An ai identification method, system, medium and product for pipeline defects

Abstract

The application discloses an AI identification method and system for pipeline defects, a medium and a product, and relates to the fields of artificial intelligence and image processing.The method comprises the following steps: controlling a detection robot to travel in a target pipeline in a first lighting mode, acquiring first image data and inputting the first image data into a defect identification model to obtain a defect position and a defect type; when the defect type is a preset type, preset flash parameters are called to trigger flash, and a defect preview image of the defect position is acquired; the defect preview image is input into a defect verification model corresponding to the defect type to obtain a defect existence probability; when the defect existence probability exceeds a preset probability threshold, supplementary lighting parameters are determined according to the defect type, and the detection robot is controlled to acquire second image data in a corresponding second lighting mode to determine a defect detection result; and after a defect detection report is generated based on the defect detection result, the first lighting mode is adjusted back. By implementing the application, the endurance of pipeline detection can be improved.

Description

Technical Field

[0001] This application relates to the fields of artificial intelligence and image processing, and in particular to an AI-based method, system, medium, and product for identifying pipeline defects. Background Technology

[0002] With the rapid development of urban infrastructure, underground pipe networks are becoming increasingly large and complex, making the safe operation of pipelines a crucial issue for urban management. During long-term operation, pipeline systems can experience various types of damage, such as cracks, corrosion, and deformation. If these damages are not detected and addressed promptly, they can lead to serious accidents such as pipeline leaks and ruptures, affecting the normal operation of the city and threatening public safety.

[0003] In related technologies, pipeline inspection primarily utilizes pipeline inspection robots. These robots are equipped with lighting and CCTV camera systems for internal inspection. In remote environments, the robots require power from outdoor portable power sources and use their onboard lighting and camera systems to acquire images of the pipeline's inner wall. To ensure reliable inspection, the system employs high image resolution to capture subtle changes in the pipeline's inner wall. As the inspection robot moves within the pipeline, it continuously acquires image data and transmits it in real-time to a control terminal for analysis. This technology employs an intelligent lighting control system, which adapts to different inspection scenarios through automatic exposure adjustment and light source control.

[0004] However, due to the diverse types of pipeline defects, lighting control strategies in related technologies often tend to use relatively high reference lighting intensity to ensure the detection rate of different defect types. While this lighting strategy improves detection reliability, it can cause unnecessary energy consumption when detecting pipe sections with no defects or only macroscopic defects that do not require high lighting intensity. Summary of the Invention

[0005] This application provides an AI-based method, system, medium, and product for identifying pipeline defects, which can improve the endurance of pipeline inspection.

[0006] In a first aspect, this application provides an AI-based method for identifying pipeline defects, applied to a pipeline inspection system. The method includes: controlling an inspection robot to move within a target pipeline in a first lighting mode to acquire first image data; inputting the first image data into a defect identification model to obtain the defect location and defect type of a suspected defect area; when the defect type is a preset type, calling preset flashing parameters corresponding to the defect type, and acquiring a defect preview image of the defect location while triggering flashing with the preset flashing parameters; inputting the defect preview image into a defect verification model corresponding to the defect type to obtain the probability of defect presence; when the probability of defect presence exceeds a preset probability threshold, determining supplementary lighting parameters based on the defect type, and controlling the inspection robot to acquire second image data in a corresponding second lighting mode based on the supplementary lighting parameters; inputting the first image data and the second image data into the defect identification model to obtain a defect detection result; and after generating a defect detection report based on the defect detection result, controlling the inspection robot to adjust back to the first lighting mode.

[0007] In the above embodiments, the pipeline inspection system performs routine inspections using an energy-saving first lighting mode, triggering flash preview and supplementary lighting only when a suspected defect is found; secondary confirmation is performed through a defect verification model, avoiding unnecessary supplementary lighting, and after confirming the defect, it is promptly adjusted to a dedicated second lighting mode for precise inspection; after the inspection is completed, the first lighting mode is immediately restored, achieving precise control of lighting energy consumption and significantly improving the endurance of the inspection robot.

[0008] In conjunction with some embodiments of the first aspect, in some embodiments, the step of inputting the first image data into the defect recognition model to obtain the defect location and defect type of the suspected defect region specifically includes: performing brightness and contrast standardization processing on the first image data to generate a standardized image; extracting defect features from the standardized image based on a deep convolutional neural network to determine the suspected defect region and the defect location of the suspected defect region; and identifying and classifying the suspected defect region based on a support vector machine to determine the defect type of the suspected defect region.

[0009] In the above embodiments, the pipeline inspection system performs brightness and contrast standardization processing on the image data to eliminate image differences under different lighting conditions; it uses a deep convolutional neural network to extract defect features and combines them with a support vector machine for classification and recognition, thus constructing an accurate and reliable two-stage defect recognition process, which significantly improves the accuracy and reliability of defect detection.

[0010] In conjunction with some embodiments of the first aspect, in some embodiments, after inputting the first image data and the second image data into the defect recognition model to obtain the defect detection result, the method further includes: extracting the defect type and geometric appearance parameters of the target defect from the defect detection result; calculating the defect influence coefficient of the target defect based on the defect type and geometric appearance parameters, combined with preset pipeline parameters; and calculating the defect quantitative score of the target pipeline based on the defect influence coefficients of all defects in the pre-detection section of the target pipeline that has been detected.

[0011] In the above embodiments, the pipeline inspection system calculates the defect impact coefficient based on the defect type and geometric appearance parameters, and performs a comprehensive evaluation in conjunction with pipeline parameters, thereby realizing a quantitative evaluation of the degree of pipeline defects, providing a scientific basis for pipeline maintenance decisions, and effectively guiding the rational allocation of maintenance priorities.

[0012] In conjunction with some embodiments of the first aspect, in some embodiments, after the step of calculating the defect quantification score of the target pipeline based on the defect influence coefficient of all defects in the pre-inspection section, the method further includes: when the defect quantification score is lower than a preset score threshold, calculating the third lighting mode of the inspection robot in the post-inspection section based on the defect type of all defects in the pre-inspection section; controlling the inspection robot to travel in the target pipeline in the third lighting mode and acquire images.

[0013] In the above embodiments, the pipeline inspection system dynamically adjusts the subsequent inspection strategy based on the defect quantification score of the pre-inspection section. When the defect level is large, a third lighting mode is adopted, which avoids frequent switching of lighting modes and improves inspection efficiency.

[0014] In conjunction with some embodiments of the first aspect, in some embodiments, the step of calling preset flash parameters corresponding to the defect type when the defect type is a preset type, and obtaining a defect preview image of the defect location while triggering a flash with the preset flash parameters, specifically includes: when the defect type is a preset type, determining the reflectivity and grayscale value of the suspected defect area, and calculating the minimum illumination energy threshold; determining the flash intensity and duration based on the minimum illumination energy threshold and the defect type, and generating the intensity timing sequence and triggering timing sequence of the flash illumination; triggering the flash with the intensity timing sequence and triggering timing sequence, and obtaining a defect preview image of the defect location.

[0015] In the above embodiments, the pipeline inspection system calculates the minimum illumination energy threshold based on the reflectivity and grayscale value of the defect area, and precisely controls the flash intensity and duration, which not only ensures the quality of the preview image, but also avoids over-illumination and optimizes energy utilization efficiency.

[0016] In conjunction with some embodiments of the first aspect, in some embodiments, after inputting the first image data and the second image data into the defect recognition model to obtain the defect detection result, the method further includes: collecting odometer data and image shooting angles of the detection robot during its movement; and calculating defect location data of the target defect based on the odometer data, the image shooting angles, and the defect location of the corresponding target defect in the defect detection result.

