Underwater pipeline detection system and detection data transmission and processing method thereof

Abstract

The invention relates to the technical field of underwater pipeline detection, in particular to an underwater pipeline detection system and a detection data transmission and processing method of the underwater pipeline detection system. The real-time high-definition detection and centimeter-level accurate positioning of the defects of the inner wall of the pipeline are realized; an intelligent processing algorithm based on multi-scale texture analysis and dynamic threshold segmentation can accurately identify and classify abnormal features such as corrosion, cracks and the like; according to the method, an abnormal distribution matrix and a risk diffusion model are constructed, a quantitative evaluation report and a repair priority scheme are automatically generated, an intelligent detection scheme integrating data acquisition, transmission, processing and decision support is finally formed, and the accuracy, efficiency and reliability of underwater pipeline detection are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater pipeline detection, and in particular to an underwater pipeline detection system and a detection data transmission and processing method thereof. Background Art

[0002] Traditional underwater pipeline inspection systems often use towed cameras and cables to transmit video signals. These systems suffer from issues such as bulky and easily corroded cables, signal susceptibility to interference, and low positioning accuracy. Cable entanglement can be particularly problematic in long or complex pipelines, limiting viewing angles and preventing accurate, real-time defect location. Existing technologies rely on copper cables or wireless communications. The former's weight and tensile strength limit towing efficiency, while the latter suffers from water attenuation, leading to image delays or distortion. Positioning often relies on inertial navigation or sonar, making it difficult to achieve centimeter-level accuracy in narrow pipelines. Furthermore, traditional systems lack efficient data fusion mechanisms, requiring video streams and positioning information to be processed independently. This results in defect analysis relying on manual comparison, which is inefficient and prone to overlooking spatial and temporal correlations. To address these issues, a lightweight, highly reliable underwater inspection system with intelligent data processing capabilities is urgently needed. By leveraging multimodal data fusion and adaptive transmission technology, it can achieve real-time pipeline defect location, accurate analysis, and risk prediction, thereby improving inspection efficiency and decision-making reliability. Summary of the Invention

[0003] The present invention overcomes the deficiencies of the prior art and provides an underwater pipeline detection system and a detection data transmission and processing method thereof.

[0004] In order to achieve the above-mentioned purpose, the technical solution adopted by the present invention is: A first aspect of the present invention discloses an underwater pipeline detection system, the detection system comprising a rotating camera, a waterproof meter, a display, a core glass fiber line, and a frame; A wire drum is rotatably mounted on the frame, the core glass fiber wire is wound around the wire drum, and the rotating camera is connected to the movable end of the core glass fiber wire; The rotating camera includes a camera, a transmission gear, a stepping motor, a tail cover, a PC anti-scratch transparent cover, and a radio frequency signal transmitting coil. The camera is embedded in a camera base, and the camera base is covered by a camera cover. The camera base and the transmission gear are connected by a conductive slip ring. The transmission gear and the stepping motor are in mutual transmission connection. The stepper motor is fixed by a motor base, and the end of the motor base is connected to a spring connection upper seat, the spring connection upper seat is fixedly connected to one end of the spring, and the other end of the spring is connected to the spring connection lower seat.

[0005] Furthermore, the rotating camera is provided with several waterproof glue-filled sections at preset positions from the head to the tail, namely glue-filled section A, glue-filled section B, glue-filled section C and glue-filled section D.

[0006] Furthermore, the waterproof meter counter is installed on the frame, and the waterproof meter counter includes a meter counter surface cover, a joint shaft and a meter counter bottom cover. A meter counter PCBA board is installed in the meter counter surface cover, and the meter counter PCBA board is connected to the magnetic induction fixed base. The magnetic induction fixed base is connected to the joint shaft through the meter counter limit plate, and the joint shaft is connected to the meter counter bottom cover through the meter counter bearing. The meter counter bottom cover is also provided with a waterproof aviation head and a waterproof plug.

[0007] Furthermore, the display includes a functional area and a display area, the functional area includes a bottom shell and a face shell, a battery compartment is provided on the base, a waterproof cotton strip is provided on the battery compartment, the base and the face shell can be interlocked with each other, and the interlocking area between the base and the face shell is also provided with a waterproof cotton strip; a silicone button is provided on the base, and a button waterproof pressure plate is mounted on the silicone button.

[0008] Furthermore, the display area includes a first shell and a second shell, the first shell and the second shell can be embedded in each other, and a waterproof cotton strip and a waterproof EVA piece are provided in the mutually embedded area of the first shell and the second shell.

[0009] A second aspect of the present invention discloses a method for transmitting and processing detection data of an underwater pipeline detection system, which is applied to any of the above-mentioned underwater pipeline detection systems and comprises the following steps: Collect real-time video stream data inside the pipeline and simultaneously obtain radio frequency positioning signals. Perform multi-channel encoding on the video stream data and radio frequency signals and transmit them to the ground processing terminal. Perform texture feature analysis on the received video stream data, extract abnormal areas based on the texture features of the inner wall of the pipeline, and generate key feature information containing abnormal location markers; Combined with the cable retraction and extension length data, a three-dimensional spatial coordinate system is constructed in the pipeline, and key feature information is dynamically bound to the corresponding spatial coordinates to generate comprehensive detection information; Comprehensive detection information is stored in a structured manner, and a detection report is generated based on the distribution density and spatial correlation of abnormal areas, with recommended repair priorities and potential risk levels marked.

[0010] Furthermore, real-time video stream data inside the pipeline is collected, and radio frequency positioning signals are obtained synchronously. The video stream data and radio frequency signals are multi-channel encoded and transmitted to the ground processing terminal. Specifically: The real-time video stream data inside the pipeline is collected by a rotating camera and the radio frequency positioning signal of the rotating camera is collected synchronously; Dividing the real-time video stream data into continuous video frame groups according to a preset duration, and extracting the initial timestamp and radio frequency signal phase offset of each video frame group; Compress each video frame group into a compressed data block, and convert the RF phase offset into a RF frequency domain feature vector, which is then bound to the corresponding compressed data block into a multi-channel data packet; Real-time analysis of the signal attenuation characteristics of the RF positioning signal, construction of a phase equalization matrix, phase pre-correction of the RF frequency domain feature vectors in multi-channel data packets, and dynamic allocation of transmission bandwidth weights for video data blocks; The phase-equalized data packets are input into a cross-interleaved error correction encoder, which generates an error correction code sequence according to the priority of the video data, and the sequence is divided into redundant sub-packets, which are transmitted in parallel through independent cores of the core glass fiber line; The ground processing terminal reassembles the redundant sub-packets according to the initial timestamp, repairs the transmission loss data using the error correction code sequence, and reversely corrects the RF frequency domain eigenvector based on the phase equalization matrix to obtain a spatiotemporally synchronized video stream and RF positioning signal.

