An underground pipeline depth ultrasonic detection method based on multi-channel echo cooperative fusion
By employing a multi-channel echo fusion ultrasonic testing method, the accuracy and robustness issues of underground pipeline burial depth detection have been resolved, achieving high-precision, low-cost burial depth measurement, which is suitable for rapid detection of urban underground pipe networks.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-21
- Publication Date
- 2026-07-10
AI Technical Summary
Existing underground pipeline burial depth detection technologies suffer from low accuracy, poor robustness, high hardware dependence, and poor adaptability to soil conditions. In particular, they are difficult to achieve accurate measurements with centimeter-level precision in old urban areas and complex geological conditions.
A multi-channel echo collaborative fusion method is adopted. By deploying an array of ultrasonic transducers on the ground, cascaded sliding mean noise reduction, Hilbert transform envelope extraction and adaptive noise threshold detection are used to identify candidate pipeline reflection peaks. Combined with neighborhood clustering and signal-to-noise ratio weighted fusion, multi-channel redundant information is obtained to resist soil scattering and single-channel bias, so as to realize burial depth measurement.
It achieves a burial depth measurement accuracy of 1cm under different soil conditions, reduces hardware costs, is highly adaptable, and is suitable for rapid and accurate detection in municipal engineering projects.
Smart Images
Figure CN122362392A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the technical field of underground pipeline detection and non-contact measurement, and in particular to an ultrasonic detection method for underground pipeline burial depth. It is geared towards the ultrasonic burial depth detection signal processing of underground pipe networks such as urban water supply, gas transmission and distribution, and centralized heating, and can be applied to scenarios such as pipeline survey before municipal engineering construction, pipeline operation and maintenance inspection, and pipeline location after geological disasters. Background Technology
[0002] Urban underground spaces are densely packed with various pipelines, including water supply, drainage, gas, heating, and power cables, forming an indispensable lifeline network for urban operation. With the continuous advancement of urbanization and the increasing intensity of underground space development, accidents involving accidental damage to existing pipelines during construction excavation are frequent. These accidents can range from minor water and gas outages to serious gas leaks or even explosions. One root cause of these frequent accidents is the inaccuracy of the actual burial depth information of the pipelines—the design values recorded in as-built drawings often deviate significantly from the actual depth after subsequent settlement and backfilling. Therefore, how to quickly and accurately obtain the true burial depth of pipelines without excavating the surface has become an urgent practical problem that needs to be solved in the field of municipal engineering.
[0003] The commonly used pipeline detection methods in current engineering projects mainly fall into the following categories. The first category is electromagnetic induction pipeline detectors. These work by applying an alternating electromagnetic field to the metal pipeline or utilizing the pipeline's own power frequency current to receive and measure the magnetic field gradient at the ground surface to infer the pipeline's location and burial depth. This method is only applicable to metallic conductor pipelines and is ineffective for non-metallic pipes such as PE and HDPE pipes. Furthermore, in older urban areas with densely intersecting pipelines, electromagnetic coupling between adjacent pipelines can severely affect the accuracy of depth readings. The second category is ground-penetrating radar (GPR) technology. This method emits broadband electromagnetic pulses into the ground and infers pipeline location by analyzing the delay of the reflected waves. Theoretically, this method is compatible with pipelines of various materials, but its measurement accuracy is extremely sensitive to soil moisture content, clay content, and underground metallic debris. In high-water-content clay layers, electromagnetic wave attenuation is drastic, significantly reducing the effective detection depth. The equipment is also expensive, and operation and interpretation require extensive experience. The third category includes acoustic and vibration methods, such as correlators and listening rods commonly used in pipeline leak detection. These methods rely on pipe vibration signals rather than the propagation of sound waves in the soil, and therefore cannot directly provide the precise distance between the pipeline and the ground. Furthermore, while manual excavation of test pits yields reliable results, the cost of road construction is high, and the recovery period is long, making it almost unacceptable in busy urban areas.
