An underwater pipeline detection system and a detection data transmission and processing method thereof

Abstract

The present application relates to underwater pipeline detection technology field, especially a kind of underwater pipeline detection system and its detection data transmission and processing method, this method through the cooperative work of rotating camera and radio frequency positioning signal, combine core glass fiber line multi-channel transmission and three-dimensional space coordinate mapping technology, realized the real-time high-definition detection and centimeter-level accurate positioning of pipeline inner wall defect;Based on the intelligent processing algorithm of multi-scale texture analysis and dynamic threshold segmentation, corrosion, crack and other abnormal features can be accurately identified and classified;By constructing abnormal distribution matrix and risk diffusion model, automatically generating quantitative evaluation report and repair priority scheme, finally form the intelligent detection scheme integrated with data acquisition, transmission, processing and decision support, improve the accuracy, efficiency and reliability of underwater pipeline detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of underwater pipeline detection, and particularly relates to an underwater pipeline detection system and a detection data transmission and processing method thereof. BACKGROUND

[0002] Traditional underwater pipeline detection systems mostly adopt a drag type camera to cooperate with a cable to transmit a video signal, and have problems such as that the cable is heavy and easy to be corroded, the signal is easy to be interfered, and the positioning accuracy is low, especially in long distance or complex pipelines, the view angle is limited due to the winding of the cable, and the defect position cannot be positioned in real time and accurately. In the prior art, the video transmission mostly depends on copper cable or wireless communication, the former limits the dragging efficiency due to the weight and tensile resistance, and the latter causes image delay or distortion due to the influence of water body attenuation; and the positioning mostly depends on inertial navigation or sonar, and it is difficult to realize centimeter level accuracy in narrow pipelines. In addition, the traditional system lacks an efficient data fusion mechanism, and the video stream and the positioning information are often processed independently, which leads to that the defect analysis depends on manual comparison, and the efficiency is low and the risk of missing the space correlation is easy. In view of the above problems, there is an urgent need for an underwater detection system which is light, high reliability and has intelligent data processing capability, realizes real-time positioning, accurate analysis and risk prediction of defects in the pipeline through multi-modal data fusion and adaptive transmission technology, so as to improve the detection efficiency and decision reliability. SUMMARY

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

[0004] To achieve the above-mentioned purpose, the technical scheme adopted by the present application is as follows:

[0005] The present application discloses an underwater pipeline detection system, which comprises a rotating camera, a waterproof meter counter, a display, a core glass fiber line and a rack.

[0006] A wire reel is rotatably installed on the rack, and the core glass fiber line is wound on the wire reel, and the rotating camera is connected with the movable end of the core glass fiber line.

[0007] The rotating camera comprises a camera, a transmission gear, a stepping motor, a tail cover, a PC anti-scratch transparent cover and a radio frequency signal emitting coil, the camera is embeddedly installed on a camera base, and the camera base is covered by a camera face cover; the camera base and the transmission gear are connected through a conductive slip ring; and the transmission gear and the stepping motor are connected in transmission.

[0008] The stepping motor is fixed through a motor base, a spring connecting upper seat is connected at the end of the motor base, one end of the spring connecting upper seat is fixedly connected with the spring, and the other end of the spring is connected with a spring connecting lower seat.

[0009] Further, the rotating camera is provided with waterproof glue filling sections at preset positions from head to tail, which are glue filling section A, glue filling section B, glue filling section C and glue filling section D respectively.

[0010] Further, the waterproof meter installed on the rack, the waterproof meter includes meter cover, joint shaft and meter bottom cover, the meter PCBA board is installed in the meter cover, the meter PCBA board is connected with the magnetic sensing fixed base, the magnetic sensing fixed base is connected with the joint shaft through the meter limiting piece, the joint shaft is connected with the meter bottom cover through the meter bearing, the waterproof aviation head and the waterproof plug are also arranged on the meter bottom cover.

[0011] Further, the display includes a function area and a display area, the function area includes a bottom shell and a surface shell, a battery compartment is arranged on the base, waterproof cotton is arranged on the battery compartment, the base and the surface shell can be embedded with each other, the mutual embedding area of the base and the surface shell is also provided with waterproof cotton; a silica gel key is arranged on the base, and a key waterproof pressing plate is sleeved on the silica gel key.

[0012] Further, the display area includes a first shell and a second shell, the first shell and the second shell can be embedded with each other, waterproof cotton and waterproof EVA are arranged in the mutual embedding area of the first shell and the second shell.

[0013] The second aspect of the application discloses a detection data transmission and processing method of an underwater pipeline detection system, which is applied to any one of the underwater pipeline detection systems and includes the following steps:

[0014] Real-time video stream data inside the pipeline are collected, and radio frequency positioning signals are synchronously acquired, and after the video stream data and the radio frequency signals are multi-channel encoded, they are transmitted to a ground processing terminal;

[0015] Texture feature analysis is performed on the received video stream data, an abnormal area is extracted based on the texture features of the inner wall of the pipeline, and key feature information containing an abnormal position mark is generated;

[0016] Combined with cable winding and unwinding length data, a three-dimensional space coordinate system inside the pipeline is constructed, and the key feature information is dynamically bound with corresponding space coordinates to generate comprehensive detection information;

[0017] The comprehensive detection information is stored in a structured manner, a detection report is generated based on the distribution density and spatial correlation of the abnormal area, and a recommended repair priority and a potential risk level are marked.

[0018] Further, real-time video stream data inside the pipeline is collected, and a radio frequency positioning signal is synchronously acquired, and the video stream data and the radio frequency signal are multi-channel encoded and then transmitted to a ground processing terminal, specifically:

[0019] Real-time video stream data inside the pipeline is collected by a rotating camera and a radio frequency positioning signal of the rotating camera is synchronously collected;

[0020] The real-time video stream data is divided into continuous video frame groups according to a preset time length, and an initial time stamp and a radio frequency signal phase offset of each video frame group are extracted;

[0021] Each video frame group is compressed into a compressed data block, and the radio frequency phase offset is converted into a radio frequency frequency domain feature vector, which is bound to the corresponding compressed data block as a multi-channel data packet;

[0022] The signal attenuation characteristics of the radio frequency positioning signal are analyzed in real time, a phase equalization matrix is constructed, the radio frequency frequency domain feature vector in the multi-channel data packet is pre-corrected in phase, and the transmission bandwidth weight of the video data block is dynamically allocated;

[0023] The data packet after phase equalization is input into a cross-interleaved error correction encoder, an error correction code sequence is generated according to the video data priority, the sequence is divided into redundant sub-packets, and the redundant sub-packets are transmitted in parallel through independent core wires of a core glass fiber line;

[0024] The ground processing terminal reorganizes the redundant sub-packets according to the initial time stamp, repairs the transmission loss data by using the error correction code sequence, and reversely corrects the radio frequency frequency domain feature vector based on the phase equalization matrix, to obtain a time-space synchronous video stream and a radio frequency positioning signal.

