A multi-sensor fusion detection device and method for corrosion defects of an oil pipeline

Abstract

The application discloses a kind of petroleum pipeline corrosion defect multi-sensor fusion detection device and detection method, it is related to oil and gas pipeline detection technical field, including control pipeline inside detector along pipeline axial operation, synchronous acquisition magnetic flux leakage signal, uncoupling electromagnetic ultrasonic signal and high-frequency eddy current signal, and obtain the relative positioning data of odograph and inertial measurement unit;Extract the weld features in high-frequency eddy current signal as absolute position mark;Through synchronous acquisition magnetic flux leakage, electromagnetic ultrasonic and eddy current signal, utilize weld features to correct positioning error and realize data space-time alignment, introduce physical information neural network embedded physical law to carry out feature level fusion, construct the three-dimensional digital twin model of corrosion defect, to accurately calculate stress concentration coefficient and predict pipeline remaining life, effectively solve the problems of positioning drift, low multi-source data fusion degree and defect quantization inaccuracy in traditional detection, improve the precision and safety of pipeline corrosion detection.

Description

Technical Field

[0001] This invention relates to the field of oil and gas pipeline inspection technology, and in particular to a multi-sensor fusion detection device and method for detecting corrosion defects in oil pipelines. Background Technology

[0002] Oil and gas pipeline inspection technology refers to the technical means of comprehensively assessing the condition of the pipeline body, the performance of the anti-corrosion coating, and the operating environment to ensure the safe operation of the pipeline, without affecting or minimizing the impact on normal transportation. This technology mainly encompasses the use of principles such as magnetic flux leakage, ultrasonic waves, and X-rays to detect corrosion and weld defects in the pipeline body; the use of potential testing and anti-corrosion coating leak detection methods to evaluate the effectiveness of the anti-corrosion system; and the use of gas detection and negative pressure waves to monitor leaks. The aim is to promptly identify potential safety hazards and guide maintenance, thereby extending pipeline life and preventing accidents.

[0003] Existing technologies suffer from problems such as difficulty in accurately aligning multi-sensor data in time and space, and poor long-distance detection and positioning accuracy. Traditional detection methods often cannot guarantee strict synchronization of signals from different physical fields such as magnetic leakage, ultrasound, and eddy current at the same time and location. Furthermore, relying solely on wheel odometers can easily lead to cumulative errors, resulting in inaccurate defect location calibration. In addition, existing technologies mostly adopt a single data-driven mode, lack the constraints of physical laws, are easily affected by noise interference, and produce false alarms. Moreover, they can usually only output two-dimensional projections or single depth values, which cannot restore the true three-dimensional shape of the defect and cannot meet the needs of refined evaluation. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a multi-sensor fusion detection method for corrosion defects in oil pipelines. This solves the problems of existing technologies, which mostly adopt a single data-driven mode, lack physical law constraints, are easily affected by noise interference and produce false alarms, and can usually only output two-dimensional projection or a single depth value, which cannot restore the true three-dimensional shape of the defect and cannot meet the needs of refined evaluation.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a multi-sensor fusion detection method for corrosion defects in oil pipelines, comprising, The detector inside the pipeline is controlled to move along the pipeline axis, and simultaneously collect leakage magnetic field signals, uncoupled electromagnetic ultrasonic signals and high-frequency eddy current signals, and obtain the relative positioning data of the odometer and the inertial measurement unit. Weld features in high-frequency eddy current signals are extracted as absolute position markers. When the absolute position markers are detected, the cumulative error of the relative positioning data is reset, and the entire sensor array is triggered to perform millisecond-level dynamic latching to obtain multi-source raw data with spatiotemporal alignment features. Multi-source raw data is input into a pre-constructed physical information neural network model for feature-level fusion. The electromagnetic field physical laws and sound wave propagation equations embedded in the model loss function are used as physical constraints to output a multi-modal fusion feature vector. Based on multimodal fusion feature vectors, three-dimensional voxelization inversion is performed to extract the volume loss, bottom contour and surface morphology features of the defect, and to reconstruct a digital twin model of the corrosion defect in three-dimensional space. Based on the digital twin model, the geometric morphological parameters and stress concentration factor of corrosion defects are calculated, and quantitative assessment results of pipeline corrosion defects are generated.