[0004] Ultrasonic testing technology, with its non-invasive nature and ability to penetrate various media, has been successfully applied in industrial fields. However, extending it to the field of underground pipeline depth measurement faces several unique challenges: First, soil is not a homogeneous isotropic medium; the sound velocity varies greatly with factors such as moisture content, compaction degree, and soil composition, resulting in significant errors in distance conversion for single measurements. Second, the surface layer contains irregular scattering bodies such as gravel, cavities, and layer interfaces, which easily generate a large number of false echoes, interfering with the reflected signals truly originating from the pipeline's outer wall. Third, unlike reflection testing of metal pipe walls, ultrasonic attenuation in soil is much higher than in steel and aluminum media, resulting in a significantly lower signal-to-noise ratio. When using the pulse-echo method with only a single transducer for depth measurement, the superposition of these interference factors easily leads to large fluctuations in depth estimation, making it difficult to meet the engineering requirements for centimeter-level accuracy.
[0005] In existing literature, some scholars have attempted to improve ultrasonic ranging performance using signal enhancement or intelligent recognition algorithms. For example, wavelet decomposition is used to improve the signal-to-noise ratio of weak echoes, or convolutional neural networks are used to automatically distinguish between echoes and noise. These approaches each have their advantages, but also limitations: the basis functions and decomposition levels of wavelet denoising are highly dependent on prior knowledge, requiring repeated parameter tuning under different soil conditions; deep learning methods require a large number of labeled soil echo training sets, but it is difficult to obtain sufficient depth-labeled sample data in actual underground pipeline scenarios, and the performance of trained models degrades significantly when transferred to new working conditions. Furthermore, none of the above approaches utilize multi-channel redundancy information to suppress random biases in single measurements; when a channel happens to be strongly interfered with by a local scatterer, the overall results may produce unacceptable errors.
[0006] In summary, the current field of trenchless inspection of underground pipeline burial depth lacks a method that balances high precision, robustness, low hardware dependence, and adaptability to soil conditions. There is an urgent need to develop ultrasonic inspection technology capable of simultaneously acquiring multiple observations under a single ground deployment and eliminating random interference and systematic biases through signal-level collaborative fusion. Summary of the Invention
[0007] To address the low accuracy of existing ultrasonic testing methods for underground pipeline burial depth, this invention proposes a multi-channel echo collaborative fusion-based ultrasonic testing method for underground pipeline burial depth. The core idea of this method is as follows: Several ultrasonic transducers are deployed in an array on the ground. Each element emits ultrasonic pulses into the ground and collects reflected echoes from the outer wall of the pipeline. The echo sequences of each channel are subjected to a signal processing chain consisting of cascaded moving average noise reduction, Hilbert transform envelope extraction, and adaptive noise threshold detection to identify the candidate pipeline echo peak values for each channel. Then, based on the transit time, the peak values are converted into independent initial burial depth values for each channel, and each initial value is assigned a quality weight represented by the echo signal-to-noise ratio. Finally, the initial values of all channels are fed into a neighborhood clustering process constrained by depth tolerance to screen out the burial depth estimates with the highest consistency. These estimates are then weighted and fused according to the signal-to-noise ratio to obtain the final pipeline burial depth. This multi-channel redundant cross-calibration effectively resists soil scattering and single-channel random deviations, achieving a measurement accuracy of 1 cm.
[0008] To achieve the above objectives, the technical solution of the present invention is implemented as follows:
[0009] An ultrasonic testing method for underground pipeline burial depth based on multi-channel echo collaborative fusion, comprising the following steps:
[0010] S1: Multiple sets of ultrasonic transducers are deployed on the ground in a predetermined spatial configuration in the area to be tested. Each set of ultrasonic transducers emits ultrasonic pulses into the ground and receives reflected echo signals from the outer wall of the underground pipeline. The reflected echo signals are conditioned and converted from analog to digital at the front end to obtain a multi-channel digital echo sequence.
[0011] S2: For each group of digital echo sequences, cascaded sliding mean denoising, Hilbert transform envelope extraction, and adaptive threshold detection based on noise-based low-level statistics are performed sequentially to identify and retain candidate pipe reflection peak points;
[0012] S3: Based on the transit time and medium sound velocity corresponding to the candidate reflection peak points of each channel, the initial burial depth estimate of each channel is calculated by time-depth conversion, and the echo signal-to-noise ratio corresponding to each candidate reflection peak point is used as a quality index.