[0025] Further, texture feature analysis is performed on the received video stream data, an abnormal area is extracted based on the texture features of the inner wall of the pipeline, and key feature information containing an abnormal position marker is generated, specifically:

[0026] Non-uniform illumination correction and gray scale equalization preprocessing are performed on the video stream frame by frame, multi-scale texture gradient maps are generated by a multi-direction Gabor filter bank based on the prior texture features of the inner wall material of the pipeline, and gradient amplitude and direction distribution features are extracted;

[0027] The multi-scale texture gradient maps are processed based on an adaptive region growing algorithm and combined with a gradient amplitude mutation threshold and a direction consistency constraint condition, the abnormal area is segmented out, and the boundary coordinates and area proportion of the abnormal area are marked;

[0028] The spatiotemporal evolution trajectory of the abnormal area in the continuous video frames is extracted according to the boundary coordinates of the abnormal area, a morphological feature model based on three-dimensional curvature change and surface roughness is constructed, and the geometric parameter difference between the corrosion area and the crack is distinguished;

[0029] Based on the geometric parameters of the morphological feature model, dynamic segmentation thresholds are set for the corrosion area and cracks respectively, and the accuracy of the abnormal boundary is optimized by genetic algorithm to generate a refined segmentation mask containing abnormal type labels;

[0030] The segmentation mask is spatiotemporally aligned with the spatial coordinates of the radio frequency positioning signal, and based on the geometric parameters and evolution trajectory of the abnormal area, key feature information including position, type, size and risk level is generated and embedded into the metadata channel of the video stream.

[0031] Further, combined with the cable length data, a three-dimensional coordinate system in the pipeline is constructed, and the key feature information and the corresponding spatial coordinates are dynamically bound to generate comprehensive detection information, specifically:

[0032] Based on the interval distribution of the phase change rate of the radio frequency positioning signal and the video stream timestamp, the non-uniformly sampled radio frequency signal and the video frame sequence are aligned to generate a set of spatiotemporally synchronized reference points;

[0033] Between the synchronization reference points, a non-uniform sampling cubic spline interpolation method is used to reconstruct the continuous phase trajectory of the radio frequency signal, and combined with the pulse count data of the cable length, a pipeline motion trajectory vector containing axial displacement and radial displacement is generated;

[0034] The motion trajectory vector is mapped to the pipeline preset geometric model, a three-dimensional coordinate system with the pipeline center line as the reference is constructed based on Cartesian coordinate conversion, and the spatial coordinate labels corresponding to each video frame are output;

[0035] The spatial distribution of the key feature information is mapped to the three-dimensional coordinate system, and the nearest neighbor association relationship between the feature points and the coordinate labels is established using a dynamic spatial hash grid algorithm, and the radial displacement weight coefficient of the feature points relative to the pipeline inner wall is calculated;

[0036] According to the radial displacement weight coefficient, the key feature information is spatially topologically sorted to generate comprehensive detection information including spatiotemporal correlation, abnormal distribution density and risk diffusion trend, and it is encoded as an incremental pipeline topology fingerprint and embedded in the video stream metadata.

[0037] Further, the comprehensive detection information is stored in a structured manner, and based on the distribution density and spatial correlation of the abnormal area, a detection report is generated, and the recommended repair priority and potential risk level are labeled, specifically:

[0038] The abnormal position, type and spatial coordinate label in the comprehensive detection information are mapped to the three-dimensional gridded model of the pipeline to generate an abnormal distribution matrix containing spatiotemporal correlation weight, record the abnormal density, type weight of each grid node and the topological connection relationship of adjacent nodes;

[0039] Based on the abnormal distribution matrix, an abnormal density value of each preset sub-region in the pipeline is obtained, and a preset sub-region with an abnormal density value higher than a preset density value is marked as a high-density abnormal region;

[0040] In combination with the pipeline material properties of the high-density abnormal region, a potential diffusion path of the defect in the high-density abnormal region along the axial and radial diffusion of the pipeline is determined;

[0041] According to the diffusion path length of the potential diffusion path, a repair urgency score function is defined, and a multi-level repair priority sequence is generated by a Pareto frontier optimization algorithm in combination with engineering constraint conditions;

[0042] Based on the cross probability of the abnormal type weight and the potential diffusion path, a composite risk index of each abnormal region is determined, and is divided into three risk level labels of critical risk, high risk and observation level according to a preset threshold;

[0043] The repair priority sequence, the risk level label and the original abnormal distribution matrix are encoded into an incremental hierarchical storage format according to the pipeline topological structure, a dynamically updated detection report is generated, and quick retrieval and visual backtracking according to spatial coordinates or risk levels are supported.

[0044] The underwater pipeline detection system and the detection data transmission and processing method provided by the present application have the following beneficial effects: through the cooperative work of the rotating camera and the radio frequency positioning signal, in combination with the multi-channel transmission of the core glass fiber and the three-dimensional space coordinate mapping technology, the real-time high-definition detection and centimeter-level accurate positioning of the defects on the inner wall of the pipeline are realized; based on the intelligent processing algorithm of multi-scale texture analysis and dynamic threshold segmentation, the abnormal features such as corrosion and cracks can be accurately identified and classified; by constructing the abnormal distribution matrix and the risk diffusion model, the quantitative evaluation report and the repair priority scheme are automatically generated, and finally the intelligent detection scheme integrating data acquisition, transmission, processing and decision support is formed, and the accuracy, efficiency and reliability of the underwater pipeline detection are improved. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can also obtain other drawings of embodiments according to these drawings without creative labor.

[0046] Figure 1 It is the first overall structure schematic diagram of the detection system;

[0047] Figure 2 It is the second overall structure schematic diagram of the detection system;

[0048] Figure 3 This is a schematic diagram of the exploded structure of the rotating camera in this detection system;

[0049] Figure 4 This is an exploded structural diagram of the waterproof meter counter in this testing system;

[0050] Figure 5 This is an exploded view of the display functional area in this testing system;

[0051] Figure 6 This is a schematic diagram of the exploded structure of the display area of ​​the monitor in this detection system;

[0052] Figure 7 This is a schematic diagram of the coil structure in this testing system. Detailed Implementation

[0053] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0054] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0055] like Figure 1 As shown, the first aspect of the present invention discloses an underwater pipeline inspection system, the inspection system comprising a rotating camera 1, a waterproof meter counter 2, a display 3, a core glass fiber cable 4, and a frame 5;

[0056] like Figure 2 , Figure 7 As shown, a wire reel 101 is rotatably mounted on the frame, the core glass fiber wire 4 is wound on the wire reel 101, and the rotating camera 1 is connected to the movable end of the core glass fiber wire 4; the wire reel 101 is rotatably mounted on the frame 5 through a wire reel rotation bearing 102; a stainless steel wire clip 103 and an aviation head waterproof connector 104 are also provided on the wire reel.