[0017] In the above embodiments, the pipeline detection system combines odometer data and image shooting angle to achieve precise location of defects, improve the accuracy of defect location marking, and provide a reliable location reference for subsequent maintenance work.

[0018] In conjunction with some embodiments of the first aspect, in some embodiments, after the step of calculating the defect location data of the target defect based on the odometer data, image shooting angle, and defect location corresponding to the target defect in the defect detection results, the method further includes: acquiring pipe images captured by the inspection robot during its movement; extracting pipe structural feature points, including joints and welds, from the pipe images; determining the point distribution characteristics of the pipe structural feature points; acquiring the radius parameters of the pipe cross-section and the material change area data along the pipe; combining the odometer data, point distribution characteristics, radius parameters, and material change area data to determine the positioning reference information of each image position in the pipe images; and correcting the defect location data based on the positioning reference information.

[0019] In the above embodiments, the pipeline network detection system establishes a complete positioning reference system by extracting pipeline structural feature points and material change information, and corrects the defect positioning data by combining multi-dimensional positioning parameters, thereby further improving the accuracy of defect positioning and providing strong support for precise maintenance.

[0020] In a second aspect, embodiments of this application provide a pipeline network detection system, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, which includes computer instructions, and the one or more processors call the computer instructions to cause the pipeline network detection system to perform the method described in the first aspect and any possible implementation thereof.

[0021] Thirdly, embodiments of this application provide a computer program product containing instructions that, when the computer program product is run on a pipeline network detection system, cause the pipeline network detection system to perform the method described in the first aspect and any possible implementation thereof.

[0022] Fourthly, embodiments of this application provide a computer-readable storage medium including instructions that, when executed on a pipeline network detection system, cause the pipeline network detection system to perform the method described in the first aspect and any possible implementation thereof.

[0023] It is understood that the pipeline monitoring system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the methods provided in the embodiments of this application. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods, and will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0025] 1. By adopting an AI-based multi-level lighting control strategy, including an energy-saving lighting mode for routine inspection, a flash preview mode for suspected defects, and a dedicated lighting mode after defect confirmation, precise control and on-demand allocation of lighting energy consumption are achieved. This effectively solves the problem of excessive energy consumption caused by the continuous use of high-intensity lighting by inspection robots in order to ensure the detection rate in existing technologies, thereby significantly improving the battery life of the inspection equipment. The system performs preliminary screening of images through a defect recognition model, triggers flash preview only when suspected defects are found, and activates the dedicated lighting mode only after secondary confirmation by a defect verification model. This multi-level linkage lighting control mechanism ensures inspection quality while maximizing energy savings.

[0026] 2. By adopting a quantitative evaluation mechanism based on defect characteristics, including a comprehensive analysis of defect types, geometric appearance parameters, and pipeline parameters, a scientific pipeline condition evaluation system has been established. This effectively solves the problem of a lack of unified evaluation standards in existing technologies, which leads to a lack of basis for maintenance decisions. As a result, maintenance resources are allocated more rationally and maintenance efficiency is improved. By calculating the impact coefficient of each defect and combining it with pipeline characteristics for weighted calculation, a quantitative score that objectively reflects the overall condition of the pipeline is obtained, providing a reliable basis for prioritizing pipeline maintenance and formulating maintenance plans.

[0027] 3. Due to the adoption of an intelligent flash parameter control strategy, including illumination energy calculation based on the reflection characteristics of the defect area and a precise timing control mechanism, the flash illumination effect is optimized. This effectively solves the problem of energy waste or unstable image quality caused by fixed flash illumination parameters in existing technologies, thereby achieving high efficiency and high quality in preview image acquisition. By analyzing the reflectivity and grayscale characteristics of the defect area, the minimum required illumination energy is calculated, and the intensity and duration of the flash are precisely controlled accordingly, ensuring that a clear defect preview image is obtained with minimal energy consumption. Attached Figure Description

[0028] Figure 1 This is a flowchart illustrating an AI-based method for identifying pipeline defects in an embodiment of this application.

[0029] Figure 2 This is another flowchart illustrating the AI-based method for identifying pipeline defects in this application.

[0030] Figure 3 This is a schematic diagram of the physical device structure of a pipeline network detection system in the embodiments of this application. Detailed Implementation

[0031] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification of this application, the singular expressions “a,” “an,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to any or all possible combinations including one or more of the listed items.

[0032] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0033] To facilitate understanding, the application scenarios of the embodiments of this application are described below.

[0034] A city is currently conducting a large-scale survey of its underground pipe network. Traditional pipe network inspection methods mainly rely on manual visual inspection or simple camera recording, which are inefficient and prone to missing defects. Especially in pipe environments with complex lighting conditions, inspectors struggle to accurately identify early defects such as minute cracks and corrosion spots on the pipe walls. Furthermore, due to the lack of standardized inspection procedures and intelligent analysis tools, inspection results often depend on the inspectors' experience and judgment, resulting in strong subjectivity and poor repeatability. In addition, the frequent adjustment of lighting equipment during the inspection process not only increases working time but also easily leads to missed or false detections, failing to meet the efficiency and accuracy requirements of large-scale pipe network surveys.

[0035] In related technologies, pipe inner wall defects can be detected using inspection robots with fixed lighting. This method requires manual observation of images and defect assessment in real time, resulting in low detection efficiency and susceptibility to subjective factors. Furthermore, because the lighting parameters are fixed, it is difficult to adapt to the imaging requirements of different types of defects, easily leading to missed detections and false positives. The following describes a scenario using AI-based pipe defect recognition methods from related technologies.

[0036] A certain inspection unit uses an inspection robot with fixed lighting for pipeline inspection. The robot is equipped with an LED ring light that provides constant power illumination and captures pipeline images through a high-definition camera. Inspectors view the images in real time via a remote control system, marking suspicious areas for focused observation. However, in actual inspections, it was found that the fixed lighting mode is difficult to adapt to the imaging requirements of different types of defects. For example, for crack-like defects, the existing lighting intensity is often insufficient to highlight minute crack features; while for corrosion-like defects, uniform lighting easily causes reflection interference, affecting the identification of defect boundaries. Due to the lack of intelligent analysis and adaptive control capabilities, inspectors need to frequently manually adjust camera parameters, significantly reducing inspection efficiency.

[0037] The AI-based pipeline defect recognition method described in this application combines preliminary scanning with a first illumination mode with fine imaging using targeted flash illumination, achieving highly efficient and accurate defect detection. This not only significantly improves detection efficiency but also effectively reduces energy consumption. Particularly in complex pipeline environments, the system can adaptively adjust illumination parameters based on defect characteristics to ensure the acquisition of optimal quality defect images. The following describes scenarios where the AI-based pipeline defect recognition method of this application is used.

[0038] A pipeline operation unit uses the intelligent detection system described in this application for pipeline inspection. The system first performs a preliminary scan using an energy-saving primary lighting mode. When a suspected defect is detected, it automatically switches to a targeted flash mode for detailed imaging. For example, upon detecting a suspected crack, the system immediately calculates the optimal flash parameters and triggers the flash under precisely controlled timing, successfully capturing clear crack details. Simultaneously, the system's deep learning model automatically analyzes image features, accurately identifies the defect type, and assesses its severity. This intelligent detection method not only improves detection efficiency but also significantly reduces the missed detection rate, providing reliable data support for pipeline maintenance decisions.