[0011] Furthermore, texture feature analysis is performed on the received video stream data, and abnormal areas are extracted based on the texture features of the inner wall of the pipe, generating key feature information containing abnormal position markers, specifically: The video stream is preprocessed frame by frame with non-uniform illumination correction and grayscale equalization. Based on the prior texture characteristics of the pipe inner wall material, a multi-scale texture gradient map is generated through a multi-directional Gabor filter bank to extract the gradient amplitude and directional distribution characteristics. The multi-scale texture gradient map is processed based on an adaptive region growing algorithm in combination with a gradient amplitude mutation threshold and a direction consistency constraint condition to segment abnormal regions and mark the boundary coordinates and area proportions of the abnormal regions; Based on the boundary coordinates of the abnormal area, its spatiotemporal evolution trajectory in continuous video frames is extracted, and a morphological feature model based on three-dimensional curvature changes and surface roughness is constructed to distinguish the geometric parameter differences between corrosion areas and cracks; Based on the geometric parameters of the morphological feature model, dynamic segmentation thresholds are set for corrosion areas and cracks respectively. The accuracy of anomaly boundaries is optimized through a genetic algorithm to generate a refined segmentation mask containing anomaly type labels. The segmentation mask and the spatial coordinates of the RF positioning signal are aligned in time and space. Based on the geometric parameters and evolution trajectory of the abnormal area, key feature information including location, type, size and risk level is generated and embedded in the metadata channel of the video stream.

[0012] Furthermore, combined with the cable retraction and extension length data, a three-dimensional spatial coordinate system is constructed in the pipeline, and the key feature information is dynamically bound to the corresponding spatial coordinates to generate comprehensive detection information, specifically: Based on the phase change rate of the RF positioning signal and the interval distribution of the video stream timestamp, the non-uniformly sampled RF signal and video frame sequence are aligned to generate a set of spatiotemporal synchronization reference points; Between the synchronous reference points, the non-uniform sampling cubic spline interpolation method is used to reconstruct the continuous phase trajectory of the radio frequency signal. Combined with the pulse counting data of the cable retraction and extension length, the pipeline motion trajectory vector containing the axial displacement and radial offset is generated. The motion trajectory vector is mapped to the preset geometric model of the pipeline, and a three-dimensional spatial coordinate system based on the pipeline centerline is constructed based on Cartesian coordinate transformation. The spatial coordinate label corresponding to each video frame is output; The spatial distribution of key feature information is mapped to a three-dimensional coordinate system. The dynamic spatial hash grid algorithm is used to establish the nearest neighbor relationship between feature points and coordinate labels, and the radial offset weight coefficient of the feature points relative to the inner wall of the pipe is calculated. Key feature information is spatially topologically sorted according to the radial offset weight coefficient to generate comprehensive detection information including spatiotemporal correlation, anomaly distribution density and risk diffusion trend, which is then encoded into an incremental pipeline topology fingerprint embedded in the video stream metadata.

[0013] Furthermore, the comprehensive detection information is stored in a structured manner, and a detection report is generated based on the distribution density and spatial correlation of the abnormal area, with the recommended repair priority and potential risk level marked, specifically: Mapping the anomaly location, type, and spatial coordinate labels in the comprehensive detection information to the pipeline 3D grid model, generating an anomaly distribution matrix containing spatiotemporal correlation weights, and recording the anomaly density, type weight, and topological connection relationship of adjacent nodes for each grid node; Based on the anomaly distribution matrix, the anomaly density value of each preset sub-region in the pipeline is obtained, and the preset sub-region with an anomaly density value higher than the preset density value is marked as a high-density anomaly region; The potential diffusion paths of defects in high-density abnormal areas along the axial and radial directions of the pipeline are determined by combining the material properties of the pipeline in the high-density abnormal areas. The repair urgency scoring function is defined according to the diffusion path length of the potential diffusion path, and a multi-level repair priority sequence is generated through the Pareto frontier optimization algorithm combined with engineering constraints. Based on the intersection probability of the anomaly type weight and the potential diffusion path, the composite risk index of each anomaly area is determined and divided into three risk level labels: critical risk, high risk and observation level according to the preset threshold; The repair priority sequence, risk level labels and original anomaly distribution matrix are encoded into an incremental hierarchical storage format according to the pipeline topology structure, generating a dynamically updateable inspection report that supports rapid retrieval and visual backtracking by spatial coordinates or risk level.

[0014] The present invention solves the technical defects existing in the background technology and has the following beneficial effects: the underwater pipeline detection system and its detection data transmission and processing method provided by the present invention, through the coordinated work of the rotating camera and the radio frequency positioning signal, combined with the core glass fiber multi-channel transmission and three-dimensional spatial coordinate mapping technology, realize real-time high-definition detection and centimeter-level precise positioning of pipeline inner wall defects; the intelligent processing algorithm based on multi-scale texture analysis and dynamic threshold segmentation can accurately identify and classify abnormal features such as corrosion and cracks; by constructing an abnormal distribution matrix and a risk diffusion model, a quantitative assessment report and a repair priority plan are automatically generated, and finally an intelligent detection solution integrating data acquisition, transmission, processing and decision support is formed, thereby improving the accuracy, efficiency and reliability of underwater pipeline detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, without paying any creative work, they can also obtain drawings of other embodiments based on these drawings.

[0016] Figure 1 This is a schematic diagram of the first overall structure of the detection system; Figure 2 This is a second overall structural diagram of the detection system; Figure 3 This is a schematic diagram of the explosion structure of the rotating camera in this detection system; Figure 4 This is a schematic diagram of the explosion structure of the waterproof meter counter in this detection system; Figure 5 This is a schematic diagram of the explosion structure of the display functional area in this detection system; Figure 6 This is a schematic diagram of the explosion structure of the display area of the detection system; Figure 7 This is a schematic diagram of the structure of the wire reel in this detection system. DETAILED DESCRIPTION

[0017] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present application and the features therein can be combined with each other.

[0018] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0019] like Figure 1 As shown, the first aspect of the present invention discloses an underwater pipeline detection system, which includes a rotating camera 1, a waterproof meter 2, a display 3, a core glass fiber line 4 and a frame 5; like Figure 2 、 Figure 7 As shown, a wire drum 101 is rotatably mounted on the frame, the core glass fiber wire 4 is wound on the wire drum 101, and the rotating camera 1 is connected to the movable end of the core glass fiber wire 4; the wire drum 101 is rotatably mounted on the frame 5 through a wire drum rotating bearing 102; a stainless steel wire clip 103 and an aviation head waterproof connector 104 are also provided on the wire drum.

[0020] It should be noted that the cable reel mechanism of this underwater pipeline inspection system utilizes a cable reel rotating bearing 102 to achieve flexible rotation, allowing the core glass fiber cable 4 wound thereon to be smoothly retracted and extended. A stainless steel cable clamp 103 secures the cable end to prevent loosening, and an aviation-head waterproof connector 104 ensures waterproof and reliable power and signal transmission. The operating principle is as follows: the cable reel 101 rotates freely on the frame 5, and the bearing 102 reduces frictional resistance, controlling the retracted and extended length of the core glass fiber cable 4 to adjust the detection position of the rotating camera 1. Simultaneously, the waterproof connector 104 maintains a tight seal on the line. The bearing 102 ensures smooth rotation, the stainless steel cable clamp 103 provides mechanical fixation, and the waterproof connector 104 enables a secure electrical connection in underwater environments, collectively ensuring reliable operation of the system during underwater pipeline inspection.

