features in the spectrogram. This allows the YOLOv5 model to quickly and accurately extract key features from the spectrogram and identify characteristic patterns of gas leaks, improving classification accuracy and meeting the needs of real-time detection.

[0039] In one implementation, the PConv module is introduced to replace the traditional Bottleneck module, which can reduce computational complexity and memory usage. Combining it with the CBS module effectively reduces memory consumption while maintaining feature diversity, achieving lightweight feature extraction and improving the real-time performance of the model. In one implementation, during feature extraction, feature diversity is preserved by adding feature maps processed by the PConv and CBS modules to the initial feature maps. This residual connection ensures the integrity of feature information and avoids information loss caused by lightweight processing, thereby reducing computational costs while maintaining the model's detection performance.

[0040] In one implementation, a SPAB structure is introduced at the end of feature extraction, further enhancing the spatial representation of features through four consecutive SPAB modules. The SPAB module utilizes activation functions and residual connections to significantly improve the quality of feature information without requiring additional parameters, while also suppressing the interference of redundant noise, thereby improving the model's ability to capture and represent gas leak signal features.

[0041] In one implementation, see Figure 3 , Figure 3 Schematic diagram of the target neck structure of another DAS-based chemical gas leak safety detection method provided in an embodiment of the present invention; the DBS module is composed of a DWConv module, a batch normalization layer (BatchNormalization) and an activation function (such as ReLU or SiLU); the CBS module is composed of a convolutional layer (ConvolutionalLayer), a batch normalization layer (Batch Normalization) and an activation function (such as ReLU or SiLU); the PConv module is a partial convolution (Partial Convolution) structure, the core idea of which is to only calculate the valid part of the input feature map in the convolution operation, thereby significantly reducing the amount of calculation and memory usage; the SPAB module is a structure designed based on a fast parameter-free self-attention mechanism, which aims to enhance the spatial expression ability of features through the attention mechanism while avoiding the introduction of additional parameters; the C3 module is composed of a CBS module and multiple stacked Bottleneck modules; Concat stands for splicing, and Upsample stands for upsampling.

[0042] In one embodiment, the working principle of the target neck structure includes: Obtain a feature tensor set of the backbone network, input the first feature tensor into the DBS module to obtain a fourth feature tensor, upsample the fourth tensor to obtain a target fourth tensor, and concatenate the four feature tensors, the target fourth feature tensor, and the second feature tensor to obtain a fifth feature tensor; the feature tensor set includes: the first feature tensor, the second feature tensor, and the third feature tensor; The fifth feature tensor is sequentially input into the target C3 module and the DBS module to obtain a sixth feature tensor, the sixth feature tensor is upsampled to obtain a target sixth feature tensor, and the third feature tensor and the target sixth feature tensor are concatenated to obtain a seventh feature tensor; The seventh feature tensor is input into the target CSPStage module to obtain the eighth feature tensor, the eighth feature tensor is input into the CBS module to obtain the ninth feature tensor, the ninth feature tensor and the sixth feature tensor are concatenated to obtain the tenth feature tensor, the tenth feature tensor is input into the target CSPStage module to obtain the first target output, and the eighth feature tensor is used as the second target output.

[0043] The working process of the target CSPStage module includes: Obtain an initial target feature map, input the initial target feature map into a 1×1Conv module to obtain a first target feature map, input the first target feature map into a target Conv module to obtain a second target feature map, input the second target feature map into a Biformer module and a 3×3Conv module in sequence to obtain a third target feature map, add the first target feature map and the third target feature map to obtain a fourth target feature map, concatenate the fourth target feature map and the first target feature map to obtain a fifth target feature map, input the fifth target feature map into a 1×1Conv module to obtain a sixth target feature map, and use the sixth target feature map as the output of the target CSPStage module.

[0044] In one implementation, multi-scale feature fusion is achieved by concatenating feature tensors of different scales. Specifically, the upsampled fourth feature tensor is concatenated with the second feature tensor, and the upsampled sixth feature tensor is concatenated with the third feature tensor. This operation integrates feature information at different levels, enhancing the model's ability to perceive multi-scale features and thus improving its ability to capture narrowband information features in the time-frequency spectrogram of the gas leak signal.

