An ultrasonic detection method for accurate positioning of underground pipelines

Abstract

The present application discloses an ultrasonic detection method for accurate positioning of underground pipelines, which uses a signal analysis and coding algorithm based on deep learning to decompose and transform ultrasonic receiving signals and extract time-frequency features, and at the same time calculates the kurtosis distribution and extracts the kurtosis distribution features of the ultrasonic receiving signals, and intelligently determines the pipeline location information based on the fast dynamic time-frequency analysis between the time-frequency features of each ultrasonic receiving signal and the kurtosis distribution features of the ultrasonic receiving signal. In this way, it is possible to accurately dig out the anomalies caused by underground pipelines hidden in the background signal, and extract useful information related to the underground pipelines from the background signal, effectively overcoming the shortcomings of traditional signal processing technology in extraction accuracy and specific anomaly recognition, thereby improving the positioning accuracy of the detection equipment.

Description

Technical Field

[0001] The present application relates to the field of intelligent detection, and more specifically, to an ultrasonic detection method for accurately locating underground pipelines. Background Art

[0002] In the process of urban construction and infrastructure maintenance, accurate positioning of underground pipelines is of great significance. With the continuous advancement of urbanization, underground pipelines are densely distributed. Traditional underground pipeline detection methods such as electromagnetic method and ground penetrating radar method are more effective in detecting metal pipelines, but have limitations in detecting non-metallic pipelines, especially non-metallic pipelines with great burial depths.

[0003] Specifically, the electromagnetic method is mainly used to detect underground metal pipelines and pipelines with metal frames, but it can usually only detect pipelines within a buried depth of 5 meters. The ground penetrating radar method can be used to detect metal and non-metallic pipelines, but its detection effect is easily affected by the site environment and geological conditions, and the detection depth is limited, generally not exceeding 3 meters in soft soil areas. The above methods are not effective for detecting deeply buried underground pipelines.

[0004] Therefore, an optimized ultrasonic detection scheme based on the water-perforating method for precise positioning of underground pipelines is desired. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides an ultrasonic detection method for precise positioning of underground pipelines, which decomposes and transforms the ultrasonic receiving signal and extracts time-frequency features by adopting a signal analysis and coding algorithm based on deep learning, and at the same time calculates the kurtosis distribution and extracts the kurtosis distribution features of the ultrasonic receiving signal, so as to intelligently obtain the pipeline location information based on the fast dynamic time-frequency query representation between the time-frequency features of each ultrasonic receiving signal and the kurtosis distribution features of the ultrasonic receiving signal. The scheme of the present application can accurately dig out the abnormal features generated by the underground pipeline hidden in the background signal, and extract useful information related to the underground pipeline from the complex background, effectively overcoming the shortcomings of traditional signal processing technology in extraction accuracy and specific abnormal features, thereby improving the detection and positioning accuracy.

[0006] According to one aspect of the present application, there is provided an ultrasonic detection method for precise positioning of underground pipelines, comprising: collecting pipeline construction data in a target area; based on the pipeline construction data, drilling holes on both sides of the target pipeline using a water jet drilling technique to obtain two working holes; arranging an ultrasonic transmitter in one of the two working holes, and arranging an ultrasonic receiver in the other working hole; emitting an ultrasonic pulse signal through the ultrasonic transmitter, and receiving an ultrasonic receiving signal through the ultrasonic receiver; transmitting the ultrasonic receiving signal to a data processing terminal for analysis to determine the underground pipeline positioning result; wherein the ultrasonic transmitter is used to transmit the ultrasonic pulse ... The ultrasonic receiving signal is transmitted to a data processing terminal for analysis to determine the underground pipeline positioning result, including: decomposing and transforming the ultrasonic receiving signal and extracting the signal time-frequency feature to obtain a set of time-frequency features of the ultrasonic receiving signal; calculating the kurtosis distribution and extracting the feature of the ultrasonic receiving signal to obtain the kurtosis distribution feature of the ultrasonic receiving signal; performing a dynamic time-frequency query response by fast semantic matching on the set of time-frequency features of the ultrasonic receiving signal and the kurtosis distribution feature of the ultrasonic receiving signal to obtain the time-frequency query response feature of the ultrasonic receiving signal, and obtaining the pipeline location information based on the time-frequency query response feature of the ultrasonic receiving signal.

[0007] Compared with the prior art, the ultrasonic detection method for accurate positioning of underground pipelines provided by the present application uses a signal analysis and coding algorithm based on deep learning to decompose and transform the ultrasonic receiving signal and extract the time-frequency features, and at the same time calculates the kurtosis distribution and extracts the kurtosis distribution features of the ultrasonic receiving signal, so as to intelligently obtain the pipeline location information based on the fast dynamic time-frequency query representation between the time-frequency features of each ultrasonic receiving signal and the kurtosis distribution features of the ultrasonic receiving signal. The scheme of the present application can accurately mine the abnormal features generated by the underground pipeline hidden in the signal, extract useful information related to the underground pipeline from the complex background, and effectively overcome the shortcomings of traditional signal processing technology in extraction accuracy and specific abnormal features, thereby improving the detection and positioning accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0009] Figure 1 The present invention is a flowchart of an ultrasonic detection method for accurately locating underground pipelines according to an embodiment of the present application.

[0010] Figure 2 Schematic diagram of data flow of an ultrasonic detection method for precise positioning of underground pipelines according to an embodiment of the present application.

