A coal gas pipeline micro-leakage point cloud positioning method based on a voiceprint space spectrum

Abstract

The present application relates to the technical field of industrial pipeline safety detection, and particularly relates to a coal gas pipeline micro-leakage point cloud positioning method based on a voiceprint spatial spectrum. The present application uses an unmanned inspection vehicle or unmanned aerial vehicle carrying a microphone array, a reference microphone and a laser radar to patrol along a pipeline; simultaneously collects environmental audio signals and pipeline three-dimensional point cloud data. LMS adaptive filtering is combined with a reference channel noise cancellation strategy to suppress strong background noise in a factory area, and extract suspected leakage voiceprints; a frequency domain delay sum beam forming algorithm is used to calculate the spatial spectrum of a sound source and generate a three-dimensional sound pressure thermal map. A rigid body transformation matrix is used to complete the registration of the acoustic thermal map and the laser point cloud coordinates, the sound source area is segmented according to the sound pressure threshold, the regional centroid is used as the leakage point coordinate output, and the leakage position is labeled on a three-dimensional platform and the voiceprint data is associated. The present application effectively realizes the three-dimensional high-precision positioning of a leakage point in a noisy industrial environment, and is suitable for large-scale coal gas pipeline network automatic inspection in steel enterprises.

Description

Technical Field

[0001] This invention relates to the field of industrial pipeline safety inspection technology, and in particular to a method for locating micro-leakage points in gas pipelines based on acoustic spatial spectrum. Background Technology

[0002] As an important secondary energy source for steel enterprises, coal gas has a large-scale and complex pipeline network. At present, the detection of coal gas pipeline leaks at home and abroad mainly relies on manual inspection and fixed gas detectors, which has the following technical defects: (1) Manual inspection is inefficient and has many blind spots, making it difficult to detect micro-leakages in high-altitude and hidden parts; (2) Fixed sensors have limited coverage and cannot achieve large-scale mobile detection; (3) Existing acoustic imaging technology is easily submerged in the strong background noise environment of steel plants (blast furnace noise, vehicle noise, mechanical vibration, etc.), and the reliability of detection is greatly reduced; (4) The location of leak points depends on the experience judgment of operators and lacks accurate mapping methods with three-dimensional spatial models. Summary of the Invention

[0003] This invention aims to solve the technical problems of low reliability and poor positioning accuracy of gas pipeline micro-leakage detection in strong background noise environment in the prior art, and provides a gas pipeline micro-leakage point cloud positioning method based on acoustic spatial spectrum.

[0004] Therefore, the present invention provides the following technical solution: A method for locating micro-leakage points in gas pipelines based on acoustic signature spatial spectrum includes the following steps: S1. Control the mobile inspection platform equipped with a microphone array and lidar to move along the preset inspection path; S2. During the movement of the mobile inspection platform, the environmental audio signal A(t) and the three-dimensional laser point cloud data P(x, y, z) of the gas pipeline are collected simultaneously through the microphone array and lidar. S3. Using a pre-stored gas leak soundprint feature library, extract the suspected leak soundprint S(t) from the environmental audio signal A(t) using an adaptive filtering algorithm. S4. Calculate the spatial spectrum of the suspected leakage acoustic waveform S(t) based on the beamforming algorithm, and generate a three-dimensional thermal map of the sound pressure level H(x, y, z). S5. Perform spatial coordinate registration between the three-dimensional heat map H(x, y, z) of sound pressure and the three-dimensional laser point cloud data P(x, y, z); S6. In the registered point cloud model, the sound source region is segmented according to the sound pressure level threshold, and the continuous point cloud region where the sound pressure level exceeds the preset alarm threshold is identified. The centroid coordinates of the region are output as the location of the leakage point. S7. Render and annotate the leak location in a 3D visualization platform and associate it with the leak acoustic signature data.

[0005] Furthermore, in step S1, the mobile inspection platform is an unmanned inspection vehicle or a drone.

[0006] Furthermore, in step S2, during the synchronous acquisition of environmental audio signals and three-dimensional laser point cloud data of the gas pipeline along the inspection path, a unified reference timestamp is added to the acquired environmental audio signals and three-dimensional laser point cloud data through a timestamp alignment module, establishing a one-to-one time-series mapping relationship between the environmental audio signals and the three-dimensional laser point cloud data, thereby achieving millisecond-level time synchronization between the environmental audio signals and the three-dimensional laser point cloud data.