[0045] In one implementation, a target CSPStage module constructed using a two-layer routing attention mechanism is introduced to capture cross-sectional feature information. This attention mechanism enhances feature representation by focusing on important regions in the feature map, enabling the network to more effectively learn and extract narrowband information features. This mechanism, combined with the spatial feature extraction capabilities of the convolutional layer, improves the model's performance and generalization. A residual connection is implemented by adding the output feature tensor to the original input feature tensor. This design helps address the vanishing gradient problem in deep networks, allowing the network to learn a residual mapping between input and output, thereby improving the model's training stability and convergence speed, and enhancing the model's ability to learn complex signal features.

[0046] In one implementation, a target Conv module performs a preliminary transformation on features, reducing the number of computational parameters while enhancing the network's nonlinear representation capabilities. This lightweight design expands the feature space without significantly increasing the computational burden, providing a richer information foundation for subsequent feature extraction and further improving the model's efficiency and performance. The target Conv module consists of two branches: the first branch consists of a 1×1 Conv module and a batch normalization layer, and the second branch consists of a 1×1 Conv module, a 1×3 Conv module, a 3×1 Conv module, and a batch normalization layer. The output of the first branch is summed with the output of the second branch to serve as the output of the target Conv module. 1×1 convolution, 1×3 convolution, and 3×1 convolution are used to maintain the same receptive field, while enhancing the network's nonlinear representation capabilities and reducing the number of computational parameters.

[0047] Based on the same inventive concept, the present invention also provides a DAS-based chemical gas leakage safety detection system. Figure 4 , Figure 4 The framework diagram of a DAS-based chemical gas leak safety detection system provided in an embodiment of the present invention includes: a global detection module, a local detection module, and a safety warning module: The global detection module is used to obtain the global signal data of the target area and perform security detection based on the global signal data to obtain anomaly detection results; The local detection module is used to determine the sub-region based on the abnormality detection results and obtain the local signal data of the sub-region; the target region contains multiple sub-regions; The safety warning module is used to detect local signal data to obtain local detection results and issue safety warnings based on the local detection results.

[0048] According to an embodiment of the present invention, a DAS-based chemical gas leak safety detection system is provided. The DAS collects global signals and uses its distributed characteristics to quickly identify abnormal areas in the pipeline, reducing the detection workload and improving response efficiency. The DAS works in conjunction with the infrasonic sensor. The former provides the abnormal direction, while the latter focuses on local details to accurately capture micro-leak characteristics.

[0049] In one embodiment, the system further includes: a data acquisition module and a sample library construction module: The data acquisition module is used to acquire historical data, pre-process the historical data to obtain target historical data, determine the length of the sliding window, and segment the target historical data according to the sliding window to obtain multiple signal segments; the historical data is the leakage sample signal under different pressure and leakage aperture conditions; The sample library construction module is used to extract each signal segment to obtain multiple upper envelopes, perform quadratic Gaussian smoothing and Lagrangian downsampling on each upper envelope to obtain multiple target upper envelopes, and construct a sample library based on each target upper envelope; the signal segments and the upper envelopes correspond one to one.

[0050] In one embodiment, the global detection module includes: a first operation module, a second operation module, a third operation module, a fourth operation module and a fifth operation module: A first operation module is configured to obtain a plurality of intrinsic mode functions (IMFs) by performing variational mode decomposition on the global signal data, determine the frequency of each IMF, and reconstruct the IMFs with a frequency less than a preset frequency to obtain a noise reduction signal; A second operation module is configured to segment the noise reduction signal according to the sliding window to obtain a plurality of segmented signals, and perform a target operation on each segmented signal to obtain a real-time signal; the target operation includes upper envelope extraction, quadratic Gaussian smoothing, and Lagrangian downsampling operations, and the real-time signals include x1 and x2; The third operation module is used to match the real-time signal according to the sample library to obtain the minimum cumulative distance matrix D1 and D2, calculate the total cumulative distance matrix according to D1 and D2, and determine the matching signal according to the total cumulative distance matrix; a fourth operation module, configured to determine a step size of the sliding window, repeatedly execute the second operation module and the third operation module, and update the real-time signal and the matching signal to obtain a signal sequence; the signal sequence includes a plurality of matching signals; The fifth operation module is used to set a leakage warning threshold. If the matching signals in the signal sequence are all smaller than the leakage warning threshold, an abnormality label is added to the global signal data to obtain an abnormality detection result.