[0011] Figure 3 This is a flowchart of sub-step S5 of the detection process of the ultrasonic detection method for accurate positioning of underground pipelines according to an embodiment of the present application.

[0012] Figure 4 This is a flowchart of sub-step S53 of the detection process of the ultrasonic detection method for accurate positioning of underground pipelines according to an embodiment of the present application. DETAILED DESCRIPTION

[0013] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0014] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0015] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0016] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0017] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0018] Ultrasonic detection is a technology that uses sound waves with a frequency higher than the upper limit of human hearing (about 20kHz) to detect objects, measure distances or obtain information. The application of ultrasonic detection in underground pipeline positioning is mainly focused on technologies that can effectively identify and accurately locate pipelines, especially for non-metallic pipelines made of PE, PVC, PCCP, etc. that are buried deep and difficult to reach with traditional methods.

[0019] It should be understandable that deep-buried non-metallic pipelines are widely used in many urban infrastructure fields such as water supply, drainage, and gas transmission. Accurately grasping the location information of these pipelines is crucial to ensuring the safe operation of urban lifelines, avoiding construction damage, reducing maintenance costs, and improving urban management efficiency. When faced with complex geological conditions, existing ultrasonic detection technology has difficulty effectively extracting sufficiently accurate and targeted characteristic signals from the received ultrasonic signals to achieve high-precision positioning. For example, when ultrasonic signals propagate underground, they are affected by various interference factors such as uneven medium composition, groundwater level fluctuations, and reflection and scattering from other surrounding objects, making the received signals complex, and traditional signal processing methods cannot fully explore the key features hidden in them that are related to underground pipelines.

[0020] Based on this, in the technical solution of the present application, an ultrasonic detection method for accurately locating underground pipelines is proposed. Figure 1 The present invention is a flowchart of an ultrasonic detection method for accurately locating underground pipelines according to an embodiment of the present application. Figure 2 FIG. 1 is a data flow diagram of an ultrasonic detection method for accurately locating underground pipelines according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to an embodiment of the present application, an ultrasonic detection method for precise positioning of underground pipelines includes: S1, collecting pipeline construction data in a target area; S2, based on the pipeline construction data, drilling holes on both sides of the target pipeline using a water jet drilling technology to obtain two working holes; S3, deploying an ultrasonic transmitter in one of the two working holes, and deploying an ultrasonic receiver in the other working hole; S4, transmitting an ultrasonic pulse signal through the ultrasonic transmitter, and receiving an ultrasonic receiving signal through the ultrasonic receiver; S5, transmitting the ultrasonic receiving signal to a data processing terminal for analysis to determine the underground pipeline positioning result.

[0021] In particular, the S1 collects pipeline construction data in the target area. Before conducting ultrasonic detection for accurate positioning of underground pipelines, it is crucial to collect pipeline construction data in the target area. In one example, first, it is necessary to obtain construction drawings and other information about the distribution of underground pipelines as the basis for the design of the detection work. In particular, the S2 uses water-jetting drilling technology on both sides of the target pipeline based on the pipeline construction data to drill holes to obtain two working holes. In one example, first, the technicians will select the specific location of the drilling according to the pipeline construction data collected in the early stage. Through analysis, the approximate direction and burial depth range of the target pipeline can be preliminarily determined, providing a basis for the design of the detection work. On-site surveys can obtain more information, such as pipeline ground signs, geological conditions, and on-site building distribution, to further confirm the direction of the underground pipeline. After determining the direction of the pipeline, a detection profile is arranged perpendicular to the direction of the pipeline, and several working holes are set at a spacing of 1~2m on the detection profile. Two adjacent boreholes are grouped together, and ultrasonic transmitters and receivers are installed respectively. In order to ensure the safety of underground pipelines, water-jetting is used for drilling the working holes. The so-called water flushing method refers to the use of a high-pressure water pump to supply water to the drill pipe, forming a head pressure at the bottom of the drill pipe to break the soil and then flow to the ground with the water flow. When it reaches a certain depth, a working hole is formed. This method is fast, efficient, and safe, and is especially suitable for drilling operations under soft soil geological conditions. During the drilling process, a dedicated person should be responsible for real-time monitoring of the stability of the hole wall and changes in the surrounding environment. If abnormal conditions such as hole wall collapse are found, timely measures should be taken to deal with them, and the operation should be suspended if necessary. When the specified depth is successfully reached, stop the water supply and remove the drill pipe, clean the hole and place the PVC casing to create good conditions for the next step of laying ultrasonic equipment. After the drilling is completed, measure the diameter and depth of each hole to ensure that the design standards are met. Normally, the depth of the two holes should be equal and consistent with the design. Check whether there are cracks, looseness or other defects in the hole wall. If necessary, the hole wall can be reinforced by injecting cement slurry to prevent it from collapsing in subsequent operations. Use special detection tools to go deep into the bottom of the hole to check whether there are any foreign objects or water accumulation. If so, clean them up in time to ensure a clean environment inside the hole.

[0022] In particular, in S3, an ultrasonic transmitter is arranged in one of the two working holes, and an ultrasonic receiver is arranged in the other working hole. Taking into account the differences in ultrasonic propagation under different geological conditions, the technicians formulate a detailed layout plan based on the actual situation on site. During observation, they are placed at the same depth of the two test holes respectively, so as to ensure the shortest distance between the two, thereby reducing the influence of signal attenuation and noise interference. During detection, first ensure that the working hole is filled with water to provide sufficient coupling conditions for the ultrasonic probe, and make the axes of the transmitter and receiver as parallel as possible through the adjustment device to ensure that the ultrasonic transmission distance is equal.