[0057] It should be noted that the reel mechanism of this underwater pipeline inspection system achieves flexible rotation through the reel's rotating bearing 102, allowing the core glass fiber wire 4 wound on it to be smoothly wound and unwound. The stainless steel wire clip 103 secures the end of the fiber wire to prevent loosening, and the aviation-grade waterproof connector 104 ensures the waterproof reliability of power / signal transmission. Its working principle is as follows: the reel 101 rotates freely on the frame 5, and the bearing 102 reduces frictional resistance, allowing control of the winding and unwinding length of the core glass fiber wire 4 to adjust the detection position of the rotating camera 1. Simultaneously, the waterproof connector 104 maintains the sealing of the circuit. The bearing 102 ensures smooth rotation, the stainless steel wire clip 103 provides mechanical fixation, and the waterproof connector 104 achieves safe electrical connection in the underwater environment, collectively ensuring the reliable operation of the system during underwater pipeline inspection.

[0058] like Figure 3 As shown, the rotating camera includes a camera 105, a transmission gear 106, a stepper motor 107, a tail cover 108, a PC scratch-resistant transparent cover 109, and an radio frequency signal transmitting coil 110. The camera 105 is fitted onto a camera base 201, which is covered by a camera face cover 202. The camera base and the transmission gear 106 are connected by a conductive slip ring 203. The transmission gear and the stepper motor are connected for mutual transmission.

[0059] The stepper motor is fixed by a motor base 204. A spring connecting upper seat 205 is connected to the end of the motor base. The spring connecting upper seat is fixedly connected to one end of a spring 206, and the other end of the spring is connected to a spring connecting lower seat 207. The rotating camera has several waterproof glue-filling sections at preset positions from head to tail, namely glue-filling section A 208, glue-filling section B 209, glue-filling section C 301, and glue-filling section D 302.

[0060] 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 potting structure ensures the reliability of underwater operations, and the PC scratch-resistant transparent cover 109 protects the lens. Together, they achieve blind-spot-free detection inside the pipeline and long-term stable underwater operation.

[0061] A waterproof pressure plate 303 is provided between the tail cap 108 and the glue-filled section D 302; several waterproof rings 304, a pin guide plate 305, a pin plate 306, and a Wuling annular pressure plate 307 are provided at preset positions from the head to the tail of the rotating camera; an anti-pull steel wire 308 and a 5-core connecting wire 309 are provided inside the spring; the radio frequency 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 fitted on the outside of the PC anti-scratch transparent cover 109.

[0062] It should be noted that the rotating camera forms a multi-layered sealing barrier through the waterproof pressure plate 303 and the multi-layered waterproof ring 304, ensuring reliable protection in the high-pressure underwater environment. The ejector pin guide plate 305 and ejector pin plate 306 precisely position the internal components, while the Wuling annular pressure plate 307 provides structural reinforcement. The anti-tension steel wire 308 enhances the tensile strength of the spring 206 and protects the built-in 5-core connecting wire 309. The 512Hz radio frequency signal transmitting coil 110 achieves precise positioning, the light board 402 provides underwater illumination, and the anti-collision guardrail 403 protects the PC scratch-resistant transparent cover 109 from impact. The coordinated action of these components ensures both the camera's waterproof and pressure-resistant performance and the stability of signal transmission and the structural strength of the equipment, enabling the system to perform long-term stable pipeline inspection operations in complex underwater environments.

[0063] The camera base 201 and the camera cover 202 are fitted with a camera protective cover 404 on their outer sides; eight protective cover wheels 405 are installed on the surface of the camera protective cover 404.

[0064] It should be noted that the underwater pipeline inspection system uses a fiberglass core 4 to insert a rotating camera 1 into the pipeline. Driven by a stepper motor 107 and a transmission gear 106, the camera 105 rotates 360° to capture images, and the video signal is transmitted in real-time to a ground display 3 via the fiberglass core 4. A 512Hz radio frequency signal transmitting coil 110, in conjunction with ground detection equipment, achieves precise positioning. Multi-layer waterproof potting compound, waterproof rings 304, and anti-collision guardrails 403 ensure the equipment's sealing and impact resistance in high-pressure underwater environments. Thus, by replacing traditional cables with lightweight fiberglass cores, drag resistance is reduced; conductive slip rings 203 prevent cable entanglement during rotation; springs 206 and anti-tension steel wires 308 buffer tension and protect the cable; and anti-collision wheel sets 405 improve pipeline passability, making it suitable for efficient inspection and fault location of complex industrial pipelines.

[0065] like Figure 4 As shown, the waterproof meter counter 2 is mounted on the frame 5. The waterproof meter counter includes a meter counter cover 406, a connector shaft 407, and a meter counter bottom cover 408. A meter counter PCBA board 409 is installed inside the meter counter cover. The meter counter PCBA board is connected to a magnetic induction fixing base 501. The magnetic induction fixing base 501 is connected to the connector shaft 407 through a meter counter limiting piece 503. The connector shaft is connected to the meter counter bottom cover 408 through a meter counter bearing 505. A waterproof aviation head 506 and a waterproof plug 507 are also provided on the meter counter bottom cover. A bearing limiting snap ring 508 is provided on the meter counter bearing 505.

[0066] It should be noted that the waterproof meter counter 2, through the magnetic fixing base 501 and the meter counter PCBA board 409, monitors the length of the core glass fiber cable 4 in real time. The connector shaft 407 rotates flexibly via the meter counter bearing 505, and its displacement range is controlled by the meter counter limiting plate 503. The waterproof aviation head 506 and waterproof plug 507 ensure the sealing in the underwater environment, while the bearing limiting spring 508 prevents the bearing from dislodging. The waterproof meter counter 2 accurately records the cable length to locate fault points within the pipeline. Its waterproof design adapts to humid or high-pressure environments, the bearing structure ensures smooth rotation, and the overall structure is compact and reliable, providing accurate length reference data for underwater pipeline inspection.

[0067] like Figure 5 , Figure 6 As shown, the display includes a functional area 6 and a display area 7. The functional area includes a bottom shell 509 and a front 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 front shell can be interlocked, and the interlocking area of ​​the base and the front 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 fitted on the silicone button 604. A battery is provided in the battery compartment. A silicone anti-collision pad 606 is provided on the bottom of the bottom shell 509. A waterproof aviation head 506, a charging base 607, a rotating device cover 608, a webbing hanging ring 609, and a control PCBA board 701 are provided on the front shell 601.