[0007] As a preferred embodiment of the multi-sensor fusion detection method for corrosion defects in oil pipelines according to the present invention, the method involves controlling the detector inside the pipeline to move along the pipeline axis, simultaneously acquiring leakage magnetic field signals, uncoupled electromagnetic ultrasonic signals, and high-frequency eddy current signals, and obtaining the relative positioning data of the odometer and the inertial measurement unit. The specific steps are as follows: A ring array sensor assembly is deployed at the front end of the detector inside the pipeline. The ring array sensor assembly includes Hall sensors, piezoelectric transducers and eddy current coils that are uniformly distributed in the circumferential direction. The detector inside the pipeline moves axially inside the pipeline, the pipe wall is magnetized by an excitation device, and the voltage change generated by the leakage magnetic field of the pipe wall is collected by a Hall sensor to generate a leakage magnetic signal. A piezoelectric transducer is used to emit high-frequency ultrasonic waves toward the pipe wall and receive the reflected echoes to generate uncoupled electromagnetic ultrasonic signals. Eddy current field is induced on the surface of the pipe wall by using eddy current coils, and the change of eddy current field impedance is monitored to generate high-frequency eddy current signal. The number of rotations is recorded by an optical encoder installed on the detector wheel axle. The travel distance is calculated by combining the wheel diameter circumference. The triaxial acceleration and angular velocity are recorded by an inertial measurement unit. The travel distance, triaxial acceleration, and angular velocity are used as relative positioning data.

[0008] As a preferred embodiment of the multi-sensor fusion detection method for corrosion defects in oil pipelines according to the present invention, the steps are as follows: extracting weld features from high-frequency eddy current signals as absolute position markers, resetting the accumulated error of relative positioning data when the absolute position markers are detected, and triggering the entire sensor array to perform millisecond-level dynamic latching to acquire multi-source raw data with spatiotemporal alignment features. The high-frequency eddy current signal is differentiated to calculate the slope of the signal amplitude change over time. When the slope exceeds the preset weld abrupt change threshold, it is determined that a pipe circumferential weld has been detected, and an absolute position marking pulse is generated. The absolute position marker pulse is input to the central control unit. The central control unit reads the current cumulative count value of the odometer and forcibly resets the cumulative count value to zero, eliminating the cumulative error of the relative positioning data. At the moment when the absolute position marker pulse is generated, a hardware trigger command is sent to the Hall sensor, piezoelectric transducer and eddy current coil to control all sensors to complete data sampling within the same microsecond time window. The sampled magnetic leakage signal, uncoupled electromagnetic ultrasonic signal, high-frequency eddy current signal, and reset relative positioning data are all labeled with a unified timestamp tag and packaged to generate multi-source raw data with spatiotemporal alignment characteristics.

[0009] As a preferred embodiment of the multi-sensor fusion detection method for corrosion defects in oil pipelines according to the present invention, the method involves inputting multi-source raw data into a pre-constructed physical information neural network model for feature-level fusion, using the electromagnetic field physical laws and sound wave propagation equations embedded in the model loss function as physical constraints, and outputting a multi-modal fusion feature vector. The specific steps are as follows: The original data from multiple sources is divided into training and testing sets, and a multi-layer perceptron network structure containing input, hidden, and output layers is constructed. A physical residual term is introduced into the loss function of the multilayer perceptron network structure. The physical residual term consists of the magnetic vector potential equation and the acoustic wave equation derived from Maxwell's equations. Define the total loss function of the physical information neural network model. Data-driven loss With physical constraint loss The weighted sum is calculated using the following formula:

[0010]

[0011] in, Represents the total loss function. This represents the mean square error loss based on actual sensor measurement data. Represents physical constraint loss. The weighting coefficients represent the balance between data-driven factors and physical laws. Indicates the number of training samples. Indicates the first Predicted magnetic vector potential for each sample point Indicates the magnetic permeability of the pipe material. Indicates the first Current density at each sample point Indicates the first acoustic displacement field at each sample point This indicates the speed at which sound waves propagate through the pipe material. Indicates time; Minimize the total loss function using the gradient descent algorithm Update the weight parameters of the multilayer perceptron network structure until the total loss function is reached. The convergence process outputs a multimodal fusion feature vector containing information about defect depth, width, and type.

[0012] As a preferred embodiment of the multi-sensor fusion detection method for corrosion defects in oil pipelines according to the present invention, the step of performing three-dimensional voxelization inversion based on multi-modal fusion feature vectors to extract the volume loss, bottom contour, and surface morphology features of the defect, and reconstructing a digital twin model of the corrosion defect in three-dimensional space, specifically includes the following steps: Establish a three-dimensional Cartesian coordinate system for the pipe wall thickness and divide the pipe wall thickness region into several voxel units; The multimodal fusion feature vector is mapped to a three-dimensional Cartesian coordinate system. The permeability change rate of each voxel unit is obtained by using the magnetic field distortion degree corresponding to the leakage magnetic signal. The permeability change rate is then converted into the metal volume loss rate. The bottom depth coordinates of each voxel unit are obtained by using the time-of-flight difference of the acoustic waves corresponding to the uncoupled electromagnetic ultrasonic signal, and the bottom contour of the defect is determined. The surface opening area of ​​each voxel unit is obtained by using the phase angle change corresponding to the high-frequency eddy current signal, and the surface morphology of the defect is determined. The metal volume loss rate, bottom depth coordinates, and surface opening area are filled into the corresponding voxel units, and a digital twin model of the corrosion defect is generated by rendering using a volume rendering algorithm.