[0051] In one embodiment, the local detection module is further configured to convert the local signal data through a short-time Fourier transform to obtain a local spectrum graph, and perform feature extraction on the local spectrum graph through a target YOLOv5 model to obtain a local detection result; Improvements to the target YOLOv5 model include: Replace the C3 module in the YOLOv5 model with the target C3 module, and replace the neck structure of the YOLOv5 model with the target neck structure to obtain the target YOLOv5 model; The Objective C3 module works by: Obtain an initial feature map, input the initial feature map into the CBS module to obtain a first feature map, input the first feature map into the PConv module, the CBS module, and the CBS module in sequence to obtain a second feature map, and add the first feature map and the second feature map to obtain a third feature map; The initial feature map is input into the CBS module to obtain the fourth feature map, the third feature map and the fourth feature map are added to obtain the fifth feature map, the fifth feature map is input into the CBS module to obtain the sixth feature map, and the sixth feature map is sequentially input into four SPAB modules to obtain the seventh feature map.

[0052] In one embodiment, the working principle of the target neck structure includes: Obtain a feature tensor set of the backbone network, input the first feature tensor into the DBS module to obtain a fourth feature tensor, upsample the fourth tensor to obtain a target fourth tensor, and concatenate the four feature tensors, the target fourth feature tensor, and the second feature tensor to obtain a fifth feature tensor; the feature tensor set includes: the first feature tensor, the second feature tensor, and the third feature tensor; The fifth feature tensor is sequentially input into the target C3 module and the DBS module to obtain a sixth feature tensor, the sixth feature tensor is upsampled to obtain a target sixth feature tensor, and the third feature tensor and the target sixth feature tensor are concatenated to obtain a seventh feature tensor; The seventh feature tensor is input into the target CSPStage module to obtain the eighth feature tensor, the eighth feature tensor is input into the CBS module to obtain the ninth feature tensor, the ninth feature tensor and the sixth feature tensor are concatenated to obtain the tenth feature tensor, the tenth feature tensor is input into the target CSPStage module to obtain the first target output, and the eighth feature tensor is used as the second target output.

[0053] The working process of the target CSPStage module includes: Obtain an initial target feature map, input the initial target feature map into a 1×1Conv module to obtain a first target feature map, input the first target feature map into a target Conv module to obtain a second target feature map, input the second target feature map into a Biformer module and a 3×3Conv module in sequence to obtain a third target feature map, add the first target feature map and the third target feature map to obtain a fourth target feature map, concatenate the fourth target feature map and the first target feature map to obtain a fifth target feature map, input the fifth target feature map into a 1×1Conv module to obtain a sixth target feature map, and use the sixth target feature map as the output of the target CSPStage module.

[0054] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A chemical gas leakage safety detection method based on DAS, characterized in that: The method comprises: Acquiring global signal data of a target area, and performing safety detection based on the global signal data to obtain an anomaly detection result; Determine a sub-region according to the abnormality detection result, and obtain local signal data of the sub-region; the target region includes multiple sub-regions; The local signal data is detected to obtain the local detection results, and a safety warning is issued based on the local detection results.

2. A chemical gas leakage safety detection method based on DAS according to claim 1, characterized in that: Before performing safety detection according to the global signal data to obtain an abnormality detection result, the method further includes: Acquiring historical data, preprocessing the historical data to obtain target historical data, determining the length of a sliding window, and segmenting the target historical data according to the sliding window to obtain multiple signal segments; the historical data is leakage sample signals under different pressure and leakage aperture conditions; Multiple upper envelopes are obtained by extracting each signal segment, and each upper envelope is subjected to quadratic Gaussian smoothing and Lagrangian downsampling to obtain multiple target upper envelopes. A sample library is constructed based on each target upper envelope; the signal segments and upper envelopes correspond one to one.

3. A chemical gas leakage safety detection method based on DAS according to claim 1, characterized in that: Performing security detection based on the global signal data to obtain an abnormality detection result includes: Step 1: Decomposing the global signal data by variational mode decomposition to obtain multiple intrinsic mode functions, determining the frequency of each intrinsic mode function, and reconstructing the intrinsic mode functions less than a preset frequency to obtain a noise reduction signal; Step 2: Segment the denoised signal according to the sliding window to obtain multiple segmented signals, and perform target operations on each segmented signal to obtain a real-time signal; the target operations include: upper envelope extraction, quadratic Gaussian smoothing, and Lagrangian downsampling operations, and the real-time signals include x1 and x2; Step 3: Match the real-time signal according to the sample library to obtain the minimum cumulative distance matrix D1 and D2, calculate the total cumulative distance matrix according to D1 and D2, and determine the matching signal according to the total cumulative distance matrix; Step 4: Determine the step size of the sliding window, repeat steps 2 and 3, and update the real-time signal and the matching signal to obtain a signal sequence; the signal sequence includes multiple matching signals; Step 5: Set a leakage warning threshold. If all matching signals in the signal sequence are smaller than the leakage warning threshold, add an abnormality label to the global signal data to obtain an abnormality detection result.