[0023] In particular, the S4 transmits an ultrasonic pulse signal through the ultrasonic transmitter and receives an ultrasonic reception signal through the ultrasonic receiver. In one example, after the ultrasonic transmitter is started, a series of pulse signals are sent to the underground medium according to a preset program. Each pulse signal is composed of ultrasonic waves within a specific frequency range, and these waves can propagate between different types of soil, rocks, and potential target objects (such as underground pipelines). Since there are differences in the propagation speed and attenuation degree of ultrasonic waves in various media, the information of the underground target body can be inferred by analyzing the characteristics of the received reflected waveform. Therefore, selecting an appropriate frequency range is very important for improving detection accuracy. As the ultrasonic pulse signal propagates in the underground medium and interacts with the target object, a portion of the energy will return to the receiver position in the form of a reflected wave. At this time, the ultrasonic receiver located in another test hole will receive these reflected signals and convert them into electrical signals for further analysis.

[0024] In particular, in S5, the ultrasonic receiving signal is transmitted to a data processing terminal for analysis to determine the underground pipeline positioning result. Figure 3 As shown, the S5 includes: S51, decomposing and transforming the ultrasonic receiving signal and extracting the signal time-frequency features to obtain a set of time-frequency features of the ultrasonic receiving signal; S52, calculating the kurtosis distribution and extracting the features of the ultrasonic receiving signal to obtain the kurtosis distribution features of the ultrasonic receiving signal; S53, performing a dynamic time-frequency query response with fast semantic matching on the set of time-frequency features of the ultrasonic receiving signal and the kurtosis distribution features of the ultrasonic receiving signal to obtain the time-frequency query response features of the ultrasonic receiving signal, and obtaining the pipeline location information based on the time-frequency query response features of the ultrasonic receiving signal.

[0025] Specifically, the S51 decomposes and transforms the ultrasonic receiving signal and extracts the signal time-frequency features to obtain a set of time-frequency features of the ultrasonic receiving signal. In a specific example of the present application, first, the ultrasonic receiving signal is subjected to variational modal decomposition to obtain a set of intrinsic modal functions of the ultrasonic receiving signal. Considering that the actually received ultrasonic signal itself contains multiple frequency components and various types of noise, signals of different frequencies are affected by underground pipelines to different degrees during propagation. For example, high-frequency signals may be more easily scattered by small-sized obstacles, while low-frequency signals may be more likely to penetrate some obstacles. These different frequency components are mixed with each other in the received signal, which is not conducive to separately analyzing the contribution of each frequency component to positioning. Based on this, in the technical solution of the present application, the ultrasonic receiving signal is subjected to variational modal decomposition to decompose the complex signal into a limited number of sub-signals with different characteristics, and a set of intrinsic modal functions of the ultrasonic receiving signal is obtained. Therefore, through variational modal decomposition, each intrinsic modal function obtained has a relatively single frequency characteristic and time characteristic, which can more accurately capture the key features of the signal, thereby improving the understanding and analysis capabilities of the original signal.

[0026] Next, each ultrasonic receiving signal intrinsic mode function in the set of ultrasonic receiving signal intrinsic mode functions is Hilbert transformed to obtain a set of ultrasonic receiving signal Hilbert spectra. Considering that for underground pipeline positioning, the time and frequency changes of ultrasonic signals propagating underground are closely related. Different underground structures (such as different depths of pipelines, obstacles of different materials, etc.) will cause different frequency changes of ultrasonic waves at different times. Considering that the Hilbert transform can reveal detailed information such as the instantaneous frequency and instantaneous amplitude of the signal. Therefore, in order to understand the dynamic behavior of the signal in the time domain, in the technical solution of the present application, each ultrasonic receiving signal intrinsic mode function in the set of ultrasonic receiving signal intrinsic mode functions is Hilbert transformed to help identify subtle changes in the signal, such as mutation points or specific patterns, to obtain a set of ultrasonic receiving signal Hilbert spectra. It can be understood that in an underground environment, the instantaneous characteristics of an ultrasonic signal can reflect the instantaneous situation of its interaction with an underground pipeline. For example, when the ultrasonic wave encounters the pipeline boundary, its instantaneous frequency may mutate. The Hilbert spectrum obtained by Hilbert transform can better show this mutation, which helps to dig out the subtle time-frequency correlation information hidden in the intrinsic mode function of the ultrasonic receiving signal, which may be an indicator of the existence of underground pipelines. In one example, the intrinsic mode function of each ultrasonic receiving signal is Hilbert transformed to generate an analytical signal. This step not only retains the information of the original IMF, but also introduces the instantaneous amplitude and phase information. By taking the derivative of the analytical signal and converting it into an angular frequency, the instantaneous frequency can be obtained, which provides the frequency characteristics of the signal changing over time. Using the instantaneous frequency and instantaneous amplitude, a Hilbert spectrum is drawn, in which the horizontal axis represents time, the vertical axis represents frequency, and the color or grayscale represents the amplitude. Therefore, each Hilbert spectrum shows the frequency distribution and intensity change of the corresponding signal intrinsic mode function over the entire time period. Finally, the Hilbert spectra corresponding to all signal intrinsic mode functions are fused to obtain a set of Hilbert spectra of the ultrasonic receiving signal.