[0021] like Figure 3 As shown, the rotating camera includes a camera 105, a transmission gear 106, a stepping motor 107, a tail cover 108, a PC anti-scratch transparent cover 109 and a radio frequency signal transmitting coil 110. The camera 105 is embedded and mounted on a camera base 201, and the camera base is covered by a camera cover 202; the camera base and the transmission gear 106 are connected by a conductive slip ring 203; the transmission gear and the stepping motor are mutually connected in transmission; The stepper motor is fixed by a motor base 204, the end of the motor base is connected to a spring connection upper seat 205, the spring connection upper seat is fixedly connected to one end of a spring 206, and the other end of the spring is connected to a spring connection lower seat 207; the rotating camera is provided with several waterproof glue-filled sections at preset positions from the head to the tail, namely, glue-filled section A 208, glue-filled section B 209, glue-filled section C 301 and glue-filled section D 302.

[0022] It should be noted that the stepper motor 107 provides precise rotational power, the transmission gear 106 transmits torque, the conductive slip ring 203 solves the winding problem, the waterproof glue-filled structure ensures the reliability of underwater operations, and the PC anti-scratch transparent cover 109 protects the lens, together realizing no-dead-angle detection inside the pipeline and long-term stable underwater operation.

[0023] A waterproof pressure plate 303 is provided between the tail cover 108 and the glue-filled D section 302; several waterproof rings 304, ejector guide plates 305, ejector plates 306 and Wuling annular pressure plates 307 are provided at preset positions from the head to the tail of the rotating camera; an anti-tension steel wire 308 and a 5-core connecting wire 309 are provided in the spring; the RF signal transmitting coil 110 is installed on a 512Hz PCBA board 401; a light board 402 is also provided on the camera base 201; and an anti-collision guardrail 403 is provided on the outside of the PC anti-scratch transparent cover 109.

[0024] It should be noted that the rotating camera forms a multi-layered sealing barrier with a waterproof pressure plate 303 and a multi-layer waterproof ring 304, ensuring reliable protection from high-pressure underwater environments. The ejector guide plate 305 and ejector plate 306 precisely position internal components, while the Wuling annular pressure plate 307 provides structural reinforcement. The anti-strain wire 308 enhances the tensile strength of the spring 206 and protects the internal five-core connecting cable 309. The 512Hz RF signal transmitting coil 110 enables precise positioning, the light board 402 provides underwater illumination, and the anti-collision guardrail 403 protects the PC anti-scratch transparent cover 109 from impact. These components work together to ensure the camera's waterproof and pressure resistance, while also ensuring signal transmission stability and structural strength, enabling the system to achieve long-term and stable pipeline inspection operations in complex underwater environments.

[0025] A camera protective cover 404 is mounted on the outer sides of the camera base 201 and the camera cover 202 ; eight protective cover wheels 405 are mounted on the surface of the camera protective cover 404 .

[0026] It should be noted that the underwater pipeline inspection system uses a core glass fiber cable 4 to deliver a rotating camera 1 into the pipeline. Camera 105, driven by a stepper motor 107 and transmission gear 106, achieves 360-degree rotation and captures video. The video signal is transmitted in real time to a ground display 3 via the fiber cable. A 512Hz RF signal transmitting coil 110 cooperates with ground detection equipment to achieve precise positioning, while multiple layers of waterproof glue, a waterproof ring 304, and an anti-collision guardrail 403 ensure the equipment's sealing and impact resistance in high-pressure underwater environments. As can be seen, by replacing traditional cables with lightweight fiber cables, drag resistance is reduced; conductive slip rings 203 prevent cable entanglement during rotation; springs 206 and anti-tension wires 308 buffer tension and protect cables; and anti-collision wheels 405 improve pipeline passability, making it suitable for efficient inspection and fault location in complex industrial pipelines.

[0027] like Figure 4 As shown, the waterproof meter counter 2 is installed on the frame 5, and the waterproof meter counter includes a meter counter surface cover 406, a joint shaft 407 and a meter counter bottom cover 408. A meter counter PCBA board 409 is installed in the meter counter surface cover, and the meter counter PCBA board is connected to the magnetic induction fixed base 501. The magnetic induction fixed base 501 is connected to the joint shaft 407 through the meter counter limiting plate 503. The joint shaft is connected to the meter counter bottom cover 408 through the meter counter bearing 505. The meter counter bottom cover is also provided with a waterproof aviation head 506 and a waterproof plug 507; the meter counter bearing 505 is provided with a bearing limiting retaining spring 508.

[0028] It should be noted that the waterproof meter counter 2, through its magnetically fixed base 501 and meter PCBA board 409, monitors the retracted and extended length of the core glass fiber cable 4 in real time. The connector shaft 407 rotates flexibly via the meter counter bearing 505, with the meter counter limiter 503 controlling its displacement. A waterproof aviation connector 506 and waterproof plug 507 ensure a tight seal in the underwater environment, while a bearing limit spring 508 prevents the bearing from dislocating. The waterproof meter counter 2 accurately records cable length to locate fault points within the pipeline. Its waterproof design adapts to humid or high-pressure environments, while the bearing structure ensures smooth rotation. The overall structure is compact and reliable, providing accurate length reference data for underwater pipeline inspections.

[0029] like Figure 5 、 Figure 6As shown, the display includes a functional area 6 and a display area 7. The functional area includes a bottom shell 509 and a top shell 601. A battery compartment 602 is provided on the base, and a waterproof cotton strip 603 is provided on the battery compartment. The base and the top shell can be interlocked with each other, and the interlocking area between the base and the top shell is also provided with a waterproof cotton strip 603; a silicone button 604 is provided on the base, and a button waterproof pressure plate 605 is mounted on the silicone button 604; a battery is provided in the battery compartment; a silicone anti-collision pad 606 is provided at the bottom of the bottom shell 509; and a waterproof aviation head 506, a charging seat 607, a rotating device cover 608, a webbing hanging ring 609 and a control PCBA board 701 are provided on the top shell 601.

[0030] The display area comprises a first housing 702 and a second housing 703, which fit together. A waterproof cotton strip 603 and a waterproof EVA member 704 are located in the interlocking region. The second housing 703 is equipped with a sunshade aluminum plate 705, front and rear rotation damping shafts 706, an acrylic display panel 707, a 9-inch display 708, and a display driver board PCBA 709.