[0013] As a preferred embodiment of the multi-sensor fusion detection method for oil pipeline corrosion defects described in this invention, the specific steps for calculating the geometric morphological parameters and stress concentration coefficients of corrosion defects based on a digital twin model to generate quantitative assessment results of pipeline corrosion defects are as follows: Traverse the voxel units in the digital twin model and extract the maximum projected size of the corrosion defects in the axial, circumferential and radial directions as geometric morphological parameters; Based on geometric morphology parameters, a finite element analysis mesh for corrosion defects is constructed, and an internal pressure load is applied to the finite element analysis mesh. The ratio of the local peak stress at the root of the corrosion defect to the nominal circumferential stress of the pipeline is calculated and used as the stress concentration factor. The calculation formula is as follows:

[0014] in, Indicates the stress concentration factor. This represents the maximum depth of corrosion defects extracted from the digital twin model. This represents the radius of curvature at the root of the corrosion defect extracted from the digital twin model; Geometric parameters and stress concentration factor Compare with the preset pipeline safety failure threshold, when the stress concentration factor When the value exceeds the safety failure threshold, it is marked as a high-risk defect, and a data set including the defect location, size, and stress concentration factor is generated. The quantitative assessment results of pipeline corrosion defects.

[0015] As a preferred embodiment of the multi-sensor fusion detection method for corrosion defects in oil pipelines according to the present invention, the method further includes: Read the quantitative assessment results of the same pipeline location within the historical inspection cycle, and spatially register the quantitative assessment results of the same pipeline location within the historical inspection cycle with the currently generated quantitative assessment results of pipeline corrosion defects; Calculate the difference in depth between the currently generated quantitative assessment result and the quantitative assessment result of the same pipeline location in the historical inspection cycle, and combine the time interval between the two inspections to obtain the average annual expansion rate of corrosion defects. The formula for calculating the depth direction difference using the spatial registration algorithm based on weld features is as follows:

[0016] in: This represents the difference in depth between the currently generated quantitative assessment result and the quantitative assessment result of the same pipeline location within the historical monitoring cycle. This indicates the maximum depth of corrosion defects within the current inspection cycle; This indicates the maximum depth of corrosion defects at the same location on the pipeline within a historical inspection period. Based on the annual expansion rate, the remaining life of the pipeline is predicted. When the remaining life of the pipeline is lower than the preset maintenance time threshold, a pipeline maintenance early warning instruction is generated, and the detection data packet containing the annual expansion rate and digital twin model is uploaded to the remote monitoring terminal through the wireless communication module.

[0017] As a preferred embodiment of the multi-sensor fusion detection method for corrosion defects in oil pipelines according to the present invention, it includes: Multi-source data synchronous acquisition module, intelligent positioning and latching module, physical fusion and inversion module, and quantitative assessment and early warning module; The multi-source data synchronous acquisition module is used to control the detector inside the pipeline to run along the pipeline axis, synchronously acquire leakage magnetic field signals, uncoupled electromagnetic ultrasonic signals and high-frequency eddy current signals, and obtain the relative positioning data of the detector in real time through the odometer and inertial measurement unit. The intelligent positioning latching module is used to extract weld features from high-frequency eddy current signals as absolute position markers. When the absolute position marker is detected, the accumulated error of the relative positioning data is reset, and the entire sensor array is triggered to perform millisecond-level dynamic latching to obtain multi-source raw data with spatiotemporal alignment features. The physical fusion inversion module is used to input multi-source raw data into a pre-constructed physical information neural network model for feature-level fusion. It uses the electromagnetic field physical laws and sound wave propagation equations embedded in the model loss function as physical constraints to output a multi-modal fusion feature vector. Based on this feature vector, it performs three-dimensional voxelization inversion to reconstruct a digital twin model of corrosion defects. The quantitative assessment and early warning module is used to calculate the geometric morphological parameters and stress concentration coefficient of corrosion defects based on the digital twin model, generate quantitative assessment results of pipeline corrosion defects, and calculate the average annual expansion rate in combination with historical detection data. When the remaining life of the pipeline is lower than a preset threshold, a pipeline maintenance early warning instruction is generated.

[0018] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the multi-sensor fusion detection method for corrosion defects in oil pipelines as described in the first aspect of the present invention.