[0027] Furthermore, each ultrasonic receiving signal Hilbert spectrogram in the set of ultrasonic receiving signal Hilbert spectrograms is respectively passed through a signal time-frequency feature extractor based on a CNN-LSTM model to obtain a set of ultrasonic receiving signal time-frequency feature vectors as a set of ultrasonic receiving signal time-frequency features. It should be understood that in the detection process of underground pipelines, the time-frequency characteristics of ultrasonic signals will change with the characteristics of the underground environment and the pipeline itself (such as the material, shape, burial depth, etc. of the pipeline). For example, as the burial depth of the pipeline increases, the frequency change of the ultrasonic signal in the spectrogram may be delayed in time and the amplitude may decrease. Therefore, in order to more clearly understand and analyze the time-frequency dynamic change characteristics of the ultrasonic receiving signal, in the technical solution of the present application, each ultrasonic receiving signal Hilbert spectrogram in the set of ultrasonic receiving signal Hilbert spectrograms is analyzed by a signal time-frequency feature extractor based on a CNN-LSTM model to obtain a set of ultrasonic receiving signal time-frequency feature vectors. Specifically, the CNN-LSTM model combines the advantages of convolutional neural networks (CNNs) and long short-term memory networks (LSTMs). CNN can automatically extract local features of images (the Hilbert spectrogram can be regarded as a special image form, a two-dimensional time-frequency representation), and can effectively capture the characteristic patterns of different positions, frequency ranges and time intervals in the spectrogram through the convolution kernel in the convolution layer. LSTM is good at processing sequence information. Since the time-frequency information of the Hilbert spectrogram is ordered in the time dimension, LSTM can further integrate and optimize these features initially extracted by CNN in the time series, so as to better extract and understand the key time-frequency features of the signal, and provide data support for subsequent precise positioning.

[0028] Specifically, the S52 is to calculate the kurtosis distribution and extract the features of the ultrasonic receiving signal to obtain the kurtosis distribution features of the ultrasonic receiving signal. That is, in the technical solution of the present application, first, the kurtosis value distribution map of the ultrasonic receiving signal is calculated to obtain the kurtosis value distribution map of the ultrasonic receiving signal. Considering that in the detection of underground pipelines, different soils, rock distributions and possible water flows make the propagation path of ultrasonic signals complex and changeable, and the target object (such as a pipeline) usually has different characteristics from the surrounding environment signal due to its material, shape and location. Based on this, in the technical solution of the present application, the kurtosis value distribution map of the ultrasonic receiving signal is calculated to obtain the kurtosis value distribution map of the ultrasonic receiving signal. That is, the kurtosis is extremely sensitive to the mutation and discontinuity in the signal, and these anomalies often originate from the difference between the reflected signal of the target object and the environmental signal. When the distribution map is calculated, the high kurtosis value points therein represent the abnormal changes in the signal, corresponding to the uniqueness of the reflected signal of the target object. In one example, the specific steps of calculating the kurtosis value distribution diagram of the ultrasonic receiving signal include: first, preprocessing the ultrasonic receiving signal; then, dividing the preprocessed ultrasonic signal into multiple short time windows; then, for the signal segment in each time window, calculating its fourth-order normalized moment, that is, kurtosis; specifically, by calculating the average value of the fourth power of the deviation of the signal value in each time window from its mean value and normalizing it, the kurtosis value of the time window can be obtained. This process involves determining the sum of the fourth squares of all sample points in each time window, dividing it by the square of the sample variance, and finally subtracting 3 (so that the kurtosis of the normal distribution is 0) to obtain the corrected kurtosis value. Then, the sequence of the calculated kurtosis values ​​is arranged according to the corresponding time position to obtain a kurtosis value sequence; finally, according to the kurtosis value sequence, a kurtosis value distribution diagram is drawn, in which the horizontal axis represents time and the vertical axis represents the corresponding kurtosis value.

[0029] Then, the kurtosis value distribution map of the ultrasonic receiving signal is passed through a signal kurtosis distribution feature extractor based on an MCNN network to obtain an ultrasonic receiving signal kurtosis distribution feature vector as the ultrasonic receiving signal kurtosis distribution feature. Considering that the kurtosis value distribution map of the ultrasonic receiving signal reflects the difference between the underground pipeline and the surrounding environment and the change of the signal during the propagation process, and the MCNN (multi-dimensional convolutional neural network) network has a powerful feature extraction capability, it can automatically learn the complex feature patterns at different positions and scales in the kurtosis value distribution map, such as peaks and valleys of different shapes and sizes and the spatial relationship between them, etc. These features are crucial for accurately judging the characteristics of underground pipelines. Based on this, in the technical solution of the present application, the kurtosis value distribution map of the ultrasonic receiving signal is processed by a signal kurtosis distribution feature extractor based on an MCNN network to extract representative and distinguishing features from different distributions through its multiple convolutional layers and pooling layers to obtain an ultrasonic receiving signal kurtosis distribution feature vector.