[0031] The display consists of a function area 6 and a display area 7. The function area features waterproof operation via silicone buttons 604 and a waterproof pressure plate 605. A battery compartment 602 and waterproof strips 603 at the interface ensure a tight seal in humid environments. The display area features a 9-inch display 708 and an aluminum sunshade 705 for enhanced outdoor visibility. A rotating damping shaft 706 supports multiple angles. The waterproof structure (waterproof strips 603 and EVA components 704) ensures reliable underwater operation. A silicone anti-collision pad 606 cushions impacts, while a rotating damping shaft 706 facilitates viewing angle adjustment. The overall design balances waterproofness, durability, and ease of operation, making it suitable for real-time monitoring in complex environments.

[0032] A second aspect of the present invention discloses a method for transmitting and processing detection data of an underwater pipeline detection system, which is applied to any of the above-mentioned underwater pipeline detection systems and comprises the following steps: Collect real-time video stream data inside the pipeline and simultaneously obtain radio frequency positioning signals. Perform multi-channel encoding on the video stream data and radio frequency signals and transmit them to the ground processing terminal. Perform texture feature analysis on the received video stream data, extract abnormal areas (such as cracks and corrosion points) based on the texture features of the pipeline inner wall, and generate key feature information containing abnormal location markers; Combined with the cable retraction and extension length data, a three-dimensional spatial coordinate system is constructed in the pipeline, and key feature information is dynamically bound to the corresponding spatial coordinates to generate comprehensive detection information; Comprehensive detection information is stored in a structured manner, and a detection report is generated based on the distribution density and spatial correlation of abnormal areas, with recommended repair priorities and potential risk levels marked.

[0033] This method addresses the technical challenges of asynchronous transmission between video data and positioning signals, low defect recognition efficiency, and a lack of spatial correlation in underwater pipeline inspection results. By synchronizing data through multi-channel coded transmission, combined with texture feature analysis and 3D coordinate binding, it enables real-time and accurate positioning of pipeline defects, automatic identification of abnormal areas, and generation of structured inspection reports including risk levels, effectively improving inspection efficiency and decision-making reliability.

[0034] Furthermore, real-time video stream data inside the pipeline is collected, and radio frequency positioning signals are obtained synchronously. The video stream data and radio frequency signals are multi-channel encoded and transmitted to the ground processing terminal. Specifically: The real-time video stream data inside the pipeline is collected by a rotating camera and the radio frequency positioning signal of the rotating camera is collected synchronously; Dividing the real-time video stream data into continuous video frame groups according to a preset duration, and extracting the initial timestamp and radio frequency signal phase offset of each video frame group; Compress each video frame group into a compressed data block, and convert the RF phase offset into a RF frequency domain feature vector, which is then bound to the corresponding compressed data block into a multi-channel data packet; Real-time analysis of the signal attenuation characteristics of the RF positioning signal, construction of a phase equalization matrix, phase pre-correction of the RF frequency domain feature vectors in multi-channel data packets, and dynamic allocation of transmission bandwidth weights for video data blocks; The RF signal receiving module monitors the amplitude-frequency response curve of the 512Hz positioning signal during transmission through the core glass fiber cable in real time, obtaining the phase offset of each subcarrier frequency (e.g., -22° at 1.5MHz). A 4×4 complex phase equalization matrix is generated using least-squares fitting, and linear transformation compensation is performed on the real and imaginary components of the 128-dimensional RF frequency-domain feature vector in the data packet. Furthermore, the transmission bandwidth allocation of each data block is adjusted based on the I / P frame type of the video data block (e.g., I-frames are assigned a weight of 0.8) and the current channel signal-to-noise ratio (e.g., a 10% bandwidth increase when the SNR is greater than 30dB), ensuring that key video frames receive priority transmission. The compensated frequency-domain feature vectors and video data packets are then output to the next processing module via TDM time division multiplexing.

[0035] The phase-equalized data packets are input into a cross-interleaved error correction encoder, which generates an error correction code sequence according to the priority of the video data, and the sequence is divided into redundant sub-packets, which are transmitted in parallel through independent cores of the core glass fiber line; The ground processing terminal reassembles the redundant sub-packets according to the initial timestamp, repairs the transmission loss data using the error correction code sequence, and reversely corrects the RF frequency domain eigenvector based on the phase equalization matrix to obtain a spatiotemporally synchronized video stream and RF positioning signal.

[0036] For example, a rotating camera captures a 1080P video stream of the pipeline's inner wall at a rate of 30 frames per second, while simultaneously acquiring positioning signals through a 512Hz RF transmitter coil. The video stream is segmented into 15-frame groups every 0.5 seconds, and the start timestamp of each group and the corresponding RF signal phase offset (e.g., ±15°) are recorded. H.265 encoding is used to compress the video frames into data blocks averaging 2MB. The RF phase offset is converted into a 128-dimensional RF frequency-domain feature vector via a fast Fourier transform (FFT). The two are then combined to form a composite data packet. During transmission, the signal attenuation curve (e.g., -3dB / m) of the core glass fiber line in the 5MHz frequency band is monitored in real time. A 4×4 phase equalization matrix is constructed to pre-compensate the RF feature vector and dynamically allocate 70% of the transmission bandwidth for the video data. A 32-byte error correction code sequence generated using (255,223) Reed-Solomon encoding is segmented into four redundant sub-packets for parallel transmission over four optical fibers. The ground terminal reassembles the sub-packets according to the timestamp, uses error correction codes to repair data loss, and restores the RF signal through inverse matrix operations, ultimately outputting synchronized data with a time deviation of less than 1ms.

[0037] In summary, this method can achieve the direct beneficial effects of high-fidelity synchronous transmission of video and positioning signals, strong anti-interference ability, and high data integrity through multi-channel data binding, phase pre-correction, and redundant parallel transmission, ensuring that the ground terminal obtains accurate and reliable detection data.

[0038] Furthermore, texture feature analysis is performed on the received video stream data, and abnormal areas are extracted based on the texture features of the inner wall of the pipe, generating key feature information containing abnormal position markers, specifically: The video stream is preprocessed frame by frame with non-uniform illumination correction and grayscale equalization. Based on the prior texture characteristics of the pipe inner wall material, a multi-scale texture gradient map is generated through a multi-directional Gabor filter bank to extract the gradient amplitude and directional distribution characteristics. Among them, the prior texture features of the pipeline inner wall material refer to the benchmark texture model library for subsequent defect detection and comparison established by pre-collecting and analyzing the standard surface texture data of the inner wall of the pipeline of a specific material (such as metal particle distribution, weld texture, oxide layer morphology, etc.).