[0019] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the multi-sensor fusion detection method for corrosion defects in oil pipelines as described in the first aspect of the present invention.

[0020] The beneficial effects of this invention are as follows: by synchronously acquiring leakage magnetic field, electromagnetic ultrasonic and eddy current signals, using weld features to correct positioning errors and achieve data spatiotemporal alignment, introducing a physical information neural network embedded with physical laws for feature-level fusion, constructing a three-dimensional digital twin model of corrosion defects, thereby accurately calculating the stress concentration factor and predicting the remaining life of the pipeline, effectively solving the problems of positioning drift, low fusion degree of multi-source data and inaccurate defect quantification in traditional detection, and improving the accuracy and safety of pipeline corrosion detection. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart of a multi-sensor fusion detection method for corrosion defects in oil pipelines. Detailed Implementation

[0023] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0024] 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 those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0025] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0026] Reference Figure 1 As one embodiment of the present invention, this embodiment provides a multi-sensor fusion detection method for corrosion defects in oil pipelines, comprising the following steps: S1. Control the detector inside the pipeline to run along the pipeline axis, synchronously collect leakage magnetic signals, uncoupled electromagnetic ultrasonic signals and high-frequency eddy current signals, and obtain the relative positioning data of the odometer and the inertial measurement unit.

[0027] Furthermore, a ring array sensor assembly is deployed at the front end of the detector inside the pipeline. The ring array sensor assembly includes Hall sensors, piezoelectric transducers, and eddy current coils that are evenly distributed around the circumference. The detector is controlled to move axially inside the pipeline. The pipe wall is magnetized by an excitation device. The Hall sensors collect the voltage change generated by the leakage magnetic field of the pipe wall to generate a leakage magnetic signal. The piezoelectric transducer emits high-frequency ultrasonic waves into the pipe wall and receives the reflected echo to generate a coupling-free electromagnetic ultrasonic signal. The eddy current coil induces an eddy current field on the pipe wall surface, monitors the impedance change of the eddy current field, and generates a high-frequency eddy current signal. The number of rotations is recorded by an optical encoder installed on the detector's wheel axle. The travel distance is calculated by combining the wheel diameter and circumference. The triaxial acceleration and angular velocity are recorded by an inertial measurement unit. The travel distance, triaxial acceleration, and angular velocity are used as relative positioning data.

[0028] It should be noted that by deploying a ring array sensor assembly containing Hall sensors, piezoelectric transducers, and eddy current coils at the front end of the detector inside the pipeline, simultaneous sensing of multiple physical fields on the pipeline wall is achieved. Magnetic leakage signals are highly sensitive to volumetric metal losses within the pipeline wall (such as large-area corrosion and pitting), effectively reflecting the overall contour and permeability changes of defects. Uncoupled electromagnetic ultrasonic signals, which can propagate within the pipe wall without the need for acoustic coupling agents, can accurately penetrate the pipe wall and capture the depth and bottom reflection characteristics of defects, overcoming the coupling instability inherent in traditional ultrasonic testing at high speeds. High-frequency eddy current signals, utilizing the skin effect, are highly concentrated on the conductor surface, exhibiting excellent resolution for minute surface cracks and opening defects on the inner or outer surface of the pipeline. Simultaneously, by combining the travel distance, triaxial acceleration, and angular velocity obtained from the odometer and inertial measurement unit as relative positioning data, the motion attitude and trajectory of the detector can be tracked in real time in multi-dimensional space. This provides a fundamental physical coordinate reference for the spatiotemporal alignment and fusion of subsequent multi-source heterogeneous data, ensuring the continuity and integrity of the detection data during high-speed operation.

[0029] S2. Extract weld features from high-frequency eddy current signals as absolute position markers. When an absolute position marker is detected, reset the cumulative error of the relative positioning data and trigger the full sensor array to perform millisecond-level dynamic latching to obtain multi-source raw data with spatiotemporal alignment features.

[0030] Furthermore, the high-frequency eddy current signal is differentiated to calculate the slope of the signal amplitude change over time. When the slope exceeds a preset weld abrupt change threshold, a pipe circumferential weld is detected, and an absolute position marker pulse is generated. The absolute position marker pulse is input to the central control unit, which reads the current cumulative count value of the odometer and forcibly resets the cumulative count value to zero, eliminating the cumulative error of the relative positioning data. At the moment the absolute position marker pulse is generated, a hardware trigger command is sent to the Hall sensor, piezoelectric transducer, and eddy current coil to control all sensors to complete data sampling within the same microsecond time window. The sampled leakage magnetic signal, uncoupled electromagnetic ultrasonic signal, high-frequency eddy current signal, and reset relative positioning data are added with a unified timestamp tag and packaged to generate multi-source raw data with spatiotemporal alignment characteristics.