[0039] As can be seen, the AI-based pipeline defect identification method in this application not only achieves intelligent defect detection but also effectively solves the problem of unstable image quality under traditional fixed lighting, thereby significantly improving detection efficiency and accuracy. By using a deep learning model to accurately analyze defect features and combining adaptive lighting control, the objectivity and reliability of the detection results are ensured.

[0040] To facilitate understanding, the method provided in this implementation will be described in detail below, using the above scenario as an example. Please refer to [link / reference]. Figure 1 This is a flowchart illustrating an AI-based method for identifying pipeline defects in an embodiment of this application.

[0041] S101. Control the detection robot to move inside the target pipe in the first lighting mode and acquire the first image data.

[0042] The inspection robot refers to an autonomous mobile device used to travel inside a pipe and perform inspections, including a drive system, a lighting system, and an image acquisition system; the first lighting mode refers to the basic lighting configuration used by the inspection robot, mainly including parameters such as lighting intensity, lighting angle, and energy consumption; the target pipe refers to the pipe segment to be inspected; and the first image data refers to the sequence of images of the inner wall of the pipe acquired by the inspection robot in the first lighting mode.

[0043] When a pipeline inspection system initiates an inspection task, it needs to perform a preliminary scan of the target pipeline to obtain basic image data. Specifically, the pipeline inspection system first controls the inspection robot to enter the inlet of the target pipeline, activates the LED light source in the first lighting mode, and sets the basic lighting intensity (usually 200-300 lumens) and lighting angle (usually 120-degree fan-shaped illumination) suitable for routine inspection. The inspection robot moves inside the pipeline at a stable speed (usually 0.1-0.2 meters per second), while continuously acquiring images of the pipeline's inner wall through its onboard high-definition camera, generating first image data with a resolution of no less than 1920×1080 pixels.

[0044] In some embodiments, lighting control and image acquisition of the inspection robot can be implemented in various ways: Optionally, the pipeline inspection system automatically adjusts the parameter configuration of the first lighting mode according to the pipe diameter and material characteristics, including: firstly acquiring basic pipe parameter information, then calculating the optimal lighting distance and illumination angle, then determining the lighting intensity based on the pipe wall reflection characteristics, and finally generating a lighting control command sequence. Optionally, the pipeline inspection system adopts an adaptive exposure control strategy, including: real-time monitoring of image brightness distribution, dynamically adjusting camera exposure parameters, and synchronously optimizing lighting output power to ensure stable image quality. It is understood that other methods can also be used to implement motion control and data acquisition of the inspection robot, such as dynamically adjusting the travel speed according to the pipe curvature, etc., which are not limited here.

[0045] In practical applications, the inspection robot may experience reduced inspection results due to insufficient battery power. To address this, the pipeline inspection system employs an intelligent energy-saving strategy: real-time monitoring of battery level and remaining inspection mileage dynamically calculates the maximum allowable lighting power; prediction of lighting needs for subsequent pipeline sections based on historical inspection data, and rational allocation of remaining power; and automatic reduction of energy consumption for unnecessary lighting when insufficient power is detected, ensuring the completion of the designated inspection tasks. When power is sufficient, the system appropriately increases lighting intensity to improve image quality.

[0046] S102. Input the first image data into the defect recognition model to obtain the defect location and defect type of the suspected defect area.

[0047] Among them, the defect identification model refers to the deep learning model used to analyze pipeline images and identify defects, including an image preprocessing module, a feature extraction module, and a classification and recognition module; the suspected defect area refers to the image area that may have defects initially identified in the first image data; the defect location refers to the spatial coordinate information of the suspected defect area in the pipeline; and the defect type refers to the type of defect initially judged based on the defect morphology characteristics, such as cracks, corrosion, deformation, etc.

[0048] After acquiring the initial image data, the pipeline network detection system needs to perform preliminary defect detection and classification. Specifically, the system first preprocesses the initial image data, including image enhancement, noise removal, and scale normalization. Then, the processed image is input into a pre-trained defect recognition model. This model extracts image features through a multi-layer convolutional neural network, uses an attention mechanism to locate suspected defect regions, and finally outputs the location coordinates and type probability distribution of each suspected defect region through a fully connected layer.

[0049] In some embodiments, defect identification and localization can be achieved in multiple ways: Optionally, the pipeline network detection system employs a two-stage detection strategy, including: first, generating candidate regions using a region proposal network; then, using a classification network to perform fine-grained classification of the candidate regions; and finally, using a non-maximum suppression algorithm to filter out the final suspected defect regions. Optionally, the pipeline network detection system combines a multi-scale feature fusion method, including: constructing a feature pyramid network, extracting feature maps at different scales, and enhancing the model's detection capability through feature fusion to achieve accurate identification of defects of different sizes. It is understood that other methods can also be used to achieve defect detection and classification, such as introducing a domain adaptation module to handle the feature differences of different pipe materials, etc., which are not limited here.

[0050] It should be noted that this defect recognition model is trained using a large dataset of labeled pipe images, including images of normal pipes and images with different types of defects. Each image in the training data is labeled with the precise location coordinates, bounding box, defect type, and severity of the defect. Training employs an end-to-end supervised learning approach, using the cross-entropy loss function to evaluate classification error and IoU (Intersection over Union) and mean AP (mAP) to evaluate localization accuracy. Network parameters are optimized using backpropagation until the model's performance on the validation set meets preset requirements. The model employs a two-stage detection architecture. The first stage is a feature extraction network based on a deep convolutional neural network, extracting multi-scale features from the image through multi-layer convolutional and pooling operations. The second stage includes a region proposal network and a classification network. The region proposal network generates candidate defect regions, while the classification network performs fine-grained classification and location regression on these regions. The model also integrates an attention mechanism, enabling it to adaptively focus on key regions in the image. In actual detection, the model receives standardized pipe images as input and outputs the location coordinates, bounding box, type probability distribution, and confidence score for each detected defect region. For each identified defect, the model also outputs geometric feature parameters for subsequent severity assessment. The model's inference process supports real-time processing, enabling rapid analysis of the image stream acquired by the inspection robot.

[0051] In practical applications, the complexity of pipeline environments can lead to false positives in initial identification results. To address this, the pipeline inspection system employs a multi-feature verification mechanism: a defect feature library is established based on pipeline material characteristics, and the reliability of the identification results is verified through feature matching; an anomaly detection model is trained using historical inspection data to filter out identification results that do not conform to statistical patterns; when a suspicious identification result is detected, a subsequent refined inspection process is triggered to ensure the accuracy of the detection results.

[0052] S103. When the defect type is a preset type, call the preset flash parameters corresponding to the defect type, and obtain the defect preview image at the defect location while triggering the flash with the preset flash parameters.

[0053] Among them, the preset type represents the set of defect types that need to be finely inspected; the preset flash parameters represent the flash illumination parameters pre-configured for different defect types, including flash intensity, duration and triggering sequence; and the defect preview image represents a high-quality local image of the defect acquired under flash illumination.

[0054] After initially identifying suspected defects, the pipeline inspection system needs to perform detailed imaging of the localized areas to obtain more details about the defects. Specifically, when the identified defect type belongs to a preset type, the pipeline inspection system calls the corresponding flash parameter configuration, controls the flash of the inspection robot to be triggered at a specified time sequence, and simultaneously captures instantaneous images of the defect area through a high-speed camera module to ensure clear defect details are obtained under optimal lighting conditions.