[0007] Furthermore, the mobile inspection platform is also equipped with a reference microphone. In step S3, the adaptive filtering algorithm uses the industrial background noise collected by the reference microphone as the input of the filtering reference signal. By iteratively fitting the interference noise waveform through the least mean square algorithm, the industrial environmental noise component in the environmental audio signal collected by the microphone array is canceled to obtain the suspected leakage sound pattern.

[0008] Furthermore, the step S3, which involves obtaining suspected leaking voiceprints based on an adaptive filtering algorithm, includes the following steps: S31. Acquire mixed acoustic signals from the site using a microphone array. These mixed acoustic signals include gas leak acoustic signals and industrial environmental background noise. The formula for calculating the mixed signal is: d(n) = s(n) + v(n), where s(n) is the acoustic signal of gas leakage and v(n) is the background noise; S32. Collect background noise from the industrial site using a reference microphone and generate a filtered reference signal x(n); S33. Construct a transverse adaptive filter and use the least mean square (LMS) algorithm to iteratively update the weight coefficients w(n) of the transverse adaptive filter so that the output signal y(n) of the transverse adaptive filter approximates the background noise v(n). S34. Calculate the filter error signal using the following formula: e(n) = d(n) - y(n), where the error signal is an estimate of the pure gas leakage acoustic signal after removing industrial environmental noise. (n), which serves as the basis for subsequent extraction of suspected leaked acoustic signatures S(t).

[0009] Furthermore, the beamforming algorithm in step S4 employs a delay-summing beamformer, which calculates the energy distribution of the sound source in each direction by scanning the pointing angle of the microphone array. The beamforming algorithm is a frequency domain beamforming algorithm, which first converts the suspected leakage sound pattern in the time domain to the frequency domain, calculates the corresponding spatial spectrum independently at each frequency point, and then performs energy integration on the spatial spectrum of all frequency points to solve the spatial energy distribution of the broadband leakage sound source, and finally generates a three-dimensional heat map H(x, y, z) of the sound pressure level.

[0010] Furthermore, the coordinate registration in step S5 is based on the real-time pose data of the mobile platform, and the rigid body transformation matrix T between the acoustic center of the microphone array and the acquisition center of the lidar is solved in advance through a joint calibration method. acoustic→lidar Based on the transformation matrix, a coordinate transformation relationship from the acoustic coordinate system to the lidar coordinate system is constructed.

[0011] Furthermore, the gas leak acoustic signature database includes leak acoustic signature samples under different pressures and different leak orifice diameters, as well as leak acoustic signature spatial spectra under different operating conditions and background noise.

[0012] Advantages and positive effects of the present invention: This invention incorporates a reference microphone to collect native background noise from the industrial environment. Combined with the LMS (Least Mean Square) adaptive filtering algorithm, it accurately fits and cancels out complex industrial interference noises such as blast furnace roar, mechanical vibration, and vehicle traffic, effectively separating weak acoustic signals from gas leaks. This solves the technical problems of traditional acoustic detection methods where micro-leakage acoustic patterns are easily submerged in high-noise industrial environments, leading to detection failures and high false positive / false negative rates. It can stably identify minute pipeline leaks even under 85dB strong background noise conditions, significantly improving the signal-to-noise ratio and environmental adaptability of leak detection.

[0013] This invention employs a frequency-domain delay-summing beamforming algorithm to construct a three-dimensional acoustic pressure thermal map of the leakage sound source. Through a joint calibration transformation matrix of a microphone array and lidar, precise registration of acoustic spatial coordinates with the three-dimensional point cloud coordinates of the pipeline is achieved, overcoming the limitations of traditional acoustic imaging, which can only output two-dimensional azimuth angles and lacks spatial depth information. By using sound pressure threshold segmentation and region centroid solving, the leak location is accurately pinpointed, enabling centimeter-level three-dimensional localization of micro-leakage in pipelines, significantly improving positioning accuracy.

[0014] This invention uses a unified timestamp alignment mechanism to time-bind audio voiceprint data and laser point cloud data, enabling millisecond-level synchronous acquisition and one-to-one mapping of the two types of heterogeneous sensor data. This avoids the problems of sound image misalignment and positioning offset caused by equipment acquisition timing deviation during mobile inspection, and provides high-precision and highly consistent data support for subsequent steps.