[0031] It should be noted that, due to the abrupt changes in geometry and the discontinuities in material properties, the circumferential weld of a pipeline exhibits unique impedance abrupt changes in high-frequency eddy current signals. This characteristic possesses extremely high physical uniqueness. Therefore, extracting it as an absolute position marker can fundamentally solve the problem of accumulated errors caused by unavoidable mechanical slippage, vibration, or attitude deflection of the detector within the pipeline when relying solely on odometers and inertial measurement units for relative positioning. This enables periodic absolute calibration of the detector position. Simultaneously, using this absolute position marker pulse as a hardware trigger source forces the central control unit to send a synchronization command to all sensors the instant the weld is detected. This allows the leakage magnetic field, electromagnetic ultrasonic, and eddy current sensors to complete data sampling and latching within the same microsecond time window. This dynamic latching mechanism, triggered by physical characteristics, completely eliminates spatial registration errors caused by asynchronous sampling clocks or data transmission delays among multiple sensors. It ensures strict alignment of multi-source signals for the same pipeline spatial location in both time and space, providing a high-quality, time-difference-free multi-source raw data foundation for subsequent high-precision feature-level fusion.

[0032] S3. Input the multi-source raw data into the pre-constructed physical information neural network model for feature-level fusion. Use the electromagnetic field physical laws and sound wave propagation equations embedded in the model loss function as physical constraints to output a multi-modal fusion feature vector.

[0033] Furthermore, the original multi-source data is divided into training and testing sets, and a multilayer perceptron network structure containing an input layer, hidden layers, and an output layer is constructed. A physical residual term is introduced into the loss function of the multilayer perceptron network structure, consisting of the magnetic vector potential equation derived from Maxwell's equations and the acoustic wave equation. The total loss function of the physical information neural network model is defined. Data-driven loss With physical constraint loss The weighted sum is calculated using the following formula:

[0034]

[0035] in, Represents the total loss function. This represents the mean square error loss based on actual sensor measurement data. Represents physical constraint loss. The weighting coefficients represent the balance between data-driven factors and physical laws. Indicates the number of training samples. Indicates the first Predicted magnetic vector potential for each sample point Indicates the magnetic permeability of the pipe material. Indicates the first Current density at each sample point Indicates the first acoustic displacement field at each sample point This indicates the speed at which sound waves propagate through the pipe material. Representing time; minimizing the total loss function using the gradient descent algorithm. Update the weight parameters of the multilayer perceptron network structure until the total loss function is reached. The convergence process outputs a multimodal fusion feature vector containing information about defect depth, width, and type.

[0036] It should be noted that traditional pure data-driven neural networks are prone to misjudgments due to noise interference when processing complex pipeline inspection data. This solution innovatively embeds the physical laws of electromagnetic fields and sound wave propagation directly into the model's loss function as constraints, ensuring that the artificial intelligence model follows objective physical laws while learning data features. This physical information neural network effectively filters out false defect noise that does not conform to physical characteristics, improving the model's accuracy and robustness in identifying defect depth, width, and type under complex working conditions.

[0037] S4. Based on the multimodal fusion feature vector, perform three-dimensional voxelization inversion, extract the volume loss, bottom contour and surface morphology features of the defect respectively, and reconstruct the digital twin model of the corrosion defect in three-dimensional space.

[0038] Furthermore, a three-dimensional Cartesian coordinate system for the pipe wall thickness is established, dividing the pipe wall thickness region into several voxel units. Multimodal fusion feature vectors are mapped onto the three-dimensional Cartesian coordinate system. The permeability change rate of each voxel unit is obtained using the magnetic field distortion degree corresponding to the leakage magnetic field signal, and this permeability change rate is converted into a metal volume loss rate. The bottom depth coordinates of each voxel unit are obtained using the time-of-flight difference of the uncoupled electromagnetic ultrasonic signal to determine the bottom contour of the defect. The surface opening area of ​​each voxel unit is obtained using the phase angle change corresponding to the high-frequency eddy current signal to determine the surface morphology of the defect. The metal volume loss rate, bottom depth coordinates, and surface opening area are filled into the corresponding voxel units, and a digital twin model of the corrosion defect is generated by rendering using a volume rendering algorithm.

[0039] It should be noted that by mapping multimodal fusion features to a three-dimensional Cartesian coordinate system and performing voxelization, the limitations of traditional detection methods, which can only output two-dimensional projections or single depth values, are overcome. By using different physical signals to invert the volume loss, bottom contour, and surface morphology of the defect, and reconstructing them in three-dimensional space, the true three-dimensional geometric morphology of the corrosion defect can be intuitively and three-dimensionally restored. This digital twin model not only preserves all the spatial details of the defect but also provides a high-fidelity digital foundation for subsequent accurate stress analysis and residual strength assessment.