[0013] S4: Collect the initial burial depth estimates of all channels into a unified dataset, perform neighborhood clustering with the preset depth tolerance as a constraint, select the cluster with the most members as the effective depth cluster, and after the validity is determined, perform a weighted average with the echo signal-to-noise ratio of each member in the effective depth cluster as the weight to obtain the final burial depth of the pipeline.
[0014] Furthermore, cascaded moving average noise reduction is applied sequentially to each group of digitized echo sequences, including:
[0015] First-level moving average filtering: For the first... raw echo sequence of the channel With window length For coarse noise reduction, the formula is:
[0016] ;
[0017] in, Channel number, For sampling point index, This is the output signal of the first-stage moving average filter;
[0018] Second-stage moving average filtering: This applies the output signal from the first-stage moving average filtering to... With window length For fine noise reduction, the formula is:
[0019] ;
[0020] in, This is the output signal of the second-stage moving average filter. .
[0021] Furthermore, the Hilbert transform envelope is performed, including:
[0022] Hilbert transform: for the output signal of the second-stage moving average filter Performing the Hilbert transform yields the orthogonal components:
[0023] Analytical signal construction: The output signal of the second-stage moving average filter Orthogonal components Combined into analytical signals:
[0024] ;
[0025] in, The imaginary unit;
[0026] Instantaneous envelope calculation: taking the analytic signal The modulus value is used as the instantaneous envelope. .
[0027] Furthermore, the adaptive threshold calculation method based on noise-level statistics includes:
[0028] Noise statistics calculation: Take the first part of the echo sequence... The instantaneous envelope data of each sampling point interval is used as a non-reflection interval. The noise statistics of the instantaneous envelope data in the non-reflection interval are calculated, including the noise mean. and standard deviation ;
[0029] Adaptive threshold construction: Constructing detection thresholds using noise statistics The formula is:
[0030] ;
[0031] in, This is the threshold coefficient.
[0032] Furthermore, candidate pipe reflection peak points are identified and preserved, including:
[0033] In the echo sequence Search for all local maxima within the region, sort them by magnitude from largest to smallest, and select the local maximum with the largest magnitude as the first local maximum. Record the time index of the candidate channel reflection peak. .
[0034] Furthermore, based on the transit time and medium sound velocity corresponding to the candidate reflection peak points of each channel, the initial burial depth estimate for each channel is calculated through time-depth transformation, including:
[0035] ;
[0036] in, For the first The initial estimated burial depth of the channel, For the speed of ultrasonic propagation, The sampling frequency.
[0037] Furthermore, the formula for calculating the echo signal-to-noise ratio corresponding to each candidate reflection peak point is as follows:
[0038] ;
[0039] in, For the first The signal-to-noise ratio of the channel echo. For the first The instantaneous envelope at the candidate peak point of the channel. For the first The average noise level of the channel.
[0040] Furthermore, the initial burial depth estimates of all channels are aggregated into a unified dataset. Neighborhood clustering is performed with a preset depth tolerance as a constraint, and the cluster with the largest number of members is selected as the effective depth cluster, including:
[0041] S41. Constructing the dataset ,in For the first The initial estimated burial depth of the channel;
[0042] S42. For each initial burial depth estimate in dataset D Statistics satisfy constraints Number of neighboring members , =3cm is the preset depth tolerance;
[0043] S43. Select the number of neighbor members Maximum initial burial depth estimate As cluster centers, the initial burial depth estimate will be included. All constraints, including Initial burial depth estimate Included in effective depth cluster .
[0044] Furthermore, the determination of validity includes:
[0045] If effective depth cluster The measurement is considered valid if the number of members is greater than or equal to half of the total number of channels; if the effective depth cluster... If the number of members is less than half of the total number of channels, most channels will not detect valid echoes, and an invalid flag will be output, prompting you to redeploy the probes or check the coupling status.
[0046] Furthermore, a weighted average is calculated using the echo signal-to-noise ratio of each member within the effective depth cluster as the weight, including:
[0047] ;
[0048] in, The final burial depth of the pipeline.