[0015] This invention utilizes mobile platforms such as drones and unmanned inspection vehicles equipped with integrated sensing systems to achieve automated continuous inspection of long-distance, large-scale gas pipeline networks. It overcomes the drawbacks of traditional manual inspections, including low efficiency, numerous blind spots in high-altitude and concealed pipelines, and high operational risks. Compared to traditional manual inspection methods, overall inspection efficiency is increased by more than 5 times, and it is adaptable to various scenarios such as complex plant pipeline networks, high-altitude pipelines, and long-distance ground pipelines.

[0016] This invention pre-constructs a database of acoustic signatures and spatial spectrum features of gas leaks covering different leak apertures, different pipeline pressures, and different industrial background conditions. It can adaptively match and identify models for different leak levels and different field environments, effectively improving the versatility and stability of micro-leak identification and solving the problem that traditional detection methods are insufficient in identifying small aperture and low-pressure weak leaks.

[0017] This invention can render and accurately mark the leak location in real time on a 3D visualization platform, and simultaneously associate the corresponding original feature data such as leak sound patterns, sound pressure spatial spectrum, and acquisition conditions. This enables the visualization of the leak location and the traceability of detection data, and can provide accurate 3D data support for pipeline hidden danger rectification, risk assessment, pipeline network operation and maintenance, and fault review, facilitating the intelligent management of gas pipeline leakage risks throughout their entire life cycle. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 The present invention provides an overall flowchart of a method for locating micro-leaks in gas pipelines based on acoustic spatial spectrum.

[0020] Figure 2 The diagram illustrates the principle of the adaptive filtering algorithm for a gas pipeline micro-leakage point cloud localization method based on acoustic spatial spectrum provided by this invention.

[0021] Figure 3 This is a schematic diagram of coordinate system transformation for acoustic-point cloud coordinate registration in a point cloud localization method for micro-leakage in gas pipelines based on acoustic spatial spectrum, provided by the present invention.

[0022] Figure 4 This invention provides the original three-dimensional point cloud model for a point cloud localization method for micro-leakage in gas pipelines based on acoustic spatial spectrum.

[0023] Figure 5This invention provides a three-dimensional point cloud model after fusing a heat map for a method of locating micro-leaks in gas pipelines based on acoustic spatial spectrum. Detailed Implementation

[0024] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0025] This invention provides a method for locating micro-leakage points in gas pipelines based on acoustic signature spatial spectrum, such as... Figure 1 As shown, it includes the following steps: S1. Control the mobile inspection platform equipped with a microphone array and lidar to move along the preset inspection path; the mobile inspection platform is an unmanned inspection vehicle or a drone.

[0026] S2. During the movement of the mobile inspection platform, the environmental audio signal A(t) and the three-dimensional laser point cloud data P(x, y, z) of the gas pipeline are simultaneously collected through a microphone array and a lidar. During the simultaneous collection of the environmental audio signal and the three-dimensional laser point cloud data of the gas pipeline, a unified reference timestamp is added to the collected environmental audio signal and the three-dimensional laser point cloud data through a timestamp alignment module, establishing a one-to-one time-series mapping relationship between the environmental audio signal and the three-dimensional laser point cloud data, and realizing millisecond-level time synchronization between the environmental audio signal and the three-dimensional laser point cloud data.

[0027] S3. Utilize a pre-stored database of gas leak acoustic signatures, such as... Figure 2 As shown, the suspected leakage soundprint S(t) is extracted from the environmental audio signal A(t) by an adaptive filtering algorithm. The mobile inspection platform is also equipped with a reference microphone. The adaptive filtering algorithm takes the industrial background noise collected by the reference microphone as the filter reference signal input, and iteratively fits the interference noise waveform by the least mean square algorithm to cancel the industrial environmental noise component in the environmental audio signal collected by the microphone array, thus obtaining the suspected leakage soundprint.

[0028] The steps involved in obtaining suspected leaking acoustic fingerprints based on an adaptive filtering algorithm are as follows: S31. Acquire mixed acoustic signals from the site using a microphone array. These mixed acoustic signals include gas leak acoustic signals and industrial environmental background noise. The formula for calculating the mixed signal is: d(n) = s(n) + v(n), where s(n) is the acoustic signal of gas leakage and v(n) is the background noise.

[0029] S32. Collect background noise from the industrial site using a reference microphone and generate a filtered reference signal x(n).

[0030] S33. Construct a transverse adaptive filter and use the least mean square (LMS) algorithm to iteratively update the weight coefficients w(n) of the transverse adaptive filter so that the output signal y(n) of the transverse adaptive filter approximates the background noise v(n).