[0040] S5. Calculate the geometric morphological parameters and stress concentration factor of corrosion defects based on the digital twin model, and generate quantitative assessment results of pipeline corrosion defects.

[0041] Furthermore, by traversing the voxel elements in the digital twin model, the maximum projected dimensions of the corrosion defects in the axial, circumferential, and radial directions are extracted as geometric morphology parameters. Based on these geometric morphology parameters, a finite element analysis mesh for the corrosion defects is constructed, and an internal pressure load on the pipeline is applied to the mesh. The ratio of the local peak stress at the root of the corrosion defect to the nominal circumferential stress of the pipeline is calculated as the stress concentration factor. The calculation formula is as follows:

[0042] in, Indicates the stress concentration factor. This represents the maximum depth of corrosion defects extracted from the digital twin model. This represents the radius of curvature at the root of the corrosion defect extracted from the digital twin model; it also includes geometric parameters and stress concentration factors. Compare with the preset pipeline safety failure threshold, when the stress concentration factor When the value exceeds the safety failure threshold, it is marked as a high-risk defect, and a data set including the defect location, size, and stress concentration factor is generated. The quantitative assessment results of pipeline corrosion defects.

[0043] It should be noted that directly calculating the geometric parameters and stress concentration coefficients of defects based on a three-dimensional digital twin model represents a leap from simple defect detection to quantitative assessment of pipeline safety. By introducing internal pipeline pressure loads and calculating the stress concentration at the defect root, the weakening effect of defects on the overall pipeline structural strength can be accurately determined. Combining historical inspection data to calculate the average annual propagation rate and predict remaining lifespan transforms pipeline maintenance strategies from traditional periodic blind inspections to precise early warning based on real defect evolution patterns, thereby improving the safety and economy of oil pipeline operations.

[0044] This embodiment also provides a computer device applicable to the multi-sensor fusion detection method for corrosion defects in oil pipelines, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the multi-sensor fusion detection method for corrosion defects in oil pipelines as proposed in the above embodiment.

[0045] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0046] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the multi-sensor fusion detection method for corrosion defects in oil pipelines as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0047] In summary, this invention, through the simultaneous acquisition of leakage magnetic field, electromagnetic ultrasonic, and eddy current signals, the use of weld features to correct positioning errors and achieve spatiotemporal data alignment, and the introduction of a physical information neural network embedded with physical laws for feature-level fusion, constructs a three-dimensional digital twin model of corrosion defects. This allows for the accurate calculation of stress concentration coefficients and prediction of the pipeline's remaining lifespan, effectively solving the problems of positioning drift, low multi-source data fusion, and inaccurate defect quantification in traditional detection methods, thereby improving the accuracy and safety of pipeline corrosion detection.

[0048] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A multi-sensor fusion detection method for corrosion defects in oil pipelines, characterized in that: include, The detector inside the pipeline is controlled to move along the pipeline axis, and simultaneously collect leakage magnetic field signals, uncoupled electromagnetic ultrasonic signals and high-frequency eddy current signals, and obtain the relative positioning data of the odometer and the inertial measurement unit. Weld features in high-frequency eddy current signals are extracted as absolute position markers. When the absolute position markers are detected, the cumulative error of the relative positioning data is reset, and the entire sensor array is triggered to perform millisecond-level dynamic latching to obtain multi-source raw data with spatiotemporal alignment features. Multi-source raw data is input into a pre-constructed physical information neural network model for feature-level fusion. The electromagnetic field physical laws and sound wave propagation equations embedded in the model loss function are used as physical constraints to output a multi-modal fusion feature vector. Based on multimodal fusion feature vectors, three-dimensional voxelization inversion is performed to extract the volume loss, bottom contour and surface morphology features of the defect, and to reconstruct a digital twin model of the corrosion defect in three-dimensional space. Based on the digital twin model, the geometric morphological parameters and stress concentration factor of corrosion defects are calculated, and quantitative assessment results of pipeline corrosion defects are generated.