[0055] In some embodiments, flash control and image acquisition can be achieved in various ways: Optionally, the pipeline inspection system employs an intelligent exposure control strategy, including: real-time analysis of the reflection characteristics of the defect area, dynamic adjustment of flash intensity and duration, optimization of exposure parameter configuration, and ensuring image quality. Optionally, the pipeline inspection system uses a multi-flash combination scheme, including: designing multiple flash sequences, acquiring multiple images under different lighting conditions, and generating a high dynamic range defect image through image fusion technology. It is understood that other methods can also be used to achieve fine imaging, such as combining light field camera technology to obtain depth information of the defect, etc., which are not limited here.

[0056] In practical applications, flash illumination can cause energy loss and acquisition delay. To address this, the pipeline inspection system implements intelligent trigger control: establishing a mapping relationship between defect types and optimal flash parameters to avoid over-illumination; using predictive algorithms to pre-calculate flash timing to reduce the waiting time of the inspection robot; and, under energy-constrained conditions, rationally allocating flash resources through a priority management mechanism to ensure the precise inspection of critical defects.

[0057] S104. Input the defect preview image into the defect verification model corresponding to the defect type to obtain the probability of defect existence.

[0058] Among them, the defect verification model refers to a deep learning model specifically designed to verify a particular type of defect, which is trained separately for different defect types; the defect existence probability represents the confidence score of the presence of the target type defect in the defect preview image, with a value ranging from 0 to 1; and the defect type correspondence represents the mapping relationship between different defect types and their dedicated verification models.

[0059] After acquiring defect preview images, the pipeline inspection system needs to perform precise defect verification. Specifically, the system selects a corresponding dedicated verification model based on the identified defect type and inputs the defect preview image into the model for analysis. The verification model calculates the probability value of the existence of a specified type of defect in the target area through fine-grained feature extraction and multi-dimensional feature analysis. This probability value reflects the reliability of the defect identification result and is used for dynamic adjustment of subsequent detection strategies.

[0060] In some embodiments, defect verification can be implemented in multiple ways: Optionally, the pipeline network inspection system employs an ensemble learning method, including: deploying multiple verification models with different structures, calculating the defect probability separately, and obtaining the final existence probability through weighted fusion, thereby improving the reliability of the verification results. Optionally, the pipeline network inspection system combines knowledge distillation technology, including: extracting general features using a large-scale pre-trained model, transferring knowledge to a lightweight verification model, thereby improving verification efficiency while ensuring accuracy. It is understood that other methods can also be used to implement defect verification, such as introducing an expert rule system to assist in verification, etc., which are not limited here.

[0061] It should be noted that the validation model is trained separately for each specific defect type, using high-quality defect image samples verified by experts. Training labels include the precise outline of the defect, internal structural features, and severity rating. The training process places particular emphasis on model robustness, simulating defect performance under different lighting conditions and imaging angles through data augmentation techniques. Precision-recall curves (PR curves) and F1 scores are used as evaluation criteria to ensure the model has high specificity. The model employs a lightweight deep learning network architecture, focusing on extracting discriminative features for specific defect types. The network includes a feature extraction module and a validation module. The feature extraction module uses depthwise separable convolutions to reduce computation, while the validation module calculates the probability of defect presence using a multilayer perceptron. The model can also assess the uncertainty of prediction results. The validation model receives a local image of a suspected defect acquired under flash preview as input, outputting the probability value of the target type defect in that area and an estimate of the prediction uncertainty. The system only triggers subsequent fine-tuning detection when the predicted probability exceeds a preset threshold and the uncertainty is low. The model's prediction results are also used to optimize flash parameters and supplementary lighting strategies.

[0062] In practical applications, the differences in characteristics among different types of defects may affect the accuracy of verification. To address this, the pipeline inspection system adopts an adaptive verification strategy: it constructs a defect feature knowledge base containing typical feature descriptions of different types of defects; it dynamically adjusts feature extraction parameters according to defect type to highlight key features; and it automatically adjusts the verification threshold when feature matching is insufficient to ensure the reliability of verification results.

[0063] S105. When the probability of a defect exists exceeds a preset probability threshold, supplementary lighting parameters are determined according to the defect type, and the detection robot is controlled to acquire second image data in the corresponding second lighting mode based on the supplementary lighting parameters.

[0064] Among them, the preset probability threshold represents the confidence threshold value for triggering fine inspection; the supplementary lighting parameters represent the fine inspection lighting configuration determined according to the defect type, including lighting intensity, angle distribution and energy control, etc.; the second lighting mode represents the high-quality lighting scheme used for fine inspection.

[0065] After completing defect verification, the pipeline inspection system needs to perform detailed inspection of high-probability defects. Specifically, when the probability of a defect's presence exceeds a preset threshold, the system calculates the optimal supplementary lighting parameters based on the defect type's detection requirements. These parameters determine the specific configuration of the second lighting mode, including light source selection, intensity distribution, and illumination angle. The inspection robot switches to the second lighting mode based on these parameters to acquire high-quality image data.

[0066] In some embodiments, fine-grained inspection can be achieved in several ways: Optionally, the pipeline inspection system uses a multi-source collaborative illumination strategy, including: analyzing the geometric features of defects, planning the spatial layout of multiple light sources, and highlighting defect features through phase coordination between light sources to achieve comprehensive fine imaging. Optionally, the pipeline inspection system employs adaptive illumination control, including: real-time monitoring of image quality indicators, dynamic adjustment of illumination parameters, and establishment of a feedback optimization mechanism to ensure optimal acquisition of image data. It is understood that other methods can also be used to achieve fine-grained inspection, such as combining structured light technology to obtain three-dimensional information about defects, etc., which are not limited here.

[0067] In practical applications, different defects have different lighting requirements. To address this, the pipeline inspection system implements intelligent lighting configuration: establishing a mapping relationship between defect types and optimal lighting schemes; dynamically adjusting lighting strategies based on defect characteristics; and ensuring the detection quality of critical defects through lighting resource scheduling when energy is limited.

[0068] S106. Input the first image data and the second image data into the defect recognition model to obtain the defect detection result.

[0069] Among them, the defect detection result represents the final defect judgment information obtained after comprehensively analyzing the first image data and the second image data, including the precise location, type, severity and geometric parameters of the defect; the data fusion analysis represents the process of co-processing the image data acquired under the two lighting modes; and the detection credibility represents the reliability evaluation index of the final detection result.

[0070] After acquiring image data under two lighting modes, the pipeline inspection system needs to perform comprehensive analysis to draw a final conclusion. Specifically, the system first registers and preprocesses the two sets of image data to ensure the accuracy of spatial correspondence. Then, the processed image data is simultaneously input into the defect recognition model. The model, through multi-scale feature extraction and cross-modal feature fusion, fully utilizes complementary information under different lighting conditions to ultimately output highly reliable defect detection results. The detection results include not only basic defect information but also detailed geometric parameters and damage assessment data.

[0071] In some embodiments, data fusion analysis can be achieved in multiple ways: Optionally, the pipeline network detection system employs a hierarchical fusion strategy, including: image alignment and enhancement at the pixel level, multi-dimensional feature extraction and fusion at the feature level, and integration of detection results from multiple sub-models at the decision level to generate the final defect judgment. Optionally, the pipeline network detection system uses an attention-guided fusion method, including: calculating the importance weights of image regions under different lighting conditions, guiding the feature fusion process based on the weights, and highlighting the contribution of key information. It is understood that other methods can also be used to achieve result fusion, such as introducing a temporal analysis module to process the correlation information between consecutive frames, etc., which are not limited here.