[0068] The display area includes a first housing 702 and a second housing 703, which can be interlocked. A waterproof cotton strip 603 and a waterproof EVA component 704 are provided in the interlocking area of ​​the first and second housings. The second housing 703 is provided with a sunshade aluminum plate 705, a front-to-back and left-to-right rotating damping shaft 706, a display acrylic plate 707, a 9-inch display screen 708, and a display driver board PCBA board 709.

[0069] It should be noted that the display consists of a function area 6 and a display area 7. The function area achieves waterproof operation control through silicone buttons 604 and a waterproof button pressure plate 605. The battery compartment 602 and the waterproof cotton strip 603 at the fitting ensure the equipment's airtightness in humid environments. The display area uses a 9-inch display screen 708 in conjunction with a sunshade aluminum plate 705 to enhance outdoor visibility. The front, back, left, and right rotating damping shaft 706 supports multi-angle adjustment. The waterproof structure (waterproof cotton strip 603, EVA component 704) ensures reliability in underwater operations, the silicone anti-collision pad 606 buffers impacts, and the 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 needs in complex environments.

[0070] The second aspect of this invention discloses a detection data transmission and processing method for an underwater pipeline inspection system, applicable to any of the underwater pipeline inspection systems described in the present invention, comprising the following steps:

[0071] Real-time video stream data inside the pipeline is collected, and radio frequency positioning signals are acquired simultaneously. The video stream data and radio frequency signals are encoded in multiple channels and then transmitted to the ground processing terminal.

[0072] The received video stream data is subjected to texture feature analysis. Based on the texture features of the inner wall of the pipe, abnormal areas (such as cracks and corrosion points) are extracted, and key feature information containing abnormal location markers is generated.

[0073] By combining cable winding and unwinding length data, a three-dimensional spatial coordinate system inside the pipeline is constructed, and key feature information is dynamically bound to the corresponding spatial coordinates to generate comprehensive detection information;

[0074] 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 abnormal areas, with recommended repair priorities and potential risk levels marked.

[0075] It should be noted that this method solves the technical problems of asynchronous transmission of video data and positioning signals, low defect identification efficiency, and lack of spatial correlation in underwater pipeline inspection. By achieving data synchronization through multi-channel encoded transmission, combined with texture feature analysis and 3D coordinate binding, it can achieve real-time and accurate positioning of pipeline defects, automatic identification of abnormal areas, and generation of structured inspection reports containing risk levels, effectively improving inspection efficiency and decision reliability.

[0076] Furthermore, real-time video stream data inside the pipeline is collected, and radio frequency positioning signals are acquired simultaneously. The video stream data and radio frequency signals are then encoded in multiple channels and transmitted to the ground processing terminal. Specifically:

[0077] The system uses a rotating camera to collect real-time video stream data inside the pipe and simultaneously collects the radio frequency positioning signal of the rotating camera.

[0078] The real-time video stream data is divided into continuous video frame groups according to a preset duration, and the initial timestamp and radio frequency signal phase offset of each video frame group are extracted.

[0079] Each video frame group is compressed into a compressed data block, and the radio frequency phase offset is converted into a radio frequency domain feature vector and bound to the corresponding compressed data block as a multi-channel data packet.

[0080] The signal attenuation characteristics of radio frequency positioning signals are analyzed in real time, a phase equalization matrix is ​​constructed, the phase pre-correction of the radio frequency domain feature vector in the multi-channel data packets is performed, and the transmission bandwidth weight of video data blocks is dynamically allocated.

[0081] It should be noted that the amplitude-frequency response curve of the 512Hz positioning signal during transmission through the core glass fiber line is monitored in real time by the RF signal receiving module to obtain the phase offset of each subcarrier frequency point (e.g., a -22° offset at 1.5MHz). A 4×4 complex phase equalization matrix is ​​generated based on least squares fitting, and linear transformation compensation is performed on the real and imaginary parts of the 128-dimensional RF frequency domain feature vector in the data packet. At the same time, the transmission bandwidth ratio of each data block is adjusted according to the I-frame / P-frame type of the video data block (e.g., I-frames are assigned a weight coefficient of 0.8) and the current channel signal-to-noise ratio (e.g., 10% bandwidth is increased when SNR>30dB) to ensure priority transmission of key video frames. The compensated frequency domain feature vector and video data packet are output to the next processing module after time-division multiplexing via TDM.

[0082] The phase-equalized data packets are input into the cross-interleaved error correction encoder, which generates error correction code sequences according to the video data priority. The sequences are then divided into redundant sub-packets, which are transmitted in parallel through the independent core wires of the fiberglass core.

[0083] The ground processing terminal reassembles redundant sub-packets based on the initial timestamp, repairs transmission loss data using error correction code sequences, and performs reverse correction on the radio frequency domain feature vector based on the phase equalization matrix to obtain a spatiotemporally synchronized video stream and radio frequency positioning signal.

[0084] For example, a rotating camera captures a 1080P video stream of the pipe's inner wall at 30 frames per second, while simultaneously acquiring a positioning signal via a 512Hz RF transmitting coil. The video stream is divided into 15 consecutive video frame groups every 0.5 seconds, and the start timestamp of each group and the corresponding phase offset of the RF signal (e.g., ±15°) are recorded. H.265 encoding is used to compress the video frame groups into average 2MB data blocks, and a Fast Fourier Transform is used to convert the RF phase offset into a 128-dimensional RF frequency domain feature vector. These two are then bound together to form a composite data packet. During transmission, the signal attenuation curve of the core glass fiber wire in the 5MHz band (e.g., -3dB / m) is monitored in real time. A 4×4 phase equalization matrix is ​​constructed to pre-compensate the RF feature vector, and 70% of the transmission bandwidth for the video data is dynamically allocated. A 32-byte error correction code sequence generated by (255,223) Reed-Solomon encoding is divided into four redundant sub-packets and transmitted in parallel through four-core optical fibers. The ground terminal reassembles sub-packets based on timestamps, repairs data loss using error correction codes, and restores radio frequency signals through inverse matrix operations, ultimately outputting synchronization data with a time deviation of less than 1ms.

[0085] In summary, this method, through multi-channel data binding, phase pre-correction, and redundant parallel transmission, can achieve the direct benefits of high-fidelity synchronous transmission of video and positioning signals, strong anti-interference capability, and high data integrity, ensuring that the ground terminal obtains accurate and reliable detection data.