2. The multi-sensor fusion detection method for corrosion defects in oil pipelines as described in claim 1, characterized in that: The detector inside the control pipeline runs along the pipeline axis, simultaneously acquiring leakage magnetic field signals, uncoupled electromagnetic ultrasonic signals, and high-frequency eddy current signals, and obtaining the relative positioning data between the odometer and the inertial measurement unit. The specific steps are as follows: A ring array sensor assembly is deployed at the front end of the detector inside the pipeline. The ring array sensor assembly includes Hall sensors, piezoelectric transducers and eddy current coils that are uniformly distributed in the circumferential direction. The detector inside the pipeline moves axially inside the pipeline, the pipe wall is magnetized by an excitation device, and the voltage change generated by the leakage magnetic field of the pipe wall is collected by a Hall sensor to generate a leakage magnetic signal. A piezoelectric transducer is used to emit high-frequency ultrasonic waves toward the pipe wall and receive the reflected echoes to generate uncoupled electromagnetic ultrasonic signals. Eddy current field is induced on the surface of the pipe wall by using eddy current coils, and the change of eddy current field impedance is monitored to generate high-frequency eddy current signal. The number of rotations is recorded by an optical encoder installed on the detector wheel axle. The travel distance is calculated by combining the wheel diameter circumference. The triaxial acceleration and angular velocity are recorded by an inertial measurement unit. The travel distance, triaxial acceleration, and angular velocity are used as relative positioning data.

3. The multi-sensor fusion detection method for corrosion defects in oil pipelines as described in claim 2, characterized in that: The process involves extracting weld features from high-frequency eddy current signals as absolute position markers, resetting the accumulated error of relative positioning data upon detection of these markers, and triggering millisecond-level dynamic latching across the entire sensor array to acquire multi-source raw data with spatiotemporal alignment features. The specific steps are as follows: The high-frequency eddy current signal is differentiated to calculate the slope of the signal amplitude change over time. When the slope exceeds the preset weld abrupt change threshold, it is determined that a pipe circumferential weld has been detected, and an absolute position marking pulse is generated. The absolute position marker pulse is input to the central control unit. The central control unit reads the current cumulative count value of the odometer and forcibly resets the cumulative count value to zero, eliminating the cumulative error of the relative positioning data. At the moment when the absolute position marker pulse is generated, a hardware trigger command is sent to the Hall sensor, piezoelectric transducer and eddy current coil to control all sensors to complete data sampling within the same microsecond time window. The sampled magnetic leakage signal, uncoupled electromagnetic ultrasonic signal, high-frequency eddy current signal, and reset relative positioning data are all labeled with a unified timestamp tag and packaged to generate multi-source raw data with spatiotemporal alignment characteristics.

4. The multi-sensor fusion detection method for corrosion defects in oil pipelines as described in claim 3, characterized in that: The process involves inputting multi-source raw data into a pre-constructed physical information neural network model for feature-level fusion. Using the electromagnetic field physics laws and sound wave propagation equations embedded in the model's loss function as physical constraints, a multi-modal fusion feature vector is output. The specific steps are as follows: The original data from multiple sources is divided into training and testing sets, and a multi-layer perceptron network structure containing input, hidden, and output layers is constructed. A physical residual term is introduced into the loss function of the multilayer perceptron network structure. The physical residual term consists of the magnetic vector potential equation and the acoustic wave equation derived from Maxwell's equations. Define the total loss function of the physical information neural network model. Data-driven loss With physical constraint loss The weighted sum is calculated using the following formula: ; ;in, Represents the total loss function. This represents the mean square error loss based on actual sensor measurement data. Represents physical constraint loss. The weighting coefficients represent the balance between data-driven factors and physical laws. Indicates the number of training samples. Indicates the first Predicted magnetic vector potential for each sample point Indicates the magnetic permeability of the pipe material. Indicates the first Current density at each sample point Indicates the first acoustic displacement field at each sample point This indicates the speed at which sound waves propagate through the pipe material. Indicates time; Minimize the total loss function using the gradient descent algorithm Update the weight parameters of the multilayer perceptron network structure until the total loss function is reached. The convergence process outputs a multimodal fusion feature vector containing information about defect depth, width, and type.

5. The multi-sensor fusion detection method for corrosion defects in oil pipelines as described in claim 4, characterized in that: The step of performing three-dimensional voxelization inversion based on multimodal fusion feature vectors to extract the volume loss, bottom contour, and surface morphology features of the defect, and reconstructing a digital twin model of the corrosion defect in three-dimensional space, is as follows: Establish a three-dimensional Cartesian coordinate system for the pipe wall thickness and divide the pipe wall thickness region into several voxel units; The multimodal fusion feature vector is mapped to a three-dimensional Cartesian coordinate system. The permeability change rate of each voxel unit is obtained by using the magnetic field distortion degree corresponding to the leakage magnetic signal. The permeability change rate is then converted into the metal volume loss rate. The bottom depth coordinates of each voxel unit are obtained by using the time-of-flight difference of the acoustic waves corresponding to the uncoupled electromagnetic ultrasonic signal, and the bottom contour of the defect is determined. The surface opening area of ​​each voxel unit is obtained by using the phase angle change corresponding to the high-frequency eddy current signal, and the surface morphology of the defect is determined. The metal volume loss rate, bottom depth coordinates, and surface opening area are filled into the corresponding voxel units, and a digital twin model of the corrosion defect is generated by rendering using a volume rendering algorithm.