[0049] The beneficial effects of this invention are as follows: A multi-element ultrasonic array replaces a single probe, generating multiple sets of independent observation data, thus solving the problems of single-channel susceptibility to local interference and lack of verification; a two-stage sliding filter replaces traditional single-stage filtering / wavelet denoising, reducing noise while preserving peak positions, eliminating the need for manual parameter tuning; detection thresholds are automatically generated using the channel's own noise data, adapting to different soil moisture contents / soil types and improving the method's adaptability; through a strategy of independent estimation followed by collaborative fusion of multiple channels, coupled with a signal-to-noise ratio weighted average output mechanism, the invention effectively mitigates accidental measurement deviations caused by local scatterers or soil inhomogeneity in single channels. Experimental verification shows that the pipeline burial depth measurement error does not exceed 1 cm under various working conditions. This equipment is low-cost and easy to deploy. The required hardware consists only of several commercially available ultrasonic transducers and general data acquisition modules. The software algorithm has a simple structure and low computational load, allowing it to run in real-time on embedded platforms. It does not require the high hardware requirements of ground-penetrating radar and deep learning, nor does it require road excavation, making it suitable for large-scale application in municipal projects. Attached Figure Description
[0050] 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 only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0051] Figure 1 This is the overall flowchart of the present invention.
[0052] Figure 2 This is a flowchart of the echo signal processing of the present invention.
[0053] Figure 3 This is a schematic diagram of the pipeline burial depth testing platform of the present invention.
[0054] Figure 4 This is a box plot of the pipeline burial depth positioning error according to the present invention. Detailed Implementation
[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 are within the scope of protection of the present invention.
[0056] An ultrasonic detection method for underground pipeline burial depth based on multi-channel echo collaborative fusion, such as Figure 1 As shown, the steps include:
[0057] S1: Multiple sets of ultrasonic transducers are deployed on the ground in a predetermined spatial configuration in the area to be tested. Each set of ultrasonic transducers emits ultrasonic pulses into the ground and receives reflected echo signals from the outer wall of the underground pipeline. The reflected echo signals are conditioned and converted from analog to digital at the front end to obtain a multi-channel digital echo sequence.
[0058] In this embodiment, when placing the ultrasonic array probe in the area to be tested, an 8-element ultrasonic transducer is used, comprising 4 sets of ultrasonic transducer transmitters / receivers. Each ultrasonic transducer transmitter or receiver corresponds to one element. The ultrasonic transducer is a piezoelectric ceramic ultrasonic transducer with a center frequency of 50kHz, a bandwidth of 10kHz, a transmitted sound pressure level ≥110dB, and a receiving sensitivity ≥-80dB. It is adapted to the ultrasonic propagation characteristics of the soil medium and can effectively penetrate soil layers within 0.5m, meeting the needs of conventional burial depth detection for urban underground pipe networks. The ultrasonic transducers are arranged linearly and equally spaced, with a center-to-center spacing of 5cm between the elements. During deployment, the central axis of the array is perpendicular to the direction of the underground pipeline, ensuring that all elements can receive stable reflected echoes from the outer wall of the pipeline. Medical ultrasonic coupling agent is applied between the transducer radiating surface and the ground to eliminate air gaps between the probe and the ground, ensuring efficient transmission of ultrasonic energy into the strata and efficient return of echo signals to the probe, avoiding signal attenuation and noise interference caused by poor coupling. Employing a time-division single-transmit, single-receive mode, the main control module sequentially triggers four groups of ultrasonic transducers. Each group of transducers independently completes the transmit-receive process, avoiding signal crosstalk caused by simultaneous operation of multiple groups of transducers and ensuring the independence of data from each channel. Figure 3 As shown. Each array element operates in a time-division multiplexing manner: the main control module uses an STM32H7 series embedded main control chip, responsible for ultrasonic element triggering control, signal conditioning parameter configuration, ADC sampling control, and data transmission; the main control module sequentially triggers the transmitter in each ultrasonic transducer to emit ultrasonic pulses. The pulses propagate downwards into the strata and are reflected when they encounter the outer wall of an underground pipe. The reflected echoes return to the ground along the original path and are received by the receiver in the same ultrasonic transducer, forming a pulse echo mode. Each transducer transmits once, corresponding to an independent channel. The received reflected echo signals are first conditioned by a charge amplification circuit and a programmable gain circuit, and then digitized by an analog-to-digital converter, ultimately obtaining a digital echo sequence of four channels. , Provide an index for the sampling points and record the reference time index for the transmitted pulse. This serves as the time reference for calculating the burial depth of all channels.