[0086] Furthermore, texture feature analysis is performed on the received video stream data. Anomaly regions are extracted based on the texture features of the inner wall of the pipe, generating key feature information containing anomaly location markers. Specifically:

[0087] The video stream is preprocessed frame by frame with non-uniform illumination correction and grayscale equalization. Based on the prior texture features of the pipe inner wall material, a multi-scale texture gradient map is generated through a multi-directional Gabor filter bank, and the gradient magnitude and direction distribution features are extracted.

[0088] Among them, the prior texture features of the pipe inner wall material refer to the benchmark texture model library established by pre-collecting and analyzing standard surface texture data (such as metal particle distribution, weld pattern, oxide layer morphology, etc.) of the inner wall of a pipe of a specific material, which is used for subsequent defect detection comparison.

[0089] The multi-scale texture gradient map is processed based on an adaptive region growing algorithm and combined with gradient magnitude mutation threshold and direction consistency constraint to segment abnormal regions and mark the boundary coordinates and area ratio of the abnormal regions.

[0090] It should be noted that the adaptive region growing algorithm achieves abnormal region segmentation in the following way: First, pixels with gradient magnitudes exceeding a preset abrupt change threshold (e.g., 35 gray levels) in the multi-scale texture gradient map 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 between adjacent pixels is less than a set tolerance (e.g., ±15°), it is determined to be a homogeneous region and included in the growth range; during the growth process, the magnitude threshold and directional tolerance are dynamically adjusted, where the magnitude threshold decreases exponentially with the increase of the region area, and the directional tolerance is linearly compensated according to the complexity of the region's shape; finally, when the region grows to meet the following termination conditions, the abnormal region is output: 1. The average gradient magnitude of pixels at the region's edge drops 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 between the background texture of the pipe's inner wall and the actual defect region by balancing gradient magnitude abrupt changes and directional continuity constraints.

[0091] Based on the boundary coordinates of the abnormal region, its spatiotemporal evolution trajectory in continuous video frames is extracted, and a morphological feature model based on three-dimensional curvature change and surface roughness is constructed to distinguish the geometric parameter differences between the corrosion region (continuous curvature distribution) and the crack (local abrupt curvature change).

[0092] Based on the geometric parameters of the morphological feature model, dynamic segmentation thresholds are set for corrosion areas and cracks respectively. The accuracy of abnormal boundaries is optimized by a genetic algorithm to generate a refined segmentation mask containing abnormal type labels (corrosion / crack).

[0093] It should be noted that, firstly, the average radius of curvature (>100mm) of the corroded region and the local curvature extremum (<20mm) of the crack region are extracted as initial classification features based on the morphological feature model; secondly, a linear threshold function with surface roughness (Ra value 0.8-1.2μm) as the independent variable is established for the corroded region, and a nonlinear threshold mapping table based on the rate of curvature change (>5% / mm) is constructed for the crack region; then, a genetic algorithm is used for optimization, in which the population is set to 20 sets of threshold parameter combinations, and the fitness function comprehensively considers boundary continuity (requiring the proportion of closed contours >95%) and feature discriminability (inter-class difference >30%); after 15-20 generations of iteration, the optimal parameter combination is selected to generate a binary mask, and discrete noise is eliminated by morphological closing operation (3×3 structuring element), and finally, a refined segmentation result with type label (corrosion / crack) is output, in which the corroded region is labeled with a gray value of 150 and the crack region is labeled with a gray value of 250.

[0094] The spatial coordinates of the segmentation mask and the radio frequency positioning signal are spatiotemporally aligned. Based on the geometric parameters and evolution trajectory of the abnormal region, key feature information including location, type, size and risk level is generated and embedded into the metadata channel of the video stream.

[0095] For example, a 1080P video stream (30 frames / second) captured by a rotating camera first undergoes illumination correction processing based on the Retinex model to equalize the grayscale values ​​of the pipe's inner wall to the range of 50-200. Each frame is then processed using an 8-directional (0°-157.5°, 22.5° interval) and 3-scale (wavelength 16 / 32 / 64 pixels) Gabor filter bank to generate a texture feature map containing gradient magnitude (0-255) and orientation angle (0-360°). For the inner wall of a carbon steel pipe, a gradient abrupt change threshold of 35 gray levels and an orientation consistency tolerance of ±15° are set. Anomaly regions with an area greater than 50 pixels are extracted using a region growing algorithm, and the vertex coordinates of the boundary polygons are output. The motion trajectory of the anomaly regions is tracked in 10 consecutive video frames, and the radius of curvature (corrosion area > 100 mm, crack area < 20 mm) and surface roughness Ra value (corrosion area 0.8-1.2 μm, crack area > 2.5 μm) are calculated to establish a three-dimensional morphological database. Based on the material type (e.g., X70 steel), a dynamic segmentation threshold is set (grayscale difference threshold of 25±5 for corrosion area and 40±10 for crack area). After 20 generations of genetic iteration optimization, a binary mask with a precision of 0.5mm is generated. Finally, the abnormal information (e.g., crack @3.2mm×15mm, coordinates X125.6m / Y-30°Z+5mm) is written into the SEI metadata area of ​​the H.264 video stream.

[0096] In summary, this method, through multi-scale texture analysis and dynamic threshold segmentation, can achieve automatic and accurate identification and classification of defects in the inner wall of pipelines, accurately distinguish abnormal features such as corrosion and cracks, and generate structured feature information containing location, type, and risk level, thereby improving detection accuracy and analysis efficiency.

[0097] Furthermore, by combining cable winding and unwinding length data, a three-dimensional spatial coordinate system is constructed within the pipeline, and key feature information is dynamically bound to the corresponding spatial coordinates to generate comprehensive inspection information, specifically:

[0098] Based on the phase change rate of the radio frequency positioning signal and the interval distribution of the video stream timestamps, the non-uniformly sampled radio frequency signal and video frame sequence are aligned to generate a set of spatiotemporal synchronization reference points.

[0099] Between synchronization reference points, the continuous phase trajectory of the radio frequency signal is reconstructed using non-uniform sampling cubic spline interpolation. Combined with pulse count data of cable winding and unwinding length, a pipeline motion trajectory vector containing axial displacement and radial offset is generated.

[0100] The motion trajectory vector is mapped to the pre-defined geometric model of the pipe (such as a circular / rectangular cross-section), a three-dimensional spatial coordinate system based on the pipe centerline is constructed based on Cartesian coordinate transformation, and the spatial coordinate labels corresponding to each video frame are output.

[0101] The spatial distribution of key feature information (abnormal location, 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 feature points relative to the inner wall of the pipe is calculated.