[0072] In practical applications, differences in image quality under different lighting modes can affect the fusion effect. To address this, the pipeline inspection system implements adaptive fusion control: an image quality assessment mechanism is established to score the reliability of data from different sources; fusion weights are dynamically adjusted based on the quality scores; and when the data quality of a certain mode is insufficient, the weight contribution of other data sources is increased to ensure the stability of the fusion result.

[0073] S107. After generating a defect detection report based on the defect detection results, control the detection robot to switch back to the first lighting mode.

[0074] Among them, the defect detection report is a standardized document that records the detection process and results, including defect information, detection parameters, evaluation conclusions, etc.; the lighting mode adjustment indicates the process of the detection robot returning to the energy-saving basic lighting state; and the detection task status indicates the management information of the current detection progress and subsequent detection plans.

[0075] After completing the detection and analysis of the current defects, the pipeline inspection system needs to adjust the lighting mode in a timely manner and prepare for subsequent inspections. Specifically, the pipeline inspection system first generates a standard-format inspection report based on the inspection results, containing a complete description of the defects and assessment recommendations. After the report is generated, the system controls the inspection robot to reduce the lighting power, restoring it to the lower-energy-consumption first lighting mode, in preparation for the inspection of subsequent pipe sections. This dynamic lighting mode adjustment strategy can effectively balance inspection effectiveness and energy consumption.

[0076] In some embodiments, mode switching management can be implemented in multiple ways: Optionally, the pipeline network inspection system adopts a gradual adjustment strategy, including: calculating the difference between the current lighting parameters and the target parameters, designing a smooth transition curve, and avoiding the impact of sudden changes on image acquisition through step-by-step adjustments to ensure inspection continuity. Optionally, the pipeline network inspection system uses a predictive adjustment method, including: analyzing the pipeline characteristics of subsequent inspection sections, planning the timing of lighting mode switching in advance, and achieving optimal allocation of lighting resources. It is understood that other methods can also be used to achieve lighting adjustment, such as dynamically optimizing the lighting strategy based on the remaining battery power, etc., which are not limited here.

[0077] In practical applications, frequent lighting mode switching can affect system stability. To address this, the pipeline monitoring system implements intelligent mode management: it establishes a caching mechanism for lighting mode switching to avoid repeated switching within a short period; it monitors the working status of each system component to ensure coordinated operation during mode switching; and when an abnormal situation is detected, it automatically activates protection strategies to ensure the safe and reliable operation of the monitoring equipment.

[0078] The above embodiments mainly describe the basic lighting control and defect detection process. In practical applications, the system can also optimize the lighting strategy based on historical detection data, establish a correlation model between defect features and lighting parameters, and achieve a more intelligent detection process. Simultaneously, through multi-source data fusion and positioning calibration technology, the accuracy of defect location is further improved. The following section supplements the scenario described in this embodiment.

[0079] This intelligent inspection system has continuously accumulated experience and optimized its performance through long-term operation. By analyzing historical inspection data, the system has established a mapping relationship between defect characteristics and optimal lighting parameters, enabling it to automatically predict the best lighting strategy based on pipe material and environmental conditions. For example, when a change in pipe material is detected, the system smoothly adjusts the lighting parameters to ensure the continuity of inspection quality. Furthermore, the system can dynamically plan inspection paths based on defect distribution characteristics, prioritizing the inspection of high-risk areas. Through this continuously optimized intelligent inspection approach, not only has inspection efficiency been further improved, but optimal allocation of inspection resources has also been achieved, laying the foundation for intelligent operation and maintenance of large-scale pipeline networks.

[0080] In light of the above scenarios, the method provided in this implementation will now be described in more detail. Please refer to [link / reference]. Figure 2 This is another flowchart illustrating the AI-based method for identifying pipeline defects in this application.

[0081] S201. Control the detection robot to move inside the target pipe in the first lighting mode and acquire the first image data.

[0082] Referring to step S101, the pipeline inspection system will control the inspection robot to enter the target pipeline, activate the energy-saving first lighting mode, continuously move inside the pipeline and collect the first image data of the pipeline inner wall.

[0083] S202. Input the first image data into the defect recognition model to obtain the defect location and defect type of the suspected defect area.

[0084] Referring to step S102, the pipeline inspection system will input the collected first image data into the pre-trained defect recognition model, analyze and process the image, identify the location of areas where defects may exist, and preliminarily determine the type of defect.

[0085] In some embodiments, the pipeline inspection system employs a two-stage image processing and defect identification process to improve detection accuracy. Specifically, the pipeline inspection system performs brightness and contrast standardization processing on the first image data to generate a standardized image. Based on a deep convolutional neural network, defect features are extracted from the standardized image to determine suspected defect areas and their defect locations. Based on a support vector machine, the suspected defect areas are identified and classified to determine their defect types.

[0086] Here, standardization refers to the preprocessing of normalizing the brightness and contrast of an image; standardized image refers to an image with uniform brightness and contrast characteristics after preprocessing; deep convolutional neural network refers to a multi-layer neural network model used for image feature extraction; suspected defect region refers to an image region that may have defects initially identified; support vector machine refers to a machine learning model used for defect type classification; and defect type refers to the defect type label obtained after classifying the suspected defect region.

[0087] After acquiring the initial image data, the pipeline network inspection system requires multi-stage intelligent analysis and processing. Specifically, the system first performs standardization preprocessing on the image data, converting images acquired under different lighting conditions into standardized images with consistent features through methods such as histogram equalization and gamma correction. Then, a pre-trained deep convolutional neural network model is used to extract features from the standardized images. Multi-scale features are extracted through multi-layer convolution operations and pooling operations, combined with an attention mechanism to locate suspected defect areas. Finally, the extracted features are input into a support vector machine classifier, which accurately classifies suspected defect areas based on trained decision boundaries and outputs defect type labels.

[0088] In some embodiments, image analysis and defect recognition can be achieved in various ways: Optionally, the pipeline network detection system employs a multi-stage processing strategy, including: firstly, adaptive threshold segmentation to obtain preliminary candidate regions; then, morphological operations to optimize region boundaries; next, extraction of multi-dimensional feature vectors; and finally, ensemble learning methods to improve classification accuracy. Optionally, the pipeline network detection system uses deep learning enhancement methods, including: constructing a feature pyramid network to extract multi-scale features; applying a region proposal network to generate candidate boxes; accurately locating defect regions through an instance segmentation network; and combining transfer learning to improve the model's generalization ability. It is understood that other methods can also be used to achieve image analysis, such as hybrid methods combining traditional image processing algorithms and deep learning models, etc., which are not limited here.

[0089] In practical applications, the complex lighting conditions inside pipelines can lead to unstable image preprocessing results. To address this, the pipeline inspection system implements an adaptive preprocessing strategy: establishing an image quality assessment model to evaluate the preprocessing effect in real time; dynamically adjusting processing parameters based on image features; and employing a multi-scale fusion method to improve processing robustness when preprocessing anomalies are detected. For example, the system automatically reduces contrast enhancement for highly reflective areas and increases local enhancement intensity for dark details.