[0039] The multi-scale texture gradient map is processed based on an adaptive region growing algorithm in combination with a gradient amplitude mutation threshold and a direction consistency constraint condition to segment abnormal regions and mark the boundary coordinates and area proportions of the abnormal regions; It should be noted that the adaptive region growing algorithm achieves abnormal region segmentation through the following method: first, pixels in the multi-scale texture gradient map whose gradient amplitude exceeds a preset mutation threshold (e.g., 35 grayscale levels) are used as seed points. Second, a local directional consistency index is calculated based on the gradient direction angles of pixels surrounding the seed point. When the directional deviation of adjacent pixels is less than a set tolerance (e.g., ±15°), the region is identified as homogeneous and included in the growing range. During the growing process, the amplitude threshold and directional tolerance are dynamically adjusted, with the amplitude threshold exponentially decaying as the region area increases, and the directional tolerance is linearly compensated based on the complexity of the regional morphology. Finally, an abnormal region is output when the region grows to meet the following termination conditions: 1. The mean gradient amplitude of the region's edge pixels falls below 60% of the initial threshold; 2. The variance of the directional consistency index within the region exceeds a preset upper limit (e.g., 0.25). This process effectively distinguishes the background texture of the pipeline inner wall from the actual defect area by balancing the gradient amplitude mutation with the directional continuity constraint.

[0040] Based on the boundary coordinates of the abnormal area, its spatiotemporal evolution trajectory in continuous video frames is extracted, and a morphological feature model based on three-dimensional curvature changes and surface roughness is constructed to distinguish the geometric parameter differences between corrosion areas (continuous curvature distribution) and cracks (local sudden changes in curvature). Based on the geometric parameters of the morphological feature model, dynamic segmentation thresholds are set for corrosion areas and cracks respectively. The accuracy of anomaly boundaries is optimized through a genetic algorithm to generate a refined segmentation mask containing anomaly type labels (corrosion / cracks). It should be noted that, first, based on the morphological feature model, the average curvature radius of the corrosion area (>100mm) and the local curvature extreme value of the crack area (<20mm) were extracted as initial classification features. Secondly, a linear threshold function with surface roughness (Ra value 0.8-1.2μm) as the independent variable was established for the corrosion area, and a nonlinear threshold mapping table based on the curvature change rate (>5% / mm) was constructed for the crack area. Subsequently, a genetic algorithm was used for optimization, in which the population was set to 20 groups of threshold parameter combinations, and the fitness function comprehensively considered boundary continuity (requiring the proportion of closed contours to be >95%) and feature discrimination (inter-class difference >30%). After 15-20 generations of iteration, the optimal parameter combination was selected to generate a binary mask, and discrete noise was eliminated through a morphological closing operation (3×3 structuring element). Finally, a refined segmentation result with a type label (corrosion / crack) was output, where the corrosion area was marked with a grayscale value of 150 and the crack area was marked with a grayscale value of 250.

[0041] The segmentation mask and the spatial coordinates of the RF positioning signal are aligned in time and space. Based on the geometric parameters and evolution trajectory of the abnormal area, key feature information including location, type, size and risk level is generated and embedded in the metadata channel of the video stream.

[0042] For example, a 1080P video stream (30 frames per second) captured by a rotating camera first undergoes illumination correction based on the Retinex model, balancing the grayscale values of the pipe's inner wall to a range of 50-200. Each frame is then processed using a Gabor filter bank with eight orientations (0°-157.5°, intervals of 22.5°) and three scales (wavelengths of 16 / 32 / 64 pixels), generating a texture feature map consisting of gradient magnitude (0-255) and orientation angle (0-360°). For the inner wall of a carbon steel pipe, a gradient mutation threshold of 35 grayscale levels and a directional consistency tolerance of ±15° are set. Anomalous regions larger than 50 pixels are extracted using a region growing algorithm, and the vertex coordinates of the bounding polygons are output. The motion of the anomalous regions is tracked over 10 consecutive video frames, and the curvature radius (corrosion areas >100 mm, crack areas <20 mm) and surface roughness Ra values (corrosion areas 0.8-1.2 μm, crack areas >2.5 μm) are calculated to construct a 3D morphological database. A dynamic segmentation threshold (grayscale difference threshold of 25±5 for corrosion areas and 40±10 for crack areas) is set according to the material type (e.g., X70 steel). After 20 generations of genetic iterative optimization, a binary mask with a precision of 0.5 mm is generated. Finally, the abnormal information (e.g., crack @ 3.2 mm × 15 mm, coordinates X125.6 m / Y-30°Z+5 mm) is written into the SEI metadata area of the H.264 video stream.

[0043] In summary, this method can realize automatic and accurate identification and classification of pipeline inner wall defects through multi-scale texture analysis and dynamic threshold segmentation, accurately distinguish abnormal features such as corrosion and cracks, and generate structured feature information including location, type and risk level, thereby improving detection accuracy and analysis efficiency.

[0044] Furthermore, combined with the cable retraction and extension length data, a three-dimensional spatial coordinate system is constructed in the pipeline, and the key feature information is dynamically bound to the corresponding spatial coordinates to generate comprehensive detection information, specifically: Based on the phase change rate of the RF positioning signal and the interval distribution of the video stream timestamp, the non-uniformly sampled RF signal and video frame sequence are aligned to generate a set of spatiotemporal synchronization reference points; Between the synchronous reference points, the non-uniform sampling cubic spline interpolation method is used to reconstruct the continuous phase trajectory of the radio frequency signal. Combined with the pulse counting data of the cable retraction and extension length, the pipeline motion trajectory vector containing the axial displacement and radial offset is generated. The motion trajectory vector is mapped to the preset geometric model of the pipeline (such as a circular / rectangular cross-section), and a three-dimensional spatial coordinate system based on the pipeline centerline is constructed based on Cartesian coordinate transformation. The spatial coordinate label corresponding to each video frame is output; The spatial distribution of key feature information (abnormal location and type) is mapped to a three-dimensional coordinate system. The dynamic spatial hash grid algorithm is used to establish the nearest neighbor relationship between feature points and coordinate labels, and the radial offset weight coefficient of the feature points relative to the inner wall of the pipeline is calculated. The calculation formula of the radial offset weight coefficient is: ; Where, is the radial offset weight coefficient of the i-th feature point; is the distance from the i-th feature point to the inner wall of the pipe (unit: mm); The theoretical radius of the inner wall of the pipe (e.g. 300mm for a DN600 pipe); is the critical radius for risk determination (usually r min +20mm).

[0045] Key feature information is spatially topologically sorted according to the radial offset weight coefficient to generate comprehensive detection information including spatiotemporal correlation, anomaly distribution density and risk diffusion trend, which is then encoded into an incremental pipeline topology fingerprint embedded in the video stream metadata.

[0046] For example, the system uses a 512Hz RF signal (phase resolution 0.1°) and a 30fps video stream as input. Within every 5-second interval, the video frame timestamp is aligned with the RF signal zero crossing, generating a set of 50-60 synchronized reference points. For intervals of approximately 150ms between adjacent reference points, a cubic spline curve is used to fit the RF phase trajectory. Combined with cable encoder data with 200 pulses per meter, this generates a motion trajectory vector with an axial positioning accuracy of ±2cm and a radial offset accuracy of ±5mm. For a DN600 circular pipe, the trajectory vector is converted to a three-dimensional coordinate system with the pipe center as the origin (0,0,0), with an axial Z coordinate accuracy of 0.1% and a circumferential θ coordinate accuracy of 0.5°. Anomalous feature points (such as corrosion areas) are mapped to their nearest coordinate labels using a hash grid (cell size 50mm x 10°). An offset weight (normalized value from 0 to 1) is then determined from the inner wall. The resulting comprehensive inspection information includes: anomaly density distribution maps for every 10cm pipe segment; a list of risk hotspots sorted by weight coefficient; and a topological fingerprint data block (4KB of metadata embedded in each MB of video stream) that can be updated as the inspection progresses. In actual implementation, pipeline geometry and sampling frequency can be adjusted based on inspection requirements.