[0102] The formula for calculating the radial offset weighting coefficient is as follows:

[0103] ;

[0104] In the formula, The radial offset weighting coefficient for the i-th feature point; The distance from the i-th feature point to the inner wall of the pipe (unit: mm); This is the theoretical radius of the inner wall of the pipe (e.g., 300mm for a DN600 pipe). The critical radius for risk assessment (usually taken as r) min +20mm).

[0105] Based on the radial offset weighting coefficient, key feature information is spatially topologically sorted to generate comprehensive detection information that includes spatiotemporal correlation, anomaly distribution density, and risk diffusion trend. This information is then encoded as an incremental pipeline topology fingerprint and embedded into the video stream metadata.

[0106] For example, the system takes a 512Hz radio frequency signal (phase resolution 0.1°) and a 30fps video stream as input. Every 5-second interval, it selects video frame timestamps and aligns them with the zero-crossing points of the radio frequency signal to generate a set of 50-60 synchronization reference points. For intervals of approximately 150ms between adjacent reference points, it fits the radio frequency phase trajectory using a cubic spline curve and combines this with cable encoder data of 200 pulses per meter to generate 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 into a three-dimensional coordinate system with the pipe center as the origin (0,0,0), where the axial Z-coordinate accuracy is 0.1% and the circumferential θ-coordinate accuracy is 0.5°. Anomaly points (such as corrosion areas) are matched to the nearest coordinate labels using a hash grid (50mm × 10° cell size), and their offset weights from the inner wall (0-1 normalized values) are determined. The resulting comprehensive inspection information includes: an anomaly density distribution map for every 10cm pipe segment; a list of risk hotspot areas sorted by weight coefficients; and a topological fingerprint data block that can be updated as the inspection progresses (4KB of metadata embedded per MB of video stream). In actual implementation, the pipe geometry parameters and sampling frequency can be adjusted according to inspection requirements.

[0107] In summary, this method solves the technical problems of inaccurate matching of video data and spatial location information, and inaccurate defect localization in underwater pipeline inspection. By achieving spatiotemporal synchronization of radio frequency signals and cable data and three-dimensional coordinate mapping, centimeter-level precise localization of internal pipeline defects is achieved, and a fully automatic association between defect features and spatial location is established, ultimately generating a structured inspection report containing complete information such as defect distribution and risk trends.

[0108] 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 abnormal areas, with recommended remediation priorities and potential risk levels indicated, specifically:

[0109] The abnormal location, type, and spatial coordinate labels in the comprehensive detection information are mapped to the three-dimensional mesh model of the pipeline to generate an abnormal distribution matrix containing spatiotemporal correlation weights, and record the abnormal density, type weight, and topological connection relationship of each mesh node.

[0110] 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.

[0111] By combining the pipe material properties (such as metal fatigue coefficient and corrosion rate) in the high-density anomaly region, the potential diffusion paths of defects along the pipe axial and radial directions in the high-density anomaly region are determined.

[0112] A repair urgency scoring function is defined based on the length of the potential diffusion path, and a multi-level repair priority sequence is generated by combining engineering constraints (such as repair cost and construction accessibility) using the Pareto front optimization algorithm.

[0113] It should be noted that, firstly, an urgency scoring model is established with the diffusion path length as the core parameter. For every 1 meter increase in axial diffusion distance, the base score increases by 20 points, and radial diffusion to adjacent grid cells increases by 10 points. A multi-objective optimization system incorporating engineering constraints is constructed, with repair cost (in ten thousand yuan), construction accessibility (standardized score from 0 to 1), and the criticality of the pipe segment (e.g., a weight of 1.5 for the main line) as parallel optimization objectives. A Pareto optimal solution set is selected through non-dominated sorting, where each solution corresponds to a set of weighted allocation schemes (e.g., cost weight 0.4, accessibility weight 0.3, criticality weight 0.3). Finally, based on the weighted total score of each scheme in the solution set, three priority levels are assigned: Level 1 (>80 points) requires handling within 24 hours, Level 2 (60-80 points) requires handling 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 reality, a standardized transformation from technical parameters to decision-making schemes is achieved.

[0114] Based on the cross probability of anomaly type weights (such as crack depth and corrosion area ratio) and potential diffusion paths, a composite risk index is determined for each anomaly area, and it is divided into three risk level labels: critical risk, high risk, and observation level according to preset thresholds.

[0115] 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, generating dynamically updated inspection reports that support quick retrieval and visual backtracking by spatial coordinates or risk level.

[0116] For example, for DN800 steel pipes, the pipe is divided into 100mm×100mm×500mm (axial) grid units. Each grid node records: (1) abnormal density value (0-10 level, such as 5 corrosion points / m² recorded as density value 7); (2) type weight coefficient (crack 0.8, corrosion 0.5); (3) connection relationship with 6 adjacent nodes. The density threshold is set to 6, and high-density areas composed of more than 3 consecutive grids exceeding the limit are marked (such as pipe sections 35.2-36.7m from the starting point). According to the steel properties (fatigue coefficient 0.85, corrosion rate 0.2mm / year), the axial diffusion speed of defects is determined (crack 2.1m / year, corrosion 0.8m / year) and a diffusion path heat map is generated (red high-risk area diffusion distance > 1m). The repair priority is divided into 0-100 points: defects with a construction window period of <7 days and diffusion intersection are rated 90 points (level 1), and those with cost exceeding the budget by 20% are rated 60 points (level 3). Risk levels are categorized by a composite index: >85 indicates critical risk (red), 60-85 indicates high risk (yellow), and the rest are under observation (green). The final report is stored in a layered JSON format. The base layer contains raw grid data (approximately 2MB / km), and the application layer contains optimized decision data (200kB / km). It supports real-time retrieval of the 3D model via a GIS system by coordinates (e.g., X=35.6m, Y=120°) or risk level.

[0117] In summary, this method automatically generates objective and quantitative repair priority sequences and risk level classifications by establishing an anomaly distribution matrix and diffusion path model, thereby enabling intelligent assessment and scientific decision-making for pipeline defects and improving the professionalism of inspection reports and the rationality of maintenance plans.

[0118] In practical applications, the detection data transmission and processing method may further include the following steps:

[0119] Electrochemical impedance spectroscopy (EIS) and differential pulse voltammetry (DPV) signals of epiphytic microbial communities on the tube wall were acquired in real time using a microelectrode array. After wavelet multi-scale decomposition, an electrochemical fingerprint feature vector containing redox peak positions and charge transfer resistance values ​​was generated.

[0120] The local electronic state density distribution of iron on the pipe surface was inverted using synchrotron X-ray absorption near-edge structure (XANES) data, generating an electronic state distribution map of metal lattice defects, and marking the energy barrier parameters of the corrosion region.