[0030] Specifically, in S53, a dynamic time-frequency query response is performed by fast semantic matching of the set of ultrasonic receiving signal time-frequency features and the ultrasonic receiving signal kurtosis distribution features to obtain the ultrasonic receiving signal time-frequency query response features, and the pipeline location information is obtained based on the ultrasonic receiving signal time-frequency query response features. Considering that the set of ultrasonic receiving signal time-frequency feature vectors reflects the change law of the signal in the time and frequency dimensions, it is crucial to understand the dynamic process of the interaction between ultrasonic waves and various media and target pipelines during underground propagation. The ultrasonic receiving signal kurtosis distribution feature vector highlights the non-Gaussian characteristics and abnormal change points in the signal, and this information is closely related to the structural characteristics of the target object (such as an underground pipeline). Therefore, in order to quickly integrate the information they contain, thereby forming a more comprehensive and more accurate comprehensive feature representation that can describe the characteristics of the underground pipeline, so as to more accurately locate the signal characteristics related to the underground pipeline, in the technical solution of the present application, a dynamic time-frequency query response is performed by fast semantic matching of the set of ultrasonic receiving signal time-frequency feature vectors and the ultrasonic receiving signal kurtosis distribution feature vector to obtain an ultrasonic receiving signal time-frequency query response encoding vector. Specifically, in a specific example of the present application, Figure 4 As shown, the S53 includes: S531, performing a dynamic time-frequency query response by fast semantic matching on the set of the ultrasonic receiving signal time-frequency feature vectors and the ultrasonic receiving signal kurtosis distribution feature vector to obtain an ultrasonic receiving signal time-frequency query response coding vector as the ultrasonic receiving signal time-frequency query response feature; S532, obtaining the pipeline location information based on the ultrasonic receiving signal time-frequency query response coding vector.

[0031] More specifically, in S531, a dynamic time-frequency query response is performed by performing fast semantic matching on the set of the ultrasonic received signal time-frequency feature vectors and the ultrasonic received signal kurtosis distribution feature vector to obtain an ultrasonic received signal time-frequency query response encoding vector as the ultrasonic received signal time-frequency query response feature. Specifically, first, based on the ultrasonic received signal kurtosis distribution feature vector, an ultrasonic received signal positioning matching library feature vector is determined from the set of ultrasonic received signal time-frequency feature vectors. That is, in a specific example, first, the cross entropy of the ultrasonic received signal kurtosis distribution feature vector relative to each ultrasonic received signal time-frequency feature vector in the set of ultrasonic received signal time-frequency feature vectors is calculated to obtain a set of ultrasonic received signal kurtosis distribution-time-frequency feature fast query positioning factors; cross entropy, as an important indicator for measuring the difference between two probability distributions, is used here to quantify the semantic difference between the ultrasonic received signal kurtosis distribution feature vector and each ultrasonic received signal time-frequency feature vector. A smaller cross entropy value means a more similar semantic expression, so that the search range can be quickly narrowed to the most likely related area. Here, the cross entropy calculation not only considers direct semantic similarity, but also implicitly captures potential contextual relationships and semantic structures. This mechanism helps to improve retrieval efficiency, especially when facing large-scale data sets, and can significantly reduce unnecessary computational burdens while maintaining high retrieval accuracy. Then, the ultrasonic receiving signal time-frequency feature vector corresponding to the minimum value in the set of ultrasonic receiving signal kurtosis distribution-time-frequency feature fast query positioning factors is used as the ultrasonic receiving signal positioning matching library feature vector. It should be understood that the ultrasonic receiving signal time-frequency feature vector corresponding to the minimum value is considered to be semantically closest to the ultrasonic receiving signal kurtosis distribution feature vector. In this way, it can be ensured that the ultrasonic receiving signal time-frequency feature vector preliminarily screened has the highest correlation, providing a good starting point for subsequent more detailed searches. More specifically, the cross entropy of the ultrasonic receiving signal kurtosis distribution feature vector relative to each ultrasonic receiving signal time-frequency feature vector in the set of ultrasonic receiving signal time-frequency feature vectors is calculated using the following cross entropy calculation formula to obtain a set of ultrasonic receiving signal kurtosis distribution-time-frequency feature fast query positioning factors; wherein, the cross entropy calculation formula is: ;in, is the set of time-frequency feature vectors of the ultrasonic receiving signal, and is the first, second, and third in the set of time-frequency feature vectors of the ultrasonic receiving signal. and The time-frequency feature vector of the ultrasonic receiving signal, yes The eigenvalues ​​at each position in , is the kurtosis distribution eigenvector of the ultrasonic receiving signal, yes The eigenvalues ​​at each position in , yes and The number of eigenvalues ​​in , represents the logarithmic function value with base 2, and It is the first, second, and third in the set of ultrasonic receiving signal kurtosis distribution-time-frequency characteristics fast query positioning factors. and The kurtosis distribution of ultrasonic receiving signals - time-frequency characteristics quickly query the positioning factor, It is a collection of fast query positioning factors of the kurtosis distribution-time-frequency characteristics of the ultrasonic receiving signal.

[0032] Next, based on the ultrasonic receiving signal positioning matching library feature vector, determine the ultrasonic receiving signal fine-grained matching search window. That is, by constructing a small-range search area around the initially selected ultrasonic receiving signal positioning matching library feature vector, the system is able to perform in-depth analysis in a smaller and more targeted data subset. Doing so not only improves the search efficiency, but also enhances the model's ability to understand local semantic patterns. In addition, the design of the fine-grained matching search window allows the model to better handle the subtle differences between the ultrasonic receiving signal kurtosis distribution feature vector and each of the ultrasonic receiving signal time-frequency feature vectors. In particular, the vector at the center position of the ultrasonic receiving signal fine-grained matching search window is the ultrasonic receiving signal positioning matching library feature vector, and each ultrasonic receiving signal time-frequency feature vector in the ultrasonic receiving signal fine-grained matching search window is defined as the ultrasonic receiving signal fine-grained query time-frequency feature vector. More specifically, based on the ultrasonic receiving signal positioning matching library feature vector, the ultrasonic receiving signal fine-grained matching search window is determined by the following formula; wherein, the formula is: ;in, is the vector at the location of the minimum value, Locate and match the feature vector of the library for ultrasonic receiving signals. is the square of the vector norm, Indicates rounding down. is the size of the search window for fine-grained matching of ultrasonic received signals, is the fine-grained matching search window of the ultrasonic receiving signal.