[0031] S34. Calculate the filter error signal using the following formula: e(n) = d(n) - y(n), where the error signal is an estimate of the pure gas leakage acoustic signal after removing industrial environmental noise. (n) serves as the basis for extracting suspected leak sound patterns S(t) in the future. It is then combined with the pre-built gas leak sound pattern feature library for matching and identification to obtain suspected leak sound patterns S(t).

[0032] The gas leak acoustic signature database includes leak acoustic signature samples under different pressures and leak orifice diameters, as well as leak acoustic signature spatial spectra under different operating conditions and background noise.

[0033] S4. Calculate the spatial spectrum of the suspected leakage sound pattern S(t) based on the beamforming algorithm to generate a three-dimensional heat map H(x, y, z) of the sound pressure level. The beamforming algorithm uses a delay-summing beamformer to calculate the energy distribution of the sound source in each direction by scanning the pointing angle of the microphone array. The beamforming algorithm is a frequency domain beamforming algorithm. First, the suspected leakage sound pattern in the time domain is converted to the frequency domain. The corresponding spatial spectrum is calculated independently at each frequency point. Then, energy integration is performed on the spatial spectrum of all frequency points to solve the spatial energy distribution of the broadband leakage sound source. Finally, a three-dimensional heat map H(x, y, z) of the sound pressure level is generated.

[0034] S5. Perform spatial coordinate registration between the three-dimensional sound pressure thermal map H(x, y, z) and the three-dimensional laser point cloud data P(x, y, z); such as Figure 3 As shown, coordinate registration is based on real-time six-DOF pose data of the mobile platform. The rigid body transformation matrix T between the acoustic center of the microphone array and the acquisition center of the lidar is solved in advance through joint calibration. acoustic→lidar Based on the transformation matrix, a coordinate transformation relationship from the acoustic coordinate system to the lidar coordinate system is constructed.

[0035] like Figure 4-5 As shown in Figure S6, in the registered point cloud model, the sound source region is segmented according to the sound pressure level threshold, and continuous point cloud regions where the sound pressure level exceeds the preset alarm threshold are identified. The centroid coordinates of the region are output as the location of the leakage point.

[0036] S7. Render and annotate the leak location in a 3D visualization platform and associate it with the leak acoustic signature data.

[0037] Example This embodiment uses a 10km gas pipeline as an application scenario to illustrate the specific implementation process of the present invention.

[0038] Step 1: System Configuration An explosion-proof unmanned inspection vehicle is configured as a mobile platform, which integrates: (1) a 32-channel spiral microphone array with a sampling rate of 80kHz and a frequency response range of 5Hz~40kHz; (2) a 32-line lidar with a ranging accuracy of ±2cm and a scanning range of 100m; (3) a high-precision inertial navigation unit (IMU) to provide real-time pose data; and (4) an edge computing unit equipped with an NVIDIA Jetson AGX Orin processor.

[0039] Step 2: Constructing the Voiceprint Feature Database In the experimental environment, leakage simulation devices with different orifice diameters (0.5mm, 1mm, 2mm, 5mm) were used to collect 1200 sets of leakage acoustic samples within a pressure range of 0.2MPa to 1.0MPa. At the same time, 500 sets of background noise samples under typical operating conditions in the plant area were collected, including: blast furnace operation noise, vehicle traffic noise, pipeline fluid noise, etc.

[0040] Step 3: Adaptive Filter Parameter Settings The LMS filter order is set to 128, the step size factor μ = 0.01, and the convergence time is < 0.5 seconds.

[0041] Step 4: Inspection Implementation The unmanned inspection vehicle travels along a preset path at a speed of 5 km / h, continuously collecting data during the journey. When the system detects a suspicious leak signal, it automatically reduces its speed to 2 km / h for precise location.

[0042] Step 5: Output Results Taking a specific detection as an example, the system detected a suspected leak signal approximately 15 meters from the leak point. After adaptive filtering, the signal-to-noise ratio was improved from the original 3dB to 18dB. The beamforming localization results were: azimuth 37°, elevation -5°. After point cloud registration, as shown... Figure 5 As shown, the coordinates of the leak point are (X=25436.5, Y=89372.3, Z=12.6), which deviates from the actual leak location verified by manual inspection by 0.23 meters, meeting the requirement of ≤0.5 meters.