[0121] The electrochemical fingerprint feature vector and the electronic state distribution map of lattice defects are input into the graph convolutional neural network. Through topological matching of the electron-ion coupling transport path, a spatial correlation matrix between the microbial metabolic current intensity and the active sites of metal defects is established to quantify the attenuation coefficient of the corrosion barrier by the catalytic effect of biofilm.

[0122] 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 passivation film rupture probability of the microbial-metal interface are determined, and a dynamically updated corrosion energy barrier parameter spectrum is generated.

[0123] When the corrosion 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 corrosion are traced according to the spatial correlation matrix, and targeted biological inhibition strategy suggestions are generated.

[0124] For example, EIS signals (frequency range 0.1Hz-100kHz) and DPV signals (pulse amplitude 50mV) were acquired at a scan rate of 10mV / s. After being decomposed into 5 frequency bands by Db4 wavelet, feature values ​​(such as the oxidation peak at 711mV and charge transfer resistance of 2.8kΩ·cm²) were extracted. Simultaneously, XANES spectra of the iron L-edge were acquired using a third-generation synchrotron radiation source (energy 4-15keV), and the surface Fe was obtained by inversion using FEFF9 software. 3D electronic density of states distribution was used to identify highly active sites at grain boundaries with energy barriers below 1.2 eV. Inputting this data into a neural network with 3 layers of graph convolution (256-128-64 nodes) showed a spatial correlation of 0.92 between the metabolic current of sulfate-reducing bacteria (>0.15 μA / cm²) and the grain boundary defect region (barrier <0.8 eV). When the ambient pH <6.5 and dissolved oxygen <0.5 ppm, the corrosion current density threshold rose to 1.2 μA / cm², and the passivation film rupture probability >65%. The system triggered a red alert when the energy barrier parameter of a pipe section (e.g., axial 35.6 m, circumferential 120° position) remained below the safe threshold of 0.5 eV for 3 consecutive days, recommending the application of a 20 ppm concentration of molybdate corrosion inhibitor. In practical applications, the electrode spacing and spectral parameters can be adjusted according to the pipe diameter.

[0125] In summary, this method, by integrating microbial electrochemical signals with metal lattice defect analysis, establishes a correlation between biofilm activity and metal corrosion tendency. This enables early warning of microbial corrosion risks and precise location of high-risk pipe sections, while providing a scientific basis for targeted protective measures and enhancing the proactiveness and accuracy of pipeline corrosion protection.

[0126] In practical applications, the detection data transmission and processing method may further include the following steps:

[0127] The pressure pulsation signal of the fluid in the pipeline is collected in real time by a distributed fiber optic acoustic sensor array, and the flow velocity profile data is obtained by an ultrasonic flow meter to construct a flow field parameter matrix that includes pressure gradient, vorticity distribution and gas-liquid mixing phase characteristics.

[0128] The parameter matrix of the abnormal flow field region is subjected to intrinsic mode decomposition to extract the characteristic frequency components related to leakage (such as low-frequency pressure fluctuations of 0.5-5Hz and sudden increase in local vorticity) and generate leakage characteristic fingerprint spectrum.

[0129] Based on the vortex core migration trajectory and pressure gradient change direction in the leakage characteristic fingerprint spectrum, a spatiotemporal correlation equation between the leakage point location and the downstream flow field distortion region is established, and the linear correlation coefficient between leakage intensity and flow field disturbance range is quantified.

[0130] Based on the correlation coefficient distribution of the spatiotemporal correlation equation and combined with the pipeline topology (elbows, valve locations), the probability weight of each pipe segment as a leakage source is determined, and a probability heatmap of leakage points is generated.

[0131] For example, distributed fiber optic acoustic sensors (sampling rate 1kHz) are deployed every 50 meters along a DN800 oil pipeline, in conjunction with 8-channel ultrasonic flow meters (accuracy ±0.5%) installed on the pipeline cross-section, to collect fluid pressure fluctuations (range 0-2MPa) and flow velocity profiles (0-5m / s) in real time, constructing a flow field parameter matrix that includes axial pressure gradient (Pa / m), vorticity (1 / s), and gas-liquid mixture fraction (0-100%). When a low-frequency pressure fluctuation of 0.8-3.2Hz is detected in a certain pipe section (e.g., station K35+200) and a local vorticity suddenly increases to 15 / s, five intrinsic mode functions are extracted through empirical mode decomposition, and the IMF3 component (energy percentage > 40%) related to leakage is screened to generate a characteristic fingerprint spectrum. Based on the downstream migration rate of the vortex core (approximately 0.6 m / s) and the direction of pressure gradient change (30° off the axis) in the flow spectrum, a correlation equation was established between the location of the leak point (e.g., K35+180) and the 50-meter downstream flow field distortion zone, with a correlation coefficient of 0.78. Combining the flow field disturbance characteristics of the upstream bend (R=3D) of this pipe section, the leakage probability weights for each node were determined (reaching 85% at K35+180), ultimately generating a heat map marked with red (high probability), yellow (medium), and blue (low probability). In actual deployment, the sensor spacing can be adjusted according to the viscosity of the medium.

[0132] In summary, by integrating multi-source flow field data with dynamic feature analysis, it is possible to quickly identify the flow field distortion characteristics induced by leakage, and establish a precise correlation between the leakage point and the flow field anomaly, thereby achieving non-contact, high-precision leakage tracing and location.