[0090] It should be noted that the training data for this image quality assessment model comes from a large number of pipeline detection images with expert quality scores. Training labels include objective metrics such as image sharpness, contrast, and signal-to-noise ratio, as well as a comprehensive subjective score. The training process employs a multi-task learning approach, simultaneously optimizing multiple quality assessment metrics. Mean squared error (MSE) and rank consistency are used as loss functions to ensure that the model's evaluation results are consistent with human judgment. The model adopts a multi-branch network structure, including an image feature extraction branch and a quality assessment branch. The feature extraction branch uses a lightweight convolutional network to extract multi-dimensional features of the image, while the quality assessment branch calculates various quality metrics through fully connected layers. The model integrates traditional image quality assessment algorithms, such as sharpness assessment and noise estimation, forming an end-to-end quality assessment system. During detection, the model evaluates the quality of the acquired images in real time. The input is the original image, and the output includes quality scores and a comprehensive score across multiple dimensions. These evaluation results are used to dynamically adjust camera and lighting parameters to ensure image acquisition quality. When an image quality anomaly is detected, the model triggers a re-acquisition mechanism to ensure the reliability of the detection data.

[0091] S203. When the defect type is a preset type, determine the reflectivity and grayscale value of the suspected defect area and calculate the minimum lighting energy threshold.

[0092] Among them, reflectivity represents the ability of the pipe surface to reflect light; grayscale value represents the brightness level of a pixel in the image; minimum illumination energy threshold represents the minimum light energy level required to obtain a clear image of a defect; and optical property parameters represent physical quantities that describe the material's response characteristics to light.

[0093] After confirming the defect type, the pipeline inspection system needs to evaluate the optical characteristics of the target area to optimize the illumination strategy. Specifically, the system first analyzes image data of the suspected defect area, extracting surface reflectivity features and grayscale distribution. Then, based on the imaging requirements of the defect type and the optical properties of the material, it calculates the minimum illumination energy threshold that can highlight the defect features. This threshold serves as a benchmark for subsequent flash parameter settings, ensuring that over-illumination is avoided while meeting inspection requirements.

[0094] In some embodiments, optical property evaluation can be achieved in multiple ways: Optionally, the pipeline network inspection system employs a regional adaptive analysis method, including: dividing multiple local regions to calculate reflection characteristics, constructing a reflectivity gradient map, identifying material variation boundaries, and generating a regional optical property distribution map. Optionally, the pipeline network inspection system uses multi-scale photometric measurement technology, including: analyzing surface optical properties at different spatial scales, establishing a hierarchical property description model, and extracting the main influencing factors. It is understood that other methods can also be used to achieve property evaluation, such as introducing spectral analysis technology to evaluate the wavelength response characteristics of materials, etc., which are not limited here.

[0095] In practical applications, the complexity of pipeline surface conditions can affect the accuracy of optical property assessment. To address this, the pipeline inspection system implements a robust assessment strategy: establishing a surface condition classification model to identify influencing factors such as contamination and scaling; dynamically adjusting assessment parameters based on surface conditions; and employing a conservative strategy to set illumination parameters when assessment results are unstable, ensuring detection reliability.

[0096] S204. Determine the flash intensity and duration based on the minimum illumination energy threshold and defect type, and generate the intensity timing and triggering timing of the flash illumination.

[0097] Among them, flash intensity represents the instantaneous power level of flash illumination; duration represents the illumination period of a single flash; intensity timing represents the curve of flash intensity changing over time; and trigger timing represents the time control sequence for flash triggering.

[0098] After obtaining the minimum illumination energy threshold, the pipeline inspection system needs to design a detailed flash illumination scheme. Specifically, the system first calculates the flash intensity value that meets the imaging requirements based on the minimum illumination energy threshold and the characteristics of the defect type. Then, it determines the suitable illumination duration for this type of defect and generates a complete timing scheme that includes intensity changes and triggering times. This scheme needs to ensure precise synchronization between flash illumination and image acquisition to obtain the best quality defect images.

[0099] In some embodiments, flash control can be implemented in several ways: Optionally, the pipeline inspection system employs a multi-level intensity control strategy, including designing a three-segment intensity curve of preheating-main illumination-afterglow to optimize energy utilization efficiency and improve image quality. Optionally, the pipeline inspection system uses an intelligent trigger optimization method, including analyzing the motion state of the inspection robot, predicting the optimal triggering time, and achieving motion compensation and image stabilization. It is understood that other methods can also be used to implement flash control, such as dynamically adjusting flash parameters according to ambient temperature, etc., which are not limited here.

[0100] In practical applications, the timing control accuracy of flash illumination directly affects image quality. To address this, the pipeline inspection system implements high-precision timing management: establishing a hardware-level synchronization triggering mechanism to ensure microsecond-level synchronization between flash illumination and image acquisition; monitoring the response characteristics of the flash device and compensating for device delays; and automatically replanning the timing scheme when a synchronization anomaly is detected.

[0101] S205. Trigger the flash with intensity timing and trigger timing, and obtain a defect preview image of the defect location.

[0102] Among them, intensity timing trigger means that the flash intensity control is executed according to a preset time sequence; trigger timing execution means that the flash device is activated according to the planned time point; and defect preview image means that a high-quality local image is obtained under precisely controlled flash illumination conditions.

[0103] After completing the flash parameter design, the pipeline inspection system needs to precisely execute flash control and acquire images. Specifically, the system first loads the designed intensity and trigger timing sequences into the flash control module to ensure readiness. Then, when the inspection robot reaches the designated position, it executes flash triggering according to the timing scheme, simultaneously initiating high-speed image acquisition to capture transient images of the defect area under optimal illumination conditions. The entire process requires microsecond-level control precision to ensure precise synchronization between flash illumination and image acquisition.

[0104] In some embodiments, accurate imaging can be achieved in several ways: Optionally, the pipeline network detection system employs multi-frame synthesis technology, including: acquiring multiple frames of images at high speed during a single flash, and using super-resolution algorithms for image reconstruction to improve detail resolution. Optionally, the pipeline network detection system uses real-time feedback control, including: monitoring flash output characteristics, dynamically adjusting control parameters, compensating for the influence of environmental factors, and ensuring stable illumination effects. It is understood that other methods can also be used to achieve accurate imaging, such as combining light field camera technology to obtain depth information, etc., which are not limited here.

[0105] In practical applications, the instability of flash illumination can lead to fluctuations in image quality. To address this, the pipeline inspection system implements an adaptive control strategy: establishing a flash characteristic model to predict device response behavior; dynamically compensating for fluctuations based on real-time monitoring data; and automatically adjusting control parameters or triggering backup plans when abnormal responses are detected.

[0106] S206. Input the defect preview image into the defect verification model corresponding to the defect type to obtain the probability of defect existence.

[0107] Referring to step S104, the pipeline inspection system will input the acquired defect preview image into a verification model specifically trained for this type of defect, and analyze the image features through deep learning algorithms to calculate the probability value of the presence of the target type defect in the area.

[0108] S207. When the probability of a defect exists exceeds a preset probability threshold, supplementary lighting parameters are determined according to the defect type, and the detection robot is controlled to acquire second image data in the corresponding second lighting mode based on the supplementary lighting parameters.

[0109] Referring to step S105, when the probability of a defect exceeds a preset threshold, the pipeline inspection system will set appropriate supplementary lighting parameters according to the inspection requirements of the defect type and control the inspection robot to switch to the corresponding second lighting mode to re-acquire image data.

[0110] S208. Input the first image data and the second image data into the defect recognition model to obtain the defect detection result.

[0111] Referring to step S106, the pipeline inspection system will simultaneously input the image data collected under the two lighting modes into the defect recognition model, and obtain the final defect detection result through multi-angle and multi-dimensional feature analysis.

[0112] S209. Extract the defect type and geometric appearance parameters of the target defect from the defect detection results.