[0043] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for locating micro-leakage points in gas pipelines based on acoustic signature spatial spectrum, characterized in that, Includes the following steps: S1. Control the mobile inspection platform equipped with a microphone array and lidar to move along the preset inspection path; S2. During the movement of the mobile inspection platform, the environmental audio signal A(t) and the three-dimensional laser point cloud data P(x, y, z) of the gas pipeline are collected simultaneously through the microphone array and lidar. S3. Using a pre-stored gas leak soundprint feature library, extract the suspected leak soundprint S(t) from the environmental audio signal A(t) using an adaptive filtering algorithm. S4. Calculate the spatial spectrum of the suspected leakage acoustic waveform S(t) based on the beamforming algorithm, and generate a three-dimensional thermal map of the sound pressure level H(x, y, z). S5. Perform spatial coordinate registration between the three-dimensional heat map H(x, y, z) of sound pressure and the three-dimensional laser point cloud data P(x, y, z); S6. In the registered point cloud model, the sound source region is segmented according to the sound pressure level threshold, and the continuous point cloud region where the sound pressure level exceeds the preset alarm threshold is identified. The centroid coordinates of the region are output as the location of the leakage point. S7. Render and annotate the leak location in a 3D visualization platform, and associate it with the leak acoustic signature data; In step S2, during the synchronous acquisition of environmental audio signals and three-dimensional laser point cloud data of gas pipelines along the inspection path, a unified reference timestamp is added to the acquired environmental audio signals and three-dimensional laser point cloud data through a timestamp alignment module. This establishes a one-to-one time-series mapping relationship between the environmental audio signals and the three-dimensional laser point cloud data, achieving millisecond-level time synchronization between the environmental audio signals and the three-dimensional laser point cloud data. The mobile inspection platform is also equipped with a reference microphone. The adaptive filtering algorithm in step S3 takes the industrial background noise collected by the reference microphone as the filter reference signal input, and uses the least mean square algorithm to iteratively fit the interference noise waveform to cancel the industrial environmental noise component in the environmental audio signal collected by the microphone array, thereby obtaining the suspected leakage soundprint. The beamforming algorithm in step S4 uses a delay summation beamformer to calculate the energy distribution of the sound source in each direction by scanning the pointing angle of the microphone array. The beamforming algorithm is a frequency domain beamforming algorithm. First, the suspected leakage sound pattern in the time domain is converted to the frequency domain. The corresponding spatial spectrum is calculated independently at each frequency point. Then, energy integration is performed on the spatial spectrum of all frequency points to solve the spatial energy distribution of the broadband leakage sound source. Finally, a three-dimensional heat map H(x, y, z) of the sound pressure level is generated. The coordinate registration in step S5 is based on the real-time pose data of the mobile platform, and the rigid body transformation matrix T between the acoustic center of the microphone array and the acquisition center of the lidar is solved in advance through joint calibration. acoustic→lidar Based on the transformation matrix, a coordinate transformation relationship from the acoustic coordinate system to the lidar coordinate system is constructed.

2. The method for locating micro-leakage points in gas pipelines based on acoustic spatial spectrum according to claim 1, characterized in that, In step S1, the mobile inspection platform is an unmanned inspection vehicle or a drone.

3. The method for locating micro-leakage points in gas pipelines based on acoustic spatial spectrum according to claim 1, characterized in that, The step S3, which involves obtaining suspected leaking acoustic signatures based on an adaptive filtering algorithm, includes the following steps: S31. Acquire mixed acoustic signals from the site using a microphone array. These mixed acoustic signals include gas leak acoustic signals and industrial environmental background noise. The formula for calculating the mixed signal is: d(n) = s(n) + v(n), where s(n) is the acoustic signal of gas leakage and v(n) is the background noise; S32. Collect background noise from the industrial site using a reference microphone and generate a filtered reference signal x(n); S33. Construct a transverse adaptive filter and use the least mean square (LMS) algorithm to iteratively update the weight coefficients w(n) of the transverse adaptive filter so that the output signal y(n) of the transverse adaptive filter approximates the background noise v(n). S34. Calculate the filter error signal using the following formula: e(n) = d(n) - y(n), where the error signal is an estimate of the pure gas leakage acoustic signal after removing industrial environmental noise. (n), which serves as the basis for subsequent extraction of suspected leaked acoustic signatures S(t).

4. The method for locating micro-leakage points in gas pipelines based on acoustic spatial spectrum according to claim 1, characterized in that, The gas leak acoustic signature database includes leak acoustic signature samples under different pressures and different leak orifice diameters, as well as leak acoustic signature spatial spectra under different operating conditions and background noise.

Images