[0047] In summary, this method addresses the technical challenges of underwater pipeline inspection, such as the difficulty in accurately matching video data with spatial location information and inaccurate defect location. By synchronizing the time and space between RF signals and cable data, and mapping their three-dimensional coordinates, it achieves centimeter-level precision in locating internal pipeline defects. It also automatically correlates defect characteristics with spatial location, ultimately generating a structured inspection report containing comprehensive information, including defect distribution and risk trends.

[0048] Furthermore, the comprehensive detection information is stored in a structured manner, and a detection report is generated based on the distribution density and spatial correlation of the abnormal area, with the recommended repair priority and potential risk level marked, specifically: Mapping the anomaly location, type, and spatial coordinate labels in the comprehensive detection information to the pipeline 3D grid model, generating an anomaly distribution matrix containing spatiotemporal correlation weights, and recording the anomaly density, type weight, and topological connection relationship of adjacent nodes for each grid node; Based on the anomaly distribution matrix, the anomaly density value of each preset sub-region in the pipeline is obtained, and the preset sub-region with an anomaly density value higher than the preset density value is marked as a high-density anomaly region; Combined with the pipeline material properties (such as metal fatigue coefficient and corrosion rate) in the high-density abnormal area, the potential diffusion paths of defects in the high-density abnormal area along the pipeline axial and radial directions are determined; A repair urgency scoring function is defined based on the diffusion path length of potential diffusion paths, and a multi-level repair priority sequence is generated using the Pareto frontier optimization algorithm in combination with engineering constraints (such as repair cost and construction accessibility). It should be noted that a urgency scoring model was first established, with diffusion path length as the core parameter. Each 1-meter increase in axial diffusion distance corresponds to a 20-point increase in the base score, while radial diffusion to adjacent grid cells adds 10 points. A multi-objective optimization system, incorporating engineering constraints, was constructed, with repair cost (10,000 yuan), construction accessibility (standardized on a scale of 0-1), and pipeline criticality (e.g., a weight of 1.5 for the main line) as parallel optimization objectives. A non-dominated sorting algorithm was used to identify a Pareto-optimal set of solutions, each corresponding to a set of weighted options (e.g., a weight of 0.4 for cost, 0.3 for accessibility, and 0.3 for criticality). Finally, the weighted total score of each option in the set was used to prioritize the solutions into three levels: Level 1 (>80 points) requires resolution within 24 hours, Level 2 (60-80 points) requires resolution within 72 hours, and Level 3 (<60 points) is included in the annual maintenance plan. By quantifying the coupling relationship between diffusion risk and engineering practice, a standardized translation from technical parameters to decision-making solutions was achieved.

[0049] Based on the intersection probability of the anomaly type weight (such as crack depth and corrosion area ratio) and the potential diffusion path, the composite risk index of each anomaly area is determined and classified into three risk level labels: critical risk, high risk, and observation level according to the preset threshold; The repair priority sequence, risk level labels and original anomaly distribution matrix are encoded into an incremental hierarchical storage format according to the pipeline topology structure, generating a dynamically updateable inspection report that supports rapid retrieval and visual backtracking by spatial coordinates or risk level.

[0050] For example, for a DN800 steel pipeline, the pipeline is divided into 100mm×100mm×500mm (axial) grid units. Each grid node records: (1) abnormal density value (0-10 levels, such as 5 corrosion points / m² is recorded as density value 7); (2) type weight coefficient (crack 0.8, corrosion 0.5); 3) connection relationship with the adjacent 6 nodes. The density threshold is set to 6, and a high-density area consisting of more than 3 consecutive excessive grids is marked (such as the pipe section 35.2-36.7m from the starting point). Based on the steel properties (fatigue coefficient 0.85, corrosion rate 0.2mm / year), the axial diffusion rate of the defect is determined (crack 2.1m / year, corrosion 0.8m / year) to generate a diffusion path heat map (red high-risk area diffusion distance >1m). The repair priority is divided into 0-100 points: defects with a construction window period of less than 7 days and cross-diffusion are scored 90 points (level 1), and defects with a cost exceeding the budget by 20% are scored 60 points (level 3). Risk levels are categorized by composite index: >85 is critical (red), 60-85 is high (yellow), and all others are observation (green). The final report is stored in a layered JSON format, with the base layer containing the original grid data (approximately 2MB / km) and the application layer containing optimized decision-making data (200kB / km). GIS systems can access 3D models in real time by coordinates (e.g., X=35.6m, Y=120°) or risk level.

[0051] In summary, this method automatically generates an objective and quantitative repair priority sequence and risk level classification by establishing an abnormality distribution matrix and diffusion path model, realizing intelligent assessment and scientific decision-making of pipeline defects, thereby improving the professionalism of inspection reports and the rationality of maintenance plans.

[0052] In actual application, the detection data transmission and processing method may further include the following steps: The electrochemical impedance spectroscopy (EIS) and differential pulse voltammetry (DPV) signals of the microbial community attached to the pipe wall were collected in real time through a microelectrode array. After wavelet multi-scale decomposition, an electrochemical fingerprint feature vector containing the redox peak position and charge transfer resistance value was generated. The local electronic state density distribution of iron elements on the pipe surface is inverted using synchrotron radiation X-ray absorption near-edge structure (XANES) data, generating a metal lattice defect electronic state distribution map and annotating the energy barrier parameters of the corrosion area; The electrochemical fingerprint feature vector and the lattice defect electronic state distribution map are input into the graph convolutional neural network. Through the topological matching of the electron-ion coupling transmission path, a spatial correlation matrix between the microbial metabolic current intensity and the metal defect active sites is established to quantify the attenuation coefficient of the biofilm catalysis on the corrosion energy barrier. Based on the attenuation coefficient of the spatial correlation matrix, combined with the environmental pH value and dissolved oxygen concentration, the corrosion current density threshold and the probability of passive film rupture at the microorganism-metal interface are determined, and a dynamically updated corrosion energy barrier parameter map is generated. When the corrosion energy barrier parameter is lower than the preset safety threshold, an early warning signal is triggered and the three-dimensional coordinates of the corresponding pipe section are marked. At the same time, the metabolic characteristics of the microbial population that dominates the corrosion are traced based on the spatial correlation matrix, and targeted bioinhibition strategy recommendations are generated.