6. The multi-sensor fusion detection method for corrosion defects in oil pipelines as described in claim 5, characterized in that: The steps for calculating the geometric morphological parameters and stress concentration factors of corrosion defects based on a digital twin model to generate quantitative assessment results of pipeline corrosion defects are as follows: Traverse the voxel units in the digital twin model and extract the maximum projected size of the corrosion defects in the axial, circumferential and radial directions as geometric morphological parameters; Based on geometric morphology parameters, a finite element analysis mesh for corrosion defects is constructed, and an internal pressure load is applied to the finite element analysis mesh. The ratio of the local peak stress at the root of the corrosion defect to the nominal circumferential stress of the pipeline is calculated and used as the stress concentration factor. The calculation formula is as follows: ;in, Indicates the stress concentration factor. This represents the maximum depth of corrosion defects extracted from the digital twin model. This represents the radius of curvature at the root of the corrosion defect extracted from the digital twin model; Geometric parameters and stress concentration factor The stress concentration factor is compared with the preset pipeline safety failure threshold. When the value exceeds the safety failure threshold, it is marked as a high-risk defect, and a data set including the defect location, size, and stress concentration factor is generated. The quantitative assessment results of pipeline corrosion defects.

7. The multi-sensor fusion detection method for corrosion defects in oil pipelines as described in claim 6, characterized in that: The method further includes: Read the quantitative assessment results of the same pipeline location within the historical inspection cycle, and spatially register the quantitative assessment results of the same pipeline location within the historical inspection cycle with the currently generated quantitative assessment results of pipeline corrosion defects; Calculate the difference in depth between the currently generated quantitative assessment result and the quantitative assessment result of the same pipeline location within the historical inspection cycle, and combine the time interval between the two inspections to obtain the average annual expansion rate of corrosion defects. The formula for calculating the depth direction difference using the spatial registration algorithm based on weld features is as follows: ;in: This represents the difference in depth between the currently generated quantitative assessment result and the quantitative assessment result of the same pipeline location within the historical monitoring cycle. This indicates the maximum depth of corrosion defects within the current inspection cycle; This indicates the maximum depth of corrosion defects at the same location on the pipeline within a historical inspection period. Based on the annual expansion rate, the remaining life of the pipeline is predicted. When the remaining life of the pipeline is lower than the preset maintenance time threshold, a pipeline maintenance early warning instruction is generated, and the detection data packet containing the annual expansion rate and digital twin model is uploaded to the remote monitoring terminal through the wireless communication module.

8. A multi-sensor fusion detection device for corrosion defects in oil pipelines, based on the multi-sensor fusion detection method for corrosion defects in oil pipelines according to any one of claims 1 to 7, characterized in that: include, Multi-source data synchronous acquisition module, intelligent positioning and latching module, physical fusion and inversion module, and quantitative assessment and early warning module; The multi-source data synchronous acquisition module is used to control the detector inside the pipeline to run along the pipeline axis, synchronously acquire leakage magnetic field signals, uncoupled electromagnetic ultrasonic signals and high-frequency eddy current signals, and obtain the relative positioning data of the detector in real time through the odometer and inertial measurement unit. The intelligent positioning latching module is used to extract weld features from high-frequency eddy current signals as absolute position markers. When the absolute position marker is detected, the accumulated error of the relative positioning data is reset, and the entire sensor array is triggered to perform millisecond-level dynamic latching to obtain multi-source raw data with spatiotemporal alignment features. The physical fusion inversion module is used to input multi-source raw data into a pre-constructed physical information neural network model for feature-level fusion. It uses the electromagnetic field physical laws and sound wave propagation equations embedded in the model loss function as physical constraints to output a multi-modal fusion feature vector. Based on this feature vector, it performs three-dimensional voxelization inversion to reconstruct a digital twin model of corrosion defects. The quantitative assessment and early warning module is used to calculate the geometric morphological parameters and stress concentration coefficient of corrosion defects based on the digital twin model, generate quantitative assessment results of pipeline corrosion defects, and calculate the average annual expansion rate in combination with historical detection data. When the remaining life of the pipeline is lower than a preset threshold, a pipeline maintenance early warning instruction is generated.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the multi-sensor fusion detection method for corrosion defects in oil pipelines as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the multi-sensor fusion detection method for corrosion defects in oil pipelines as described in any one of claims 1 to 7.

Images