[0059] The method of this invention is not limited to an 8-element array. It can be expanded according to the detection requirements and can use 4-element, 16-element, or 32-element arrays. The more elements there are, the higher the fusion accuracy. It is suitable for the detection of ultra-deep buried pipelines. The linear array can also be changed to a ring array to adapt to pipelines with arbitrary orientation without the need to adjust the probe direction.
[0060] S2: For each group of digitized echo sequences, cascaded moving mean denoising, Hilbert transform envelope extraction, and adaptive threshold detection based on noise-level statistics are sequentially applied to identify and retain candidate pipe reflection peak points, such as... Figure 2 As shown.
[0061] In this embodiment of the application, cascaded moving average noise reduction is performed sequentially on each group of digitized echo sequences, including:
[0062] First-level moving average filtering: For the first... raw echo sequence of the channel With window length For coarse noise reduction, the formula is:
[0063] ;
[0064] in, Channel number, For sampling point index, This is the output signal of the first-stage moving average filter.
[0065] Second-stage moving average filtering: This applies the output signal from the first-stage moving average filtering to... With window length For fine noise reduction, the formula is:
[0066] ;
[0067] in, The output signal of the second-stage moving average filter is L1, which is slightly larger than L2. This is so that after the first-stage coarse noise reduction, the second stage can perform a more refined smoothing, taking into account both noise suppression depth and time resolution.
[0068] It is known that random high-frequency noise introduced by the soil propagation path can cause dense spikes in the echo waveform, which is detrimental to subsequent peak localization. Therefore, this invention introduces a two-stage cascaded moving average filter to smooth the original sequence, preserving the temporal position of the echo peaks while reducing noise, thus avoiding filtering distortion. Compared to directly using a single-stage long-window mean filter, the two-stage cascaded filter can better preserve the rising edge and peak position information of the echo pulses while achieving similar noise reduction effects, reducing peak time drift caused by excessive smoothing.
[0069] In this embodiment of the application, performing the Hilbert transform envelope includes:
[0070] Hilbert transform: for the output signal of the second-stage moving average filter Performing the Hilbert transform yields the orthogonal components:
[0071] ;
[0072] in, This is the Hilbert transform operator.
[0073] Analytical signal construction: The output signal of the second-stage moving average filter Orthogonal components Combined into analytical signals:
[0074] ;
[0075] in, It is the imaginary unit.
[0076] Instantaneous envelope calculation: taking the analytic signal The modulus value is used as the instantaneous envelope. .
[0077] It is known that the denoised signal is still a time-domain waveform containing carrier oscillations. Its alternating positive and negative values are not convenient for direct amplitude comparison. Therefore, this invention extracts the instantaneous envelope through Hilbert transform, converting the oscillating signal into a smooth amplitude curve. The instantaneous envelope is a non-negative real number, which intuitively reflects the intensity change of the echo signal. The envelope curve presents the contour of the echo pulse intensity changing with time in the form of non-negative real values, making the pipe reflection echo exhibit a prominent envelope peak, while the carrier oscillation details are eliminated, which is beneficial for subsequent peak search and amplitude comparison.
[0078] In this embodiment of the application, adaptive threshold detection based on noise low-level statistics identifies and retains candidate pipe reflection peak points, including:
[0079] Noise statistics calculation: Take the first part of the echo sequence... The instantaneous envelope data of each sampling point interval is used as a reflection-free interval, which contains no pipe reflection echoes and only environmental and circuit noise. The noise statistics of the instantaneous envelope data in the reflection-free interval are calculated, including the noise mean. and standard deviation .
[0080] Adaptive threshold construction: Constructing detection thresholds using noise statistics The formula is:
[0081] ;
[0082] in, =3.5 is the threshold coefficient, determined experimentally. This detection threshold... The physical meaning is: if the instantaneous envelope of the echo exceeds the mean of the noise layer by 3.5 standard deviations at a certain moment, then there is sufficient reason to believe that there is a significant reflector at this point.
[0083] Candidate peak determination: in the echo sequence Search for all local maxima within the region, sort them by magnitude from largest to smallest, and select the local maximum with the largest magnitude as the first local maximum. Record the time index of the candidate channel reflection peak. The selection of the strongest peak value is based on the following physical understanding: the acoustic impedance mismatch between the outer wall of the pipe and the soil is much greater than the acoustic impedance mismatch of the tiny scatterers inside the soil. Therefore, the amplitude of the pipe-reflected echo should be the largest among all echoes under normal conditions.