[0113] Among them, target defects refer to confirmed defects that require quantitative evaluation; defect type refers to the attribute classification of defects; geometric appearance parameters refer to quantitative indicators describing the morphological characteristics of defects, including size, depth, shape, etc.

[0114] After obtaining defect detection results, the pipeline inspection system needs to extract key parameters for subsequent evaluation. Specifically, the system first locates the precise position of the target defect from the detection results, and then extracts the defect's type label and detailed geometric feature parameters. These parameters include quantitative indicators such as the defect's spatial dimensions, depth distribution, and shape characteristics, providing basic data support for subsequent impact assessment.

[0115] In some embodiments, parameter extraction can be achieved in multiple ways: Optionally, the pipeline inspection system employs a multi-dimensional feature analysis method, including: constructing a defect feature vector, extracting morphological features, texture features, and statistical features, and establishing a complete defect description model. Optionally, the pipeline inspection system uses depth estimation techniques, including: reconstructing the three-dimensional structure of the defect using multi-view images, calculating geometric parameters, and evaluating the spatial distribution characteristics of the defect. It is understood that other methods can also be used to achieve parameter extraction, such as using structured light scanning to obtain high-precision topographic data, etc., which are not limited here.

[0116] In practical applications, the complex morphology of defects can affect the accuracy of parameter extraction. To address this, the pipeline inspection system implements robust parameter extraction by: establishing a defect morphology classification system and selecting appropriate feature extraction methods for different types; utilizing multi-source data fusion to improve parameter estimation accuracy; and employing interval estimation to represent results when there is uncertainty in parameter extraction.

[0117] S210. Calculate the defect influence coefficient of the target defect based on the defect type and geometric appearance parameters, combined with the preset pipeline parameters.

[0118] Among them, the preset pipeline parameters represent a set of engineering parameters describing the basic characteristics of the pipeline, including pipe diameter, wall thickness, material strength, etc.; the defect influence coefficient represents a scoring index that quantifies the degree of impact of a single defect on pipeline safety; and the geometric appearance parameters represent the quantitative characteristics of the defect, such as shape, size, and depth.

[0119] After acquiring defect characteristics, the pipeline inspection system needs to assess the severity of individual defects. Specifically, the system first reads preset pipeline basic parameters, including pipeline design parameters and material performance indicators. Then, combining the defect type characteristics and geometric parameters, it calculates the impact of the defect on the pipeline structural integrity using a mechanical model. Finally, it generates a standardized defect impact coefficient, which comprehensively considers the severity of the defect and the pipeline's load-bearing capacity.

[0120] In some embodiments, the influence coefficient can be calculated in several ways: Optionally, the pipeline network inspection system employs a multi-factor evaluation method, including: constructing a comprehensive evaluation model considering stress concentration, strength reduction, and life loss, calculating the contribution value of each influencing factor, and obtaining a weighted and fused influence coefficient. Optionally, the pipeline network inspection system uses numerical simulation technology, including: establishing a locally refined finite element model, analyzing the stress distribution changes caused by defects, and evaluating the structural safety margin. It is understood that other methods can also be used to achieve influence assessment, such as damage tolerance analysis based on fracture mechanics, etc., which are not limited here.

[0121] In practical applications, the coupling effect between defects and pipeline parameters can affect the accuracy of assessments. To address this, pipeline inspection systems implement assessment strategies that consider coupling effects: establishing a defect-pipeline interaction model to analyze the effects of geometric and material nonlinearities; adjusting assessment parameters based on service conditions; and initiating a refined assessment process when abnormal coupling phenomena are detected.

[0122] S211. Calculate the defect quantification score of the target pipeline based on the defect influence coefficient of all defects in the pre-inspection section that has been inspected.

[0123] Among them, the pre-inspection section represents the pipeline segment that has been inspected; the defect quantification score represents a comprehensive indicator reflecting the overall health status of the pipeline; and the influence coefficient set represents the dataset of influence coefficients of all defects within a specific pipeline segment.

[0124] After obtaining the impact coefficient of individual defects, the pipeline inspection system needs to assess the overall condition of the pipeline. Specifically, the system first collects the impact coefficient data of all defects found in the pre-inspection sections to establish a defect distribution database. Then, considering the spatial distribution characteristics and interaction effects of the defects, it calculates the overall quantitative score of the pipeline using a scoring model. This score reflects the cumulative degree of damage to the pipeline and can be used to guide the dynamic adjustment of inspection strategies.

[0125] In some embodiments, quantitative scoring can be achieved in multiple ways: Optionally, the pipeline inspection system employs the analytic hierarchy process (AHP), including: establishing a multi-level scoring index system, considering defect density, distribution patterns, and cumulative effects, and obtaining the final score through hierarchical weighting. Optionally, the pipeline inspection system uses statistical inference techniques, including: analyzing the statistical characteristics of defect distribution, assessing pipeline degradation trends, and predicting potential risk probabilities. It is understood that other methods can also be used to achieve quantitative scoring, such as probability assessment based on reliability theory, etc., which are not limited here.

[0126] In practical applications, the interaction between defects can lead to scoring bias. To address this, the pipeline inspection system implements a scoring mechanism that considers group effects: establishing a defect group analysis model to evaluate the combined effect of defect clusters; considering the distribution characteristics of defects in the axial and circumferential directions of the pipeline; and increasing the scoring weight of the section when a highly clustered defect group is detected.

[0127] In some embodiments, the pipeline inspection system dynamically adjusts the lighting strategy based on defect scores to improve inspection efficiency; that is, when the defect quantification score is lower than a preset score threshold, the pipeline inspection system calculates the third lighting mode for the inspection robot in the subsequent inspection section based on the defect types of all defects in the preceding inspection section; and controls the inspection robot to travel in the target pipeline and acquire images in the third lighting mode.

[0128] Among them, the defect quantification score represents a comprehensive indicator reflecting the overall health status of the pipeline; the preset scoring threshold represents the scoring threshold value that triggers the adjustment of the lighting strategy; the pre-inspection section represents the pipeline section that has been inspected; the post-inspection section represents the pipeline section to be inspected; and the third lighting mode represents a comprehensive enhanced lighting configuration scheme for efficient detection of areas with serious defects.

[0129] After completing the preliminary inspection and calculating the quantitative score, the pipeline inspection system needs to optimize subsequent inspection strategies based on the degree of pipeline damage. Specifically, the system first compares the calculated defect quantitative score with a preset threshold; a score below the threshold indicates significant pipeline damage. Then, the system analyzes all defect types and their distribution characteristics found in the preliminary inspection section, designing a comprehensive lighting scheme that accommodates the detection needs of multiple defect types, forming a third lighting mode. This mode features high illumination intensity and optimized light source configuration, simultaneously meeting the imaging requirements of different defect types and avoiding frequent switching of lighting modes. Finally, the inspection robot is controlled to continue performing the inspection tasks for subsequent sections using this mode, improving inspection efficiency.

[0130] In some embodiments, lighting strategy optimization can be achieved in several ways: Optionally, the pipeline inspection system employs a severity-based lighting configuration method, including: analyzing the severity distribution of defects in upstream sections, determining the lighting requirements for major defect types, and designing lighting parameter combinations that can simultaneously meet multiple inspection needs, achieving a balance between inspection efficiency and lighting effect. Optionally, the pipeline inspection system uses predictive optimization techniques, including: establishing a defect propagation model to predict the damage level of subsequent sections, designing adaptive lighting schemes, and adjusting the inspection strategy in advance to avoid frequent adjustments during the inspection process. It is understood that other methods can also be used to achieve lighting optimization, such as zoned lighting control strategies based on pipeline health conditions, etc., which are not limited here.