[0133] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for transmitting and processing detection data in an underwater pipeline inspection system, characterized in that: The detection system includes a rotating camera, a waterproof meter, a display, a core glass fiber cable, and a frame; A reel is rotatably mounted on the frame, the core glass fiber wire is wound on the reel, 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 stepper motor, a tail cover, a PC scratch-resistant transparent cover, and an RF signal transmitting coil. The camera is fitted onto a camera base, which is covered by a camera face cover. The camera base and the transmission gear are connected by a conductive slip ring. The transmission gear and the stepper motor are interconnected and drive each other. The stepper motor is fixed by a motor base, and a spring connecting upper seat is connected to the end of the motor base. The spring connecting upper seat is fixedly connected to one end of the spring, and the other end of the spring is connected to the spring connecting lower seat. The detection data transmission and processing method includes: Real-time video stream data inside the pipeline is collected, and radio frequency positioning signals are acquired simultaneously. The video stream data and radio frequency signals are encoded in multiple channels and then transmitted to the ground processing terminal. The received video stream data is subjected to texture feature analysis. Abnormal regions are extracted based on the texture features of the inner wall of the pipe, and key feature information containing abnormal location markers is generated. By combining cable winding and unwinding length data, a three-dimensional spatial coordinate system inside the pipeline is constructed, and key feature information is dynamically bound to the corresponding spatial coordinates to generate comprehensive detection information; 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 abnormal areas, with recommended repair priorities and potential risk levels marked. Real-time video stream data inside the pipeline is collected, and radio frequency positioning signals are acquired simultaneously. The video stream data and radio frequency signals are then encoded in multiple channels and transmitted to the ground processing terminal. Specifically: The system uses a rotating camera to collect real-time video stream data inside the pipe and simultaneously collects the radio frequency positioning signal of the rotating camera. The real-time video stream data is divided into continuous video frame groups according to a preset duration, and the initial timestamp and radio frequency signal phase offset of each video frame group are extracted. Each video frame group is compressed into a compressed data block, and the radio frequency phase offset is converted into a radio frequency domain feature vector and bound to the corresponding compressed data block as a multi-channel data packet. The signal attenuation characteristics of radio frequency positioning signals are analyzed in real time, a phase equalization matrix is ​​constructed, the phase pre-correction of the radio frequency domain feature vector in the multi-channel data packets is performed, and the transmission bandwidth weight of video data blocks is dynamically allocated. The phase-equalized data packets are input into the cross-interleaved error correction encoder, which generates error correction code sequences according to the video data priority. The sequences are then divided into redundant sub-packets, which are transmitted in parallel through the independent core wires of the fiberglass core. The ground processing terminal reassembles redundant sub-packets based on the initial timestamp, repairs transmission loss data using error correction code sequences, and performs reverse correction on the radio frequency domain feature vector based on the phase equalization matrix to obtain a spatiotemporally synchronized video stream and radio frequency positioning signal.

2. The detection data transmission and processing method of an underwater pipeline inspection system according to claim 1, characterized in that: The rotating camera has several waterproof glue-filling sections at preset positions from head to tail, namely glue-filling section A, glue-filling section B, glue-filling section C and glue-filling section D.

3. The detection data transmission and processing method of an underwater pipeline inspection system according to claim 1, characterized in that: The waterproof meter counter is mounted on the frame. The waterproof meter counter includes a meter counter cover, a connector shaft, and a meter counter bottom cover. A meter counter PCBA board is installed inside the meter counter cover. The meter counter PCBA board is connected to a magnetic induction fixing base. The magnetic induction fixing base is connected to the connector shaft through a meter counter limiting piece. The connector shaft is connected to the meter counter bottom cover through a meter counter bearing. A waterproof aviation head and a waterproof plug are also provided on the meter counter bottom cover.

4. The detection data transmission and processing method of an underwater pipeline inspection 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 front shell. A battery compartment is provided on the base, and a waterproof cotton strip is provided on the battery compartment. The base and the front shell can be interlocked. The interlocking area of ​​the base and the front shell is also provided with a waterproof cotton strip. A silicone button is provided on the base, and a button waterproof pressure plate is fitted on the silicone button.

5. The detection data transmission and processing method of an underwater pipeline inspection system according to claim 4, characterized in that: The display area includes a first housing and a second housing, which can be fitted together. A waterproof cotton strip and a waterproof EVA component are provided in the area where the first housing and the second housing fit together.

6. The detection data transmission and processing method of an underwater pipeline inspection system according to claim 1, characterized in that, Texture feature analysis is performed on the received video stream data. Anomaly regions are extracted based on the texture features of the inner wall of the pipe, and key feature information containing anomaly 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 features of the pipe inner wall material, a multi-scale texture gradient map is generated through a multi-directional Gabor filter bank, and the gradient magnitude and direction distribution features are extracted. The multi-scale texture gradient map is processed based on an adaptive region growing algorithm and combined with gradient magnitude mutation threshold and direction consistency constraint to segment abnormal regions and mark the boundary coordinates and area ratio of the abnormal regions. Based on the boundary coordinates of the abnormal region, its spatiotemporal evolution trajectory in continuous video frames is extracted, and a morphological feature model based on three-dimensional curvature change and surface roughness is constructed to distinguish the geometric parameter differences between the corrosion region and the crack. Based on the geometric parameters of the morphological feature model, dynamic segmentation thresholds are set for corrosion areas and cracks respectively. The accuracy of abnormal boundaries is optimized by a genetic algorithm to generate a refined segmentation mask containing abnormal type labels. The spatial coordinates of the segmentation mask and the radio frequency positioning signal are spatiotemporally aligned. Based on the geometric parameters and evolution trajectory of the abnormal region, key feature information including location, type, size and risk level is generated and embedded into the metadata channel of the video stream.

7. The detection data transmission and processing method of an underwater pipeline inspection system according to claim 1, characterized in that, By combining cable winding and unwinding length data, a three-dimensional spatial coordinate system is constructed inside the pipeline, and key feature information is dynamically bound to the corresponding spatial coordinates to generate comprehensive inspection information, specifically: Based on the phase change rate of the radio frequency positioning signal and the interval distribution of the video stream timestamps, the non-uniformly sampled radio frequency signal and video frame sequence are aligned to generate a set of spatiotemporal synchronization reference points. Between synchronization reference points, the continuous phase trajectory of the radio frequency signal is reconstructed using non-uniform sampling cubic spline interpolation. Combined with pulse count data of cable winding and unwinding length, a pipeline motion trajectory vector containing axial displacement and radial offset is generated. The motion trajectory vector is mapped to the preset geometric model of the pipe, a three-dimensional spatial coordinate system based on the pipe centerline is constructed based on Cartesian coordinate transformation, and the spatial coordinate labels corresponding to each video frame are 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 feature points relative to the inner wall of the pipe is calculated. Based on the radial offset weighting coefficient, key feature information is spatially topologically sorted to generate comprehensive detection information that includes spatiotemporal correlation, anomaly distribution density, and risk diffusion trend. This information is then encoded as an incremental pipeline topology fingerprint and embedded into the video stream metadata.

8. The detection data transmission and processing method of an underwater pipeline inspection system according to claim 1, characterized in that, The 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 recommended remediation priorities and potential risk levels are indicated. Specifically: The abnormal location, type, and spatial coordinate labels in the comprehensive detection information are mapped to the three-dimensional mesh model of the pipeline to generate an abnormal distribution matrix containing spatiotemporal correlation weights, and record the abnormal density, type weight, and topological connection relationship of each mesh 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. Based on the pipe material properties in the high-density anomaly region, determine the potential diffusion paths of defects along the pipe's axial and radial directions in the high-density anomaly region. 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 by combining engineering constraints and using the Pareto front optimization algorithm. Based on the cross probability of anomaly type weights and potential diffusion paths, a composite risk index is determined for each anomaly region, and it is divided into three risk level labels: critical risk, high risk, and observation level according to preset thresholds. 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, generating dynamically updated inspection reports that support quick retrieval and visual backtracking by spatial coordinates or risk level.

Images