[0084] It is known that there is a time interval between the emission of the ultrasonic pulse and the return of the reflected wave from the shallowest pipe to the ground. The initial part of the echo sequence corresponding to this interval contains almost only environmental noise and extremely weak shallow surface scattering. Therefore, this invention selects this interval to construct the detection threshold. This threshold is entirely determined by the noise statistics of the channel itself and does not require manual preset of a fixed value. When the soil moisture content increases, leading to an increase in background noise, and The threshold will increase accordingly. The threshold is automatically raised to maintain a constant false alarm control level; conversely, in dry, low-noise soil, the threshold is automatically lowered to maintain the detection sensitivity for weak pipe echoes. This adaptive characteristic allows the algorithm to operate normally under different site conditions without the need for specific parameter tuning.
[0085] S3: Based on the transit time and medium sound velocity corresponding to the candidate reflection peak points for each channel, calculate the initial burial depth estimate for each channel through time-depth conversion. Simultaneously, use the echo signal-to-noise ratio corresponding to each candidate reflection peak point as a quality indicator. Figure 2 As shown.
[0086] In this embodiment of the application, based on the transit time and medium sound velocity corresponding to the candidate reflection peak points of each channel, the initial burial depth estimate corresponding to each channel is calculated, including:
[0087] Obtain the candidate peak point index for each channel Then, combined with the reference time index of the transmitted pulse. The transit time of ultrasound in the soil medium can be converted into the pipe burial depth using the time-depth conversion formula:
[0088]
[0089] in, For the first The initial burial depth of the channel was estimated. In this embodiment, the test soil was medium-dense sandy clay with a moisture content of approximately 15%, therefore, the initial burial depth was selected. The ultrasonic propagation velocity in the sandy clay in the embodiments of this application is... The sampling frequency is 2, which reflects the complete round-trip journey of the ultrasonic wave from the ground to the pipe and back to the ground.
[0090] In this embodiment of the application, the formula for calculating the echo signal-to-noise ratio corresponding to each candidate reflection peak point is as follows:
[0091]
[0092] in, For the first The signal-to-noise ratio of the channel echo. For the first The instantaneous envelope at the candidate peak point of the channel. For the first The average noise level of the channel.
[0093] As we know, the signal-to-noise ratio (SNR) reflects the degree to which the candidate echo protrudes relative to the noise layer. A higher SNR indicates a clearer pipe reflection signal in that channel, less spurious interference, and a higher reliability of its burial depth estimate. Subsequent fusion steps will use this metric to weight each channel, quantifying signal quality to avoid low-quality data dominating the final result and improving fusion accuracy.
[0094] S4: The initial burial depth estimates for all channels are aggregated into a unified dataset. Neighborhood clustering is performed with a preset depth tolerance as a constraint. The cluster with the largest number of members is selected as the effective depth cluster. After validity determination, a weighted average is calculated using the echo signal-to-noise ratio of each member within the effective depth cluster as the weight, yielding the final burial depth of the pipeline. Figure 2 As shown.
[0095] In this embodiment of the application, step S4 specifically includes:
[0096] S41. Constructing the dataset ,in For the first The initial estimated burial depth of the channel;
[0097] S42. For each initial burial depth estimate in dataset D Statistics satisfy constraints Number of neighboring members , =3cm is the preset depth tolerance;
[0098] S43. Select the number of neighbor members Maximum initial burial depth estimate As cluster centers, the initial burial depth estimate will be included. All constraints, including Initial burial depth estimate Included in effective depth cluster ;
[0099] S44, if effective depth cluster The measurement is considered valid if the number of members is greater than or equal to half of the total number of channels; if the effective depth cluster... If the number of members is less than half of the total number of channels, it indicates that the echoes from most channels do not form a consistent direction. This may be due to reasons such as the absence of a pipeline below the measurement area, the pipeline being buried at a depth exceeding the probe's range, or extremely poor coupling conditions. Since most channels did not detect valid echoes, an invalid flag is output, prompting the probe to be repositioned or the coupling status to be checked.