[0033] Furthermore, the ultrasonic received signal kurtosis distribution feature vector and each feature vector in the ultrasonic received signal fine-grained matching search window are response-encoded to obtain the ultrasonic received signal time-frequency query response encoding vector. That is, in the technical solution of the present application, the ultrasonic received signal kurtosis distribution feature vector is linearly transformed to obtain the ultrasonic received signal kurtosis distribution query vector and the ultrasonic received signal kurtosis distribution value vector, and each ultrasonic received signal fine-grained query time-frequency feature vector in the ultrasonic received signal fine-grained matching search window is used as a key vector, and the ultrasonic received signal kurtosis distribution query vector, the ultrasonic received signal kurtosis distribution value vector and the key vector are subjected to fine-grained query encoding based on a heterogeneous converter to obtain the ultrasonic received signal time-frequency query response encoding vector. In this module, the interaction between query, key and value simulates the attention mechanism, captures the complex relationship between features through a multi-head self-attention layer, and finally outputs the ultrasonic received signal time-frequency query response encoding vector. More specifically, the ultrasonic receiving signal kurtosis distribution feature vector and each feature vector in the ultrasonic receiving signal fine-grained matching search window are response-encoded using the following response coding formula to obtain the ultrasonic receiving signal time-frequency query response coding vector; wherein the response coding formula is: ;in, and They are respectively the embedding matrix for querying the kurtosis distribution of ultrasonic received signals and the bias vector for querying the kurtosis distribution of ultrasonic received signals. is the query vector of the kurtosis distribution of the ultrasonic receiving signal, and are respectively the embedding matrix of the kurtosis distribution value of the ultrasonic receiving signal and the bias vector of the kurtosis distribution value of the ultrasonic receiving signal, is the kurtosis distribution value vector of ultrasonic receiving signal, is the first Fine-grained query of the time-frequency feature vector of ultrasonic receiving signals, for The transposed vector of yes Length, is matrix multiplication, is the activation function, is the time-frequency query response coding vector of the ultrasonic receiving signal.

[0034] More specifically, the S532 obtains the pipeline location information based on the ultrasonic receiving signal time-frequency query response coding vector. That is, in the technical solution of the present application, the ultrasonic receiving signal time-frequency query response coding vector is passed through the decoder-based underground pipeline locator to obtain the underground pipeline positioning result, and the underground pipeline positioning result includes the pipeline buried depth decoding value as the pipeline location information. It should be understood that the decoder-based underground pipeline locator can act as a "translator" to convert this complex coding vector into an underground pipeline positioning result with actual physical meaning, especially the pipeline buried depth decoding value. This is because in the encoding process, the time-frequency characteristics and kurtosis distribution characteristics have inherent mathematical and physical associations with the buried depth, location and other information of the pipeline. The decoder can use these association models learned in advance training to restore the corresponding location information from the coding vector, and realize the conversion from the signal feature domain to the spatial position domain. These positioning results can be directly used by engineers and construction personnel to guide actual engineering operations, such as determining the excavation depth, selecting the construction route, etc., so as to ensure the smooth progress of the project while reducing construction risks and costs.

[0035] Here, when the set of ultrasonic received signal time-frequency feature vectors represents the image semantic features of each ultrasonic received signal Hilbert spectrogram based on the image semantic short-range-long-range context association, and the ultrasonic received signal kurtosis distribution feature vector represents the image semantic features of the ultrasonic received signal kurtosis value distribution map, when performing dynamic time-frequency semantic query encoding based on a fast positioning mechanism, the misalignment of the dynamic time-frequency positioning mechanism under the differentiation of image semantic feature patterns will also cause the semantic query encoding richness of the ultrasonic received signal time-frequency query response encoding vector, resulting in repeated unstructured probability mapping, and thus it is necessary to improve the feature space-decoding target probability response effect of the ultrasonic received signal time-frequency query response encoding vector.

[0036] Therefore, it is preferred that when the ultrasonic receiving signal time-frequency query response coding vector is passed through the decoder-based underground pipeline locator to obtain the underground pipeline positioning result, the ultrasonic receiving signal time-frequency query response coding vector is optimized. Specifically, the optimization includes the following steps: calculating the absolute value sum of all eigenvalues ​​of the ultrasonic receiving signal time-frequency query response coding vector , and the square root of the sum of the squares of all eigenvalues ​​of the ultrasonic receiving signal time-frequency query response encoding vector , and Divide by To obtain the probability value of the ultrasonic receiving signal time-frequency query response code ,Right now: ;in, The first one represents the time-frequency query response encoding vector of the ultrasonic receiving signal The ultrasonic receiving signal time-frequency query response encoding vector Each eigenvalue in is normalized to the maximum value to obtain the probability vector of the ultrasonic receiving signal time-frequency query response coding .