4. A DAS-based chemical gas leakage safety detection method according to claim 1, characterized in that: Detecting local signal data to obtain local detection results, including: The local signal data is converted into a local spectrum through short-time Fourier transform, and the local spectrum is extracted through the target YOLOv5 model to obtain the local detection result; Improvements to the target YOLOv5 model include: The C3 module in the YOLOv5 model is replaced with the target C3 module, and the neck structure of the YOLOv5 model is replaced with the target neck structure to obtain the target YOLOv5 model; The working principle of the target C3 module includes: Obtain an initial feature map, input the initial feature map into a CBS module to obtain a first feature map, input the first feature map into a PConv module, a CBS module, and a CBS module in sequence to obtain a second feature map, and add the first feature map and the second feature map to obtain a third feature map; The initial feature map is input into the CBS module to obtain a fourth feature map, the third feature map and the fourth feature map are added to obtain a fifth feature map, the fifth feature map is input into the CBS module to obtain a sixth feature map, and the sixth feature map is sequentially input into four SPAB modules to obtain a seventh feature map.

5. The DAS-based chemical gas leakage safety detection method according to claim 1, characterized in that: The working principle of the target neck structure includes: Obtain a feature tensor set of the backbone network, input the first feature tensor into the DBS module to obtain a fourth feature tensor, upsample the fourth tensor to obtain a target fourth tensor, and concatenate the four feature tensors, the target fourth feature tensor, and the second feature tensor to obtain a fifth feature tensor; the feature tensor set includes: the first feature tensor, the second feature tensor, and the third feature tensor; Inputting the fifth feature tensor into the target C3 module and the DBS module in sequence to obtain a sixth feature tensor, upsampling the sixth feature tensor to obtain a target sixth feature tensor, and concatenating the third feature tensor and the target sixth feature tensor to obtain a seventh feature tensor; Inputting the seventh feature tensor into the target CSPStage module to obtain an eighth feature tensor, inputting the eighth feature tensor into the CBS module to obtain a ninth feature tensor, concatenating the ninth feature tensor and the sixth feature tensor to obtain a tenth feature tensor, inputting the tenth feature tensor into the target CSPStage module to obtain a first target output, and using the eighth feature tensor as a second target output; The working process of the target CSPStage module includes: Obtain an initial target feature map, input the initial target feature map into a 1×1Conv module to obtain a first target feature map, input the first target feature map into a target Conv module to obtain a second target feature map, input the second target feature map into a Biformer module and a 3×3Conv module in sequence to obtain a third target feature map, add the first target feature map and the third target feature map to obtain a fourth target feature map, concatenate the fourth target feature map and the first target feature map to obtain a fifth target feature map, input the fifth target feature map into a 1×1Conv module to obtain a sixth target feature map, and use the sixth target feature map as the output of the target CSPStage module.

6. A chemical gas leak safety detection system based on DAS, characterized in that: The system includes: a global detection module, a local detection module and a safety warning module: The global detection module is used to obtain global signal data of the target area and perform safety detection based on the global signal data to obtain an abnormality detection result; The local detection module is used to determine a sub-region according to the abnormality detection result and obtain local signal data of the sub-region; the target region includes multiple sub-regions; The safety warning module is used to detect local signal data to obtain local detection results, and issue a safety warning based on the local detection results.

7. A DAS-based chemical gas leak safety detection system according to claim 6, characterized in that: The system also includes: a data acquisition module and a sample library construction module: The data acquisition module is used to acquire historical data, pre-process the historical data to obtain target historical data, determine the length of the sliding window, and segment the target historical data according to the sliding window to obtain multiple signal segments; the historical data is leakage sample signals under different pressure and leakage aperture conditions; The sample library construction module is used to extract each signal segment to obtain multiple upper envelopes, perform quadratic Gaussian smoothing and Lagrangian downsampling on each upper envelope to obtain multiple target upper envelopes, and construct a sample library based on each target upper envelope; the signal segments and upper envelopes correspond one to one.