[0053] For example, the EIS signal (frequency range 0.1Hz-100kHz) and DPV signal (pulse amplitude 50mV) were collected at a scan rate of 10mV / s, and the characteristic values (such as the oxidation peak at 711mV and the charge transfer resistance of 2.8kΩ·cm²) were extracted after decomposition into 5 frequency bands by Db4 wavelet. The iron L-edge XANES spectrum was collected simultaneously using the third-generation synchrotron radiation source (energy 4-15keV), and the surface Fe The 3D electron density of states distribution identifies highly active sites at grain boundaries with energy barriers below 1.2 eV. This data was fed into a three-layer graph convolutional neural network (256-128-64 nodes). The output showed a spatial correlation of 0.92 between the metabolic current of sulfate-reducing bacteria (>0.15 μA / cm²) and grain boundary defects (barriers <0.8 eV). When the ambient pH is <6.5 and the dissolved oxygen is <0.5 ppm, the corrosion current density threshold rises to 1.2 μA / cm², and the probability of passive film rupture is >65%. If the energy barrier parameter of a pipe section (e.g., 35.6 m axial length, 120° circumferential position) remains below the safety threshold of 0.5 eV for three consecutive days, the system triggers a red alert and recommends the application of a 20 ppm molybdate corrosion inhibitor. In practice, the electrode spacing and spectral parameters can be adjusted based on the pipe diameter.

[0054] In summary, this method integrates microbial electrochemical signals with metal lattice defect analysis to correlate biofilm activity with metal corrosion tendency, achieves early warning of microbial corrosion risk and accurately locates high-risk pipe sections, and at the same time provides a scientific basis for targeted protective measures, thereby improving the initiative and accuracy of pipeline corrosion protection.

[0055] In actual application, the detection data transmission and processing method may further include the following steps: The distributed fiber optic acoustic sensor array collects the pressure pulsation signal of the fluid in the pipeline in real time, and combines it with the ultrasonic flowmeter to obtain the flow velocity profile data, and constructs a flow field parameter matrix containing the pressure gradient, vorticity distribution and gas-liquid mixture phase characteristics; Perform intrinsic mode decomposition on the parameter matrix of the abnormal flow field area, extract characteristic frequency components related to leakage (such as 0.5-5Hz low-frequency pressure fluctuations and local vorticity surges), and generate leakage characteristic fingerprints; Based on the vortex core migration trajectory and pressure gradient change direction in the leakage characteristic fingerprint, a spatiotemporal correlation equation between the leakage point location and the downstream flow field distortion area is established, and the linear correlation coefficient between the leakage intensity and the flow field disturbance range is quantified. Based on the correlation coefficient distribution of the spatiotemporal correlation equation and the pipeline topology (elbow and valve positions), the probability weight of each pipe section as a leakage source is determined, and a leakage point probability heat map is generated.

[0056] For example, distributed fiber-optic acoustic sensors (sampling rate 1kHz) are deployed every 50 meters along a DN800 oil pipeline. Together with eight-channel ultrasonic flowmeters (accuracy ±0.5%) installed in the pipeline sections, they collect real-time fluid pressure fluctuations (range 0-2MPa) and velocity profiles (0-5m / s). This allows the system to construct a flow field parameter matrix containing axial pressure gradient (Pa / m), vorticity (1 / s), and gas-liquid mixed phase fraction (0-100%). When low-frequency pressure fluctuations of 0.8-3.2Hz and a sudden increase in local vorticity to 15 / s are detected in a specific pipeline section (e.g., pile number K35+200), empirical mode decomposition (EMD) is used to extract five intrinsic mode functions (IMFs). The leak-related IMF3 component (energy contribution >40%) is selected to generate a signature fingerprint. Based on the downstream migration rate of the vortex core (approximately 0.6 m / s) and the direction of the pressure gradient (30° off-axis) in the map, a correlation equation was established between the leak point location (e.g., K35+180) and the flow field distortion zone 50 meters downstream, resulting in a correlation coefficient of 0.78. Combined with the flow field disturbance characteristics of the upstream elbow (R=3D) in this pipe section, the leak probability weights for each node were determined (reaching 85% at K35+180). This ultimately generated a heat map with red (high probability), yellow (medium probability), and blue (low probability). In actual deployment, sensor spacing can be adjusted based on the viscosity of the medium.

[0057] In summary, by fusing multi-source flow field data with dynamic feature analysis, we can quickly identify the flow field distortion characteristics induced by leakage, establish an accurate association between leakage points and flow field anomalies, and achieve non-contact, high-precision leak tracing and positioning.

[0058] The above are only specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the scope of protection of the present invention.

Claims

1. An underwater pipeline detection system, characterized in that: The detection system includes a rotating camera, a waterproof meter, a display, a core glass fiber line and a frame; A wire drum is rotatably mounted on the frame, the core glass fiber wire is wound around the wire drum, and the rotating camera is connected to the movable end of the core glass fiber wire; The rotating camera includes a camera, a transmission gear, a stepping motor, a tail cover, a PC anti-scratch transparent cover, and a radio frequency signal transmitting coil. The camera is embedded in a camera base, and the camera base is covered by a camera cover. The camera base and the transmission gear are connected by a conductive slip ring. The transmission gear and the stepping motor are in mutual transmission connection. The stepper motor is fixed by a motor base, and the end of the motor base is connected to a spring connection upper seat, the spring connection upper seat is fixedly connected to one end of the spring, and the other end of the spring is connected to the spring connection lower seat.

2. The underwater pipeline detection system according to claim 1, characterized in that: The rotating camera is provided with several waterproof glue-filled sections at preset positions from the head to the tail, namely glue-filled section A, glue-filled section B, glue-filled section C and glue-filled section D.

3. The underwater pipeline detection system according to claim 1, characterized in that: The waterproof meter counter is installed on the frame. The waterproof meter counter includes a meter counter surface cover, a joint shaft and a meter counter bottom cover. A meter counter PCBA board is installed in the meter counter surface cover. The meter counter PCBA board is connected to the magnetic induction fixed base. The magnetic induction fixed base is connected to the joint shaft through the meter counter limit plate. The joint shaft is connected to the meter counter bottom cover through the meter counter bearing. A waterproof aviation head and a waterproof plug are also provided on the meter counter bottom cover.

4. The underwater pipeline detection system according to claim 1, characterized in that: The display includes a functional area and a display area. The functional area includes a bottom shell and a top shell. A battery compartment is provided on the base. A waterproof cotton strip is provided on the battery compartment. The base and the top shell can be interlocked with each other. The interlocking area between the base and the top shell is also provided with a waterproof cotton strip. A silicone button is provided on the base, and a button waterproof pressure plate is mounted on the silicone button.

5. The underwater pipeline detection system according to claim 4, characterized in that: The display area includes a first shell and a second shell. The first shell and the second shell can be embedded with each other. A waterproof cotton strip and a waterproof EVA piece are provided in the mutually embedded area of the first shell and the second shell.