[0037] ;in, Indicates the maximum value in the time-frequency query response encoding vector of the ultrasonic receiving signal The first one represents the probability vector of the time-frequency query response code of the ultrasonic receiving signal feature values.

[0038] The ultrasonic receiving signal time-frequency query response coding probability vector is subtracted from the ultrasonic receiving signal time-frequency query response coding probability value, and then multiplied by the reciprocal of the ultrasonic receiving signal time-frequency query response coding probability value to obtain the ultrasonic receiving signal time-frequency query response coding domain gradient vector ,in, Indicates point reduction, Represents dot product.

[0039] The ultrasonic receiving signal time-frequency query response coding probability vector is subtracted from the unit feature vector, and then divided by the ultrasonic receiving signal time-frequency query response coding domain gradient vector to obtain the ultrasonic receiving signal time-frequency query response coding differential vector, that is: ;in, represents the unit eigenvector, represents the bit-by-bit inverse of the gradient vector of the time-frequency query response coding domain of the ultrasonic receiving signal, that is, the inverse of each eigenvalue of the gradient vector of the time-frequency query response coding domain of the ultrasonic receiving signal is calculated, Represents the differential vector of the time-frequency query response encoding of the ultrasonic receiving signal.

[0040] Calculate the power function of the ultrasonic receiving signal time-frequency query response coding domain gradient vector with the ultrasonic receiving signal time-frequency query response coding probability value as an exponent , and the exponential function of the differential vector of the time-frequency query response encoding the ultrasonic receiving signal with the natural constant as the base The weighted sum is performed to obtain the optimized ultrasonic receiving signal time-frequency query response coding vector, namely: ;in, and represents the weighted hyperparameter, Indicates point addition, Represents the optimized ultrasonic receiving signal time-frequency query response coding vector.

[0041] That is, the mutually exclusive generalized representation behavior unit of the ultrasonic receiving signal time-frequency query response coding vector is constructed by discretely differentiating the probability density domain gradient corresponding to the ultrasonic receiving signal time-frequency query response coding vector, so as to enhance the static response characteristics of the short-range correlation mesoscopic information configuration of the ultrasonic receiving signal time-frequency query response coding vector to the generative probabilistic generalized representation behavior based on the different behavior unit organization spaces under the non-uniform topological architecture of the ultrasonic receiving signal time-frequency query response coding vector, thereby ensuring the steady-state convergence of the optimization process between the generated target and the extracted features in the feature space-generation probability mapping, so as to improve the accuracy of the underground pipeline positioning result obtained by the ultrasonic receiving signal time-frequency query response coding vector through the decoder-based underground pipeline locator.

[0042] In summary, the detection process of the ultrasonic detection method for accurate positioning of underground pipelines according to the embodiment of the present application is explained, which uses a signal analysis and coding algorithm based on deep learning to decompose and transform the ultrasonic receiving signal and extract the time-frequency features, and at the same time calculates the kurtosis distribution and extracts the kurtosis distribution features of the ultrasonic receiving signal, so as to intelligently obtain the pipeline location information based on the fast dynamic time-frequency query representation between the time-frequency features of each ultrasonic receiving signal and the kurtosis distribution features of the ultrasonic receiving signal. In this way, the subtle features hidden in the signal can be accurately excavated, and useful information related to the underground pipeline can be extracted from the complex background, which effectively overcomes the shortcomings of traditional signal processing technology in extraction accuracy and specific abnormal features, thereby improving the positioning accuracy of the detection equipment.

[0043] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. An ultrasonic detection method for accurate positioning of underground pipelines, characterized in that: include: Collect pipeline construction data in the target area; based on the pipeline construction data, use water jet drilling technology to drill holes on both sides of the target pipeline to obtain two working holes; arrange an ultrasonic transmitter in one of the two working holes, and arrange an ultrasonic receiver in the other working hole; transmit an ultrasonic pulse signal through the ultrasonic transmitter, and receive an ultrasonic receiving signal through the ultrasonic receiver; transmit the ultrasonic receiving signal to a data processing terminal for analysis to determine the underground pipeline positioning result; wherein, transmitting the ultrasonic receiving signal to a data processing terminal for analysis to determine the underground pipeline positioning result includes: decomposing and transforming the ultrasonic receiving signal and extracting the signal time-frequency feature to obtain a set of ultrasonic receiving signal time-frequency features; calculating the kurtosis distribution and extracting the feature of the ultrasonic receiving signal to obtain the kurtosis distribution feature of the ultrasonic receiving signal; performing a dynamic time-frequency query response with fast semantic matching on the set of ultrasonic receiving signal time-frequency features and the kurtosis distribution feature of the ultrasonic receiving signal to obtain the ultrasonic receiving signal time-frequency query response feature, and obtaining the pipeline location information based on the ultrasonic receiving signal time-frequency query response feature.

2. The ultrasonic detection method for accurate positioning of underground pipelines according to claim 1, characterized in that: The ultrasonic transmitter and the ultrasonic receiver are arranged at the same depth.