8. A DAS-based chemical gas leak safety detection system according to claim 6, characterized in that: The global detection module includes: a first operation module, a second operation module, a third operation module, a fourth operation module and a fifth operation module: The first operating module is configured to obtain a plurality of intrinsic mode functions (IMFs) by performing variational mode decomposition on the global signal data, determine the frequency of each IMF, and reconstruct the IMFs with a frequency less than a preset frequency to obtain a noise reduction signal; The second operation module is configured to segment the noise reduction signal according to the sliding window to obtain a plurality of segmented signals, and perform a target operation on each segmented signal to obtain a real-time signal; the target operation includes upper envelope extraction, quadratic Gaussian smoothing, and Lagrangian downsampling operations, and the real-time signal includes x1 and x2; The third operation module is configured to match the real-time signal according to the sample library to obtain minimum cumulative distance matrices D1 and D2, calculate a total cumulative distance matrix according to D1 and D2, and determine a matching signal according to the total cumulative distance matrix; The fourth operation module is configured to determine a step size of the sliding window, repeatedly execute the second operation module and the third operation module, and update the real-time signal and the matching signal to obtain a signal sequence; the signal sequence includes a plurality of matching signals; The fifth operation module is configured to set a leakage warning threshold, and if all matching signals in the signal sequence are smaller than the leakage warning threshold, add an abnormality label to the global signal data to obtain an abnormality detection result.

9. A DAS-based chemical gas leak safety detection system according to claim 6, characterized in that: The local detection module is further configured to convert the local signal data into a local spectrum graph through a short-time Fourier transform, and perform feature extraction on the local spectrum graph through a target YOLOv5 model to obtain a local detection result; Improvements to the target YOLOv5 model include: The C3 module in the YOLOv5 model is replaced with the target C3 module, and the neck structure of the YOLOv5 model is replaced with the target neck structure to obtain the target YOLOv5 model; The working principle of the target C3 module includes: Obtain an initial feature map, input the initial feature map into a CBS module to obtain a first feature map, input the first feature map into a PConv module, a CBS module, and a CBS module in sequence to obtain a second feature map, and add the first feature map and the second feature map to obtain a third feature map; The initial feature map is input into the CBS module to obtain a fourth feature map, the third feature map and the fourth feature map are added to obtain a fifth feature map, the fifth feature map is input into the CBS module to obtain a sixth feature map, and the sixth feature map is sequentially input into four SPAB modules to obtain a seventh feature map.

10. A DAS-based chemical gas leakage safety detection system according to claim 6, characterized in that: The working principle of the target neck structure includes: Obtain a feature tensor set of the backbone network, input the first feature tensor into the DBS module to obtain a fourth feature tensor, upsample the fourth tensor to obtain a target fourth tensor, and concatenate the four feature tensors, the target fourth feature tensor, and the second feature tensor to obtain a fifth feature tensor; the feature tensor set includes: the first feature tensor, the second feature tensor, and the third feature tensor; Inputting the fifth feature tensor into the target C3 module and the DBS module in sequence to obtain a sixth feature tensor, upsampling the sixth feature tensor to obtain a target sixth feature tensor, and concatenating the third feature tensor and the target sixth feature tensor to obtain a seventh feature tensor; Inputting the seventh feature tensor into the target CSPStage module to obtain an eighth feature tensor, inputting the eighth feature tensor into the CBS module to obtain a ninth feature tensor, concatenating the ninth feature tensor and the sixth feature tensor to obtain a tenth feature tensor, inputting the tenth feature tensor into the target CSPStage module to obtain a first target output, and using the eighth feature tensor as a second target output; The working process of the target CSPStage module includes: Obtain an initial target feature map, input the initial target feature map into a 1×1Conv module to obtain a first target feature map, input the first target feature map into a target Conv module to obtain a second target feature map, input the second target feature map into a Biformer module and a 3×3Conv module in sequence to obtain a third target feature map, add the first target feature map and the third target feature map to obtain a fourth target feature map, concatenate the fourth target feature map and the first target feature map to obtain a fifth target feature map, input the fifth target feature map into a 1×1Conv module to obtain a sixth target feature map, and use the sixth target feature map as the output of the target CSPStage module.

Images