[0100] S45. After determining that the measurement is valid, the effective depth cluster... The initial burial depth estimate is weighted by the signal-to-noise ratio (SNR). The higher the SNR, the greater the weight. The final burial depth calculation formula is as follows:
[0101]
[0102] in, The final burial depth of the pipeline.
[0103] It can be seen that the initial burial depth estimates for each channel are obtained through S3. and corresponding signal-to-noise ratio If all four channels correctly identify the echo of the same pipe, their estimated burial depth values should be highly concentrated. However, if a channel is misled by false reflections caused by localized debris or cavities, its estimated value will deviate from the others. Neighborhood clustering aims to group together the more consistent estimates and eliminate outliers. Using signal-to-noise ratio as the weight means that channels with clearer echoes and less interference have greater weight in the final result, further improving fusion accuracy.
[0104] In this embodiment, an aluminum pipe with an outer diameter of 40 mm and a wall thickness of 4 mm was selected as the test object. The experiment was conducted in an outdoor sandy clay test trench. The pipe was horizontally buried in the backfilled and compacted sandy clay test trench, with six burial depths set from the top outer wall of the pipe to the ground: 0.1 m, 0.15 m, 0.2 m, 0.25 m, 0.3 m, and 0.35 m. Each burial depth was measured 20 times. After the lower-level computer uploaded the raw echo data from each channel to the upper-level computer, it processed the data using the method described in this invention and output the pipe burial depth. The experimental results are as follows: Figure 4 As shown in the box plot, the burial depth error distribution of 20 repeated measurements under various working conditions is presented. It can be observed that the measurement error increases with increasing distance, but the overall error remains within 1 cm. These experimental results demonstrate that the algorithm can effectively improve distance accuracy.
[0105] The method proposed in this invention has three core advantages: high measurement accuracy, strong environmental adaptability, and low deployment cost. Compared with ground-penetrating radar, the required equipment is only a few general-purpose ultrasonic transducers and a small data acquisition module, significantly reducing the system cost. Furthermore, data interpretation does not rely on operator experience, and the algorithm can automatically provide pipe depth values. Compared with electromagnetic induction pipe gauges, this method is not limited by pipe material and can generate effective echoes for various pipe materials such as steel pipes, cast iron pipes, concrete pipes, and PE pipes, making it more widely applicable. Compared with deep learning algorithms, it does not require collecting a large number of labeled training samples or GPU computing resources. The entire algorithm can complete calculations in milliseconds on an ARM embedded platform, making it suitable for integration into portable field devices.
[0106] This invention fills the gap in non-destructive testing technology for underground pipeline burial depth based on ultrasonic array multi-channel fusion. It has good engineering application prospects in urban pipeline network surveys and daily inspections, and is expected to replace some of the high-cost and low-efficiency traditional detection methods, promoting the evolution of underground pipeline information collection towards a more intelligent and economical direction.
[0107] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for ultrasonic testing of underground pipeline burial depth based on multi-channel echo collaborative fusion, characterized in that, The steps are as follows: S1: Multiple sets of ultrasonic transducers are deployed on the ground in a predetermined spatial configuration in the area to be tested. Each set of ultrasonic transducers emits ultrasonic pulses into the ground and receives reflected echo signals from the outer wall of the underground pipeline. The reflected echo signals are conditioned and converted from analog to digital at the front end to obtain a multi-channel digital echo sequence. S2: For each group of digital echo sequences, cascaded sliding mean denoising, Hilbert transform envelope extraction, and adaptive threshold detection based on noise-based low-level statistics are performed sequentially to identify and retain candidate pipe reflection peak points; S3: Based on the transit time and medium sound velocity corresponding to the candidate reflection peak points of each channel, the initial burial depth estimate of each channel is calculated by time-depth conversion, and the echo signal-to-noise ratio corresponding to each candidate reflection peak point is used as a quality index. S4: Collect the initial burial depth estimates of all channels into a unified dataset, perform neighborhood clustering with the preset depth tolerance as a constraint, select the cluster with the most members as the effective depth cluster, and after the validity is determined, perform a weighted average with the echo signal-to-noise ratio of each member in the effective depth cluster as the weight to obtain the final burial depth of the pipeline.