6. A method for transmitting and processing detection data of an underwater pipeline detection system, applied to an underwater pipeline detection system according to any one of claims 1 to 5, characterized in that: The following steps are involved: Collect real-time video stream data inside the pipeline and simultaneously obtain radio frequency positioning signals. Perform multi-channel encoding on the video stream data and radio frequency signals and transmit them to the ground processing terminal. Perform texture feature analysis on the received video stream data, extract abnormal areas based on the texture features of the inner wall of the pipeline, and generate key feature information containing abnormal location markers; Combined with the cable retraction and extension length data, a three-dimensional spatial coordinate system is constructed in the pipeline, and key feature information is dynamically bound to the corresponding spatial coordinates to generate comprehensive detection information; Comprehensive detection information is stored in a structured manner, and a detection report is generated based on the distribution density and spatial correlation of abnormal areas, with recommended repair priorities and potential risk levels marked.

7. The method for transmitting and processing detection data of an underwater pipeline detection system according to claim 6, characterized in that: Collect real-time video stream data inside the pipeline and synchronously obtain radio frequency positioning signals. Perform multi-channel encoding on the video stream data and radio frequency signals and transmit them to the ground processing terminal. Specifically: The real-time video stream data inside the pipeline is collected by a rotating camera and the radio frequency positioning signal of the rotating camera is collected synchronously; Dividing the real-time video stream data into continuous video frame groups according to a preset duration, and extracting the initial timestamp and radio frequency signal phase offset of each video frame group; Compress each video frame group into a compressed data block, and convert the RF phase offset into a RF frequency domain feature vector, and bind it with the corresponding compressed data block to form a multi-channel data packet; Real-time analysis of the signal attenuation characteristics of the RF positioning signal, construction of a phase equalization matrix, phase pre-correction of the RF frequency domain feature vectors in multi-channel data packets, and dynamic allocation of transmission bandwidth weights for video data blocks; The phase-equalized data packets are input into a cross-interleaved error correction encoder, which generates an error correction code sequence according to the priority of the video data, and the sequence is divided into redundant sub-packets, which are transmitted in parallel through independent cores of the core glass fiber line; The ground processing terminal reassembles the redundant sub-packets according to the initial timestamp, repairs the transmission loss data using the error correction code sequence, and reversely corrects the RF frequency domain eigenvector based on the phase equalization matrix to obtain a spatiotemporally synchronized video stream and RF positioning signal.

8. The method for transmitting and processing detection data of an underwater pipeline detection system according to claim 6, characterized in that: The received video stream data is subjected to texture feature analysis. The abnormal area is extracted based on the texture features of the inner wall of the pipe, and key feature information including abnormal location markers is generated. Specifically: The video stream is preprocessed frame by frame with non-uniform illumination correction and grayscale equalization. Based on the prior texture characteristics of the pipe inner wall material, a multi-scale texture gradient map is generated through a multi-directional Gabor filter bank to extract the gradient amplitude and directional distribution characteristics. The multi-scale texture gradient map is processed based on an adaptive region growing algorithm in combination with a gradient amplitude mutation threshold and a direction consistency constraint condition to segment abnormal regions and mark the boundary coordinates and area proportions of the abnormal regions; Based on the boundary coordinates of the abnormal area, its spatiotemporal evolution trajectory in continuous video frames is extracted, and a morphological feature model based on three-dimensional curvature changes and surface roughness is constructed to distinguish the geometric parameter differences between corrosion areas and cracks; Based on the geometric parameters of the morphological feature model, dynamic segmentation thresholds are set for corrosion areas and cracks respectively. The accuracy of anomaly boundaries is optimized through a genetic algorithm to generate a refined segmentation mask containing anomaly type labels. The segmentation mask and the spatial coordinates of the RF positioning signal are aligned in time and space. Based on the geometric parameters and evolution trajectory of the abnormal area, key feature information including location, type, size and risk level is generated and embedded in the metadata channel of the video stream.

9. The method for transmitting and processing detection data of an underwater pipeline detection system according to claim 6, characterized in that: Combined with the cable retraction and extension length data, a three-dimensional spatial coordinate system is constructed in the pipeline. The key feature information is dynamically bound to the corresponding spatial coordinates to generate comprehensive detection information, specifically: Based on the phase change rate of the RF positioning signal and the interval distribution of the video stream timestamp, the non-uniformly sampled RF signal and video frame sequence are aligned to generate a set of spatiotemporal synchronization reference points; Between the synchronous reference points, the non-uniform sampling cubic spline interpolation method is used to reconstruct the continuous phase trajectory of the radio frequency signal. Combined with the pulse counting data of the cable retraction and extension length, the pipeline motion trajectory vector containing the axial displacement and radial offset is generated. The motion trajectory vector is mapped to the preset geometric model of the pipeline, and a three-dimensional spatial coordinate system based on the pipeline centerline is constructed based on Cartesian coordinate transformation. The spatial coordinate label corresponding to each video frame is output; The spatial distribution of key feature information is mapped to a three-dimensional coordinate system. The dynamic spatial hash grid algorithm is used to establish the nearest neighbor relationship between feature points and coordinate labels, and the radial offset weight coefficient of the feature points relative to the inner wall of the pipe is calculated. Key feature information is spatially topologically sorted according to the radial offset weight coefficient to generate comprehensive detection information including spatiotemporal correlation, anomaly distribution density and risk diffusion trend, which is then encoded into an incremental pipeline topology fingerprint embedded in the video stream metadata.

10. The method for transmitting and processing detection data of an underwater pipeline detection system according to claim 6, characterized in that: Comprehensive detection information is stored in a structured manner. A detection report is generated based on the distribution density and spatial correlation of abnormal areas, and the recommended repair priority and potential risk level are marked. Specifically: Mapping the anomaly location, type, and spatial coordinate labels in the comprehensive detection information to the pipeline 3D grid model, generating an anomaly distribution matrix containing spatiotemporal correlation weights, and recording the anomaly density, type weight, and topological connection relationship of adjacent nodes for each grid node; Based on the anomaly distribution matrix, the anomaly density value of each preset sub-region in the pipeline is obtained, and the preset sub-region with an anomaly density value higher than the preset density value is marked as a high-density anomaly region; The potential diffusion paths of defects in high-density abnormal areas along the axial and radial directions of the pipeline are determined by combining the material properties of the pipeline in the high-density abnormal areas. The repair urgency scoring function is defined according to the diffusion path length of the potential diffusion path, and a multi-level repair priority sequence is generated through the Pareto frontier optimization algorithm combined with engineering constraints. Based on the intersection probability of the anomaly type weight and the potential diffusion path, the composite risk index of each anomaly area is determined and divided into three risk level labels: critical risk, high risk and observation level according to the preset threshold; The repair priority sequence, risk level labels and original anomaly distribution matrix are encoded into an incremental hierarchical storage format according to the pipeline topology structure, generating a dynamically updateable inspection report that supports rapid retrieval and visual backtracking by spatial coordinates or risk level.

Images