3. The ultrasonic detection method for accurate positioning of underground pipelines according to claim 2, characterized in that: The ultrasonic receiving signal is decomposed and transformed and the signal time-frequency feature is extracted to obtain a set of time-frequency features of the ultrasonic receiving signal, including: performing variational mode decomposition on the ultrasonic receiving signal to obtain a set of intrinsic mode functions of the ultrasonic receiving signal; performing Hilbert transformation on each ultrasonic receiving signal intrinsic mode function in the set of intrinsic mode functions of the ultrasonic receiving signal to obtain a set of Hilbert spectrograms of the ultrasonic receiving signal; and each ultrasonic receiving signal Hilbert spectrogram in the set of Hilbert spectrograms of the ultrasonic receiving signal is respectively passed through a signal time-frequency feature extractor based on a CNN-LSTM model to obtain a set of ultrasonic receiving signal time-frequency feature vectors as the set of time-frequency features of the ultrasonic receiving signal.

4. The ultrasonic detection method for accurate positioning of underground pipelines according to claim 3 is characterized in that: The kurtosis distribution calculation and feature extraction are performed on the ultrasonic receiving signal to obtain the kurtosis distribution feature of the ultrasonic receiving signal, including: calculating the kurtosis value distribution map of the ultrasonic receiving signal to obtain the kurtosis value distribution map of the ultrasonic receiving signal; passing the kurtosis value distribution map of the ultrasonic receiving signal through a signal kurtosis distribution feature extractor based on an MCNN network to obtain an ultrasonic receiving signal kurtosis distribution feature vector as the ultrasonic receiving signal kurtosis distribution feature.

5. The ultrasonic detection method for accurate positioning of underground pipelines according to claim 4, characterized in that: A dynamic time-frequency query response is performed by quickly semantically matching the set of time-frequency features of the ultrasonic received signal and the kurtosis distribution features of the ultrasonic received signal to obtain the ultrasonic received signal time-frequency query response features, and the pipeline location information is obtained based on the ultrasonic received signal time-frequency query response features, including: a dynamic time-frequency query response is performed by quickly semantically matching the set of time-frequency feature vectors of the ultrasonic received signal and the kurtosis distribution feature vector of the ultrasonic received signal to obtain the ultrasonic received signal time-frequency query response coding vector as the ultrasonic received signal time-frequency query response features; and the pipeline location information is obtained based on the ultrasonic received signal time-frequency query response coding vector.

6. The ultrasonic detection method for accurate positioning of underground pipelines according to claim 5, characterized in that: A dynamic time-frequency query response that performs fast semantic matching on the set of ultrasonic received signal time-frequency feature vectors and the ultrasonic received signal kurtosis distribution feature vector to obtain an ultrasonic received signal time-frequency query response encoding vector as the ultrasonic received signal time-frequency query response feature, including: determining an ultrasonic received signal positioning matching library feature vector from the set of ultrasonic received signal time-frequency feature vectors based on the ultrasonic received signal kurtosis distribution feature vector; determining an ultrasonic received signal fine-grained matching search window based on the ultrasonic received signal positioning matching library feature vector; and performing response encoding on the ultrasonic received signal kurtosis distribution feature vector and each feature vector in the ultrasonic received signal fine-grained matching search window to obtain the ultrasonic received signal time-frequency query response encoding vector.

7. The ultrasonic detection method for accurate positioning of underground pipelines according to claim 6, characterized in that: Based on the ultrasonic received signal kurtosis distribution feature vector, an ultrasonic received signal positioning matching library feature vector is determined from the set of ultrasonic received signal time-frequency feature vectors, including: calculating the cross entropy of the ultrasonic received signal kurtosis distribution feature vector relative to each ultrasonic received signal time-frequency feature vector in the set of ultrasonic received signal time-frequency feature vectors to obtain a set of ultrasonic received signal kurtosis distribution-time-frequency feature fast query positioning factors; and using the ultrasonic received signal time-frequency feature vector corresponding to the minimum value in the set of ultrasonic received signal kurtosis distribution-time-frequency feature fast query positioning factors as the ultrasonic received signal positioning matching library feature vector.

8. The ultrasonic detection method for accurate positioning of underground pipelines according to claim 7, characterized in that: The vector at the center position of the ultrasonic received signal fine-grained matching search window is the ultrasonic received signal positioning matching library feature vector, and each ultrasonic received signal time-frequency feature vector in the ultrasonic received signal fine-grained matching search window is defined as an ultrasonic received signal fine-grained query time-frequency feature vector.

9. The ultrasonic detection method for accurate positioning of underground pipelines according to claim 8, characterized in that: The ultrasonic received signal kurtosis distribution feature vector and each feature vector in the ultrasonic received signal fine-grained matching search window are response-encoded to obtain the ultrasonic received signal time-frequency query response encoding vector, including: linearly transforming the ultrasonic received signal kurtosis distribution feature vector to obtain an ultrasonic received signal kurtosis distribution query vector and an ultrasonic received signal kurtosis distribution value vector, and using each ultrasonic received signal fine-grained query time-frequency feature vector in the ultrasonic received signal fine-grained matching search window as a key vector, performing fine-grained query encoding based on a heterogeneous converter on the ultrasonic received signal kurtosis distribution query vector, the ultrasonic received signal kurtosis distribution value vector and the key vector to obtain the ultrasonic received signal time-frequency query response encoding vector.

10. The ultrasonic detection method for accurate positioning of underground pipelines according to claim 9, characterized in that: Based on the ultrasonic receiving signal time-frequency query response coding vector, the pipeline location information is obtained, including: passing the ultrasonic receiving signal time-frequency query response coding vector through a decoder-based underground pipeline locator to obtain the underground pipeline locating result, and the underground pipeline locating result includes a pipeline buried depth decoding value as the pipeline location information.

Images