2. The ultrasonic detection method for underground pipeline burial depth based on multi-channel echo collaborative fusion according to claim 1, characterized in that, For each set of digitized echo sequences, cascaded moving average noise reduction is performed sequentially, including: First-level moving average filtering: For the first... raw echo sequence of the channel With window length For coarse noise reduction, the formula is: ； in, Channel number, For sampling point index, This is the output signal of the first-stage moving average filter; Second-stage moving average filtering: This applies the output signal from the first-stage moving average filtering to... With window length For fine noise reduction, the formula is: ； in, This is the output signal of the second-stage moving average filter. .
3. The ultrasonic detection method for underground pipeline burial depth based on multi-channel echo collaborative fusion according to claim 2, characterized in that, Performing the Hilbert transform envelope includes: Hilbert transform: for the output signal of the second-stage moving average filter Performing the Hilbert transform yields the orthogonal components: Analytical signal construction: The output signal of the second-stage moving average filter Orthogonal components Combined into analytical signals: ； in, The imaginary unit; Instantaneous envelope calculation: taking the analytic signal The modulus value is used as the instantaneous envelope. .
4. The ultrasonic detection method for underground pipeline burial depth based on multi-channel echo collaborative fusion according to claim 3, characterized in that, Adaptive threshold calculation methods based on noise level statistics include: Noise statistics calculation: Take the first part of the echo sequence... The instantaneous envelope data of each sampling point interval is used as the non-reflection interval. The noise statistics of the instantaneous envelope data in the non-reflection interval are calculated, including the noise mean. and standard deviation ; Adaptive threshold construction: Constructing detection thresholds using noise statistics The formula is: ； in, This is the threshold coefficient.
5. The ultrasonic detection method for underground pipeline burial depth based on multi-channel echo collaborative fusion according to claim 4, characterized in that, Identify and retain candidate pipe reflection peaks, including: In the echo sequence Search for all local maxima within the region, sort them by magnitude from largest to smallest, and select the local maximum with the largest magnitude as the first local maximum. Record the time index of the candidate channel reflection peak. .
6. The ultrasonic detection method for underground pipeline burial depth based on multi-channel echo collaborative fusion according to claim 5, characterized in that, Based on the transit time and medium sound velocity corresponding to the candidate reflection peak points of each channel, the initial burial depth estimate for each channel is calculated through time-depth transformation, including: ； in, For the first The initial estimated burial depth of the channel, For the speed of ultrasonic propagation, Where n is the sampling frequency, and n0 is the reference time index of the transmitted pulse.
7. The ultrasonic detection method for underground pipeline burial depth based on multi-channel echo collaborative fusion according to claim 6, characterized in that, The formula for calculating the echo signal-to-noise ratio corresponding to each candidate reflection peak point is: ； in, For the first The signal-to-noise ratio of the channel echo. For the first The instantaneous envelope at the candidate peak point of the channel. For the first The average noise level of the channel.
8. The ultrasonic detection method for underground pipeline burial depth based on multi-channel echo collaborative fusion according to claim 7, characterized in that, The initial burial depth estimates of all channels are aggregated into a unified dataset. Neighborhood clustering is performed with a preset depth tolerance as a constraint. The cluster with the largest number of members is selected as the effective depth cluster, including: S41. Constructing the dataset ,in For the first The initial estimated burial depth of the channel; S42. For each initial burial depth estimate in dataset D Statistics satisfy constraints Number of neighboring members ; S43. Select the number of neighbor members Maximum initial burial depth estimate As cluster centers, the initial burial depth estimate will be included. All constraints, including Initial burial depth estimate Included in effective depth cluster .
9. The ultrasonic detection method for underground pipeline burial depth based on multi-channel echo collaborative fusion according to claim 8, characterized in that, Validity determination includes: If effective depth cluster The measurement is considered valid if the number of members is greater than or equal to half of the total number of channels; if the effective depth cluster... If the number of members is less than half of the total number of channels, most channels will not detect valid echoes, and an invalid flag will be output, prompting you to redeploy the probes or check the coupling status.
10. The ultrasonic detection method for underground pipeline burial depth based on multi-channel echo collaborative fusion according to claim 9, characterized in that, A weighted average is calculated using the echo signal-to-noise ratio of each member within the effective depth cluster as the weight, including: ； in, The final burial depth of the pipeline.