[0131] In practical applications, the spatial correlation of pipeline damage levels can affect the effectiveness of lighting strategies. To address this, pipeline inspection systems implement zoned adaptive lighting control: establishing a spatial distribution model of pipeline damage levels to analyze the continuity of defect severity; setting reasonable lighting mode switching thresholds based on damage levels; and promptly adjusting the lighting strategy to adapt to new inspection needs when significant changes in damage levels are detected. For example, when entering a severely damaged area, the system smoothly transitions to a higher-intensity lighting mode, ensuring inspection effectiveness while avoiding delays caused by frequent switching.

[0132] In some embodiments, the pipeline inspection system employs a multi-source data fusion positioning method to accurately locate defects; that is, the pipeline inspection system collects odometer data and image shooting angles during the movement of the inspection robot; and calculates the defect positioning data of the target defect based on the odometer data, image shooting angles, and the defect location corresponding to the target defect in the defect detection results.

[0133] Among them, the odometer data represents the travel distance and speed information of the inspection robot; the image shooting angle represents the spatial orientation parameter of the camera; the defect location data represents the coordinate information describing the precise spatial location of the defect in the pipeline; and the target defect represents the confirmed defect that needs to be precisely located.

[0134] During the robot's movement, the pipeline inspection system needs to collect motion parameters in real time for defect localization. Specifically, the system first uses an odometry sensor mounted on the robot to continuously record displacement, velocity, and acceleration data during movement. Simultaneously, it uses an attitude sensor to collect the pan-tilt angle of the camera, obtaining precise orientation information during image capture. Then, combining these motion parameters with the relative position information from the defect detection results, and through coordinate transformation and error compensation, the system calculates the absolute spatial coordinates of the target defect within the pipeline, generating standardized defect localization data.

[0135] In some embodiments, precise positioning can be achieved in multiple ways: Optionally, the pipeline inspection system employs a multi-sensor fusion method, including: integrating an inertial measurement unit to acquire attitude data, deploying an optical encoder to measure rotation angles, using a visual odometry to estimate relative displacement, and fusing multi-source data through a Kalman filter algorithm. Optionally, the pipeline inspection system uses visual positioning technology, including: extracting the correspondence between feature points in continuous images, reconstructing the camera motion trajectory, and combining structured light ranging data to achieve high-precision spatial positioning of defects. It is understood that other methods can also be used to achieve position measurement, such as ultrasonic or laser-based ranging and positioning, etc., which are not limited here.

[0136] In practical applications, robot slippage and vibration can lead to cumulative errors in odometry data. To address this, the pipeline inspection system implements a multi-source data correction mechanism: establishing a motion state evaluation model to detect abnormal movements in real time; using image feature matching for position correction; and triggering a global position relocation process when the cumulative error exceeds a threshold. For example, when robot slippage is detected, the system temporarily switches to a vision-based positioning method to ensure positioning accuracy.

[0137] Furthermore, in some embodiments, the pipeline inspection system constructs a complete positioning reference system to avoid inaccurate defect positioning due to single factors such as robot movement issues. Specifically, the pipeline inspection system acquires pipeline images captured by the inspection robot during its movement, extracts pipeline structural feature points including joints and welds from the pipeline images, and determines the point distribution characteristics of the pipeline structural feature points; acquires the radius parameters of the pipeline cross-section and the material change area data along the pipeline; combines the odometer data, point distribution characteristics, radius parameters, and material change area data to determine the positioning reference information for each image position in the pipeline image; and corrects the defect positioning data based on the positioning reference information.

[0138] Among them, pipeline structural feature points represent fixed landmark components on the pipeline, such as joints and welds; point distribution features represent the spatial arrangement of feature points; radius parameters represent the geometric dimensions of the pipeline cross-section; material change area data represent the material property changes along the pipeline; positioning reference information represents the benchmark dataset used for position calibration; and positioning correction represents the process of accurately calibrating the defect location based on the reference information.

[0139] After obtaining preliminary positioning results, the pipeline inspection system needs to establish a complete positioning reference system for precise calibration. Specifically, the system first analyzes the pipeline image sequence collected by the inspection robot, using image processing algorithms to identify permanent markers such as joints and welds on the pipeline, extracting the location information of these feature points and analyzing their distribution patterns. Simultaneously, it acquires basic pipeline parameter information, including cross-sectional radius and material distribution data. Then, it fuses multi-source information such as odometer data, feature point distribution, and pipeline parameters to construct a complete spatial reference coordinate system. Finally, based on this reference system, the previously calculated defect location is calibrated to obtain a more accurate positioning result.

[0140] In some embodiments, the positioning reference system can be constructed in various ways: Optionally, the pipeline network detection system adopts a multi-level reference point management strategy, including: establishing a hierarchical feature point database, performing reliability rating on feature points, constructing topological relationships between feature points, and realizing position calibration based on multiple references. Optionally, the pipeline network detection system uses dynamic reference update technology, including: real-time evaluation of the effectiveness of reference points, dynamic adjustment of reference point weights, and establishing a sliding reference window to ensure the real-time performance and accuracy of the positioning reference system. It is understood that other methods can also be used to construct the reference system, such as feature point detection and matching methods based on deep learning, etc., which are not limited here.

[0141] In practical applications, the complex structure of pipelines can lead to uneven distribution of reference feature points. To address this, the pipeline network detection system implements adaptive reference point management: it establishes a feature point quality assessment model to calculate the reliability score of each feature point; in areas with sparse feature points, it increases the weight contribution of image features; and when a section lacks reliable reference points, it constructs virtual reference points through interpolation. For example, in pipe bends or material transition zones, the system comprehensively uses multiple types of reference features and dynamically adjusts their weights based on their reliability to ensure the continuity and stability of positioning accuracy. Furthermore, the system records and analyzes historical data for positioning corrections to optimize reference point selection strategies and calibration algorithm parameters, continuously improving the performance of the positioning system.

[0142] S212. After generating a defect detection report based on the defect detection results, control the detection robot to switch back to the first lighting mode.

[0143] Referring to step S107, after generating the defect detection report, the pipeline inspection system will control the inspection robot to return to the energy-saving first lighting mode and continue to perform the inspection tasks of subsequent pipeline sections.

[0144] It's important to note that generating a defect detection report requires integrating multi-dimensional detection data: First, key information is extracted from the defect detection results, including the precise spatial location, type classification, geometric dimensions (such as length, width, and depth), and severity rating of the defect. Then, combined with basic pipeline parameters (such as pipe diameter, wall thickness, and material), the impact coefficient of each defect is calculated to assess its potential impact on the pipeline's structural integrity. The system also analyzes the spatial distribution characteristics of defects, identifying high-risk areas and areas with high defect density. Finally, this information is organized according to a standardized format to generate a comprehensive detection report containing text descriptions, data charts, and defect images.

[0145] Lighting mode adjustment is a delicate transition process. The pipeline inspection system first assesses the differences between the current lighting parameters and the first lighting mode, then designs a smooth parameter transition curve to avoid sudden changes in light intensity interfering with subsequent inspections. While reducing the main lighting intensity, the system monitors image quality indicators in real time to ensure that basic image acquisition quality is maintained during the transition. If the inspection robot is about to enter a new inspection section, the system will pre-calculate the optimal lighting parameter configuration to achieve seamless switching. For different types of lighting equipment (such as LED light sources and auxiliary lighting), the system will coordinate and control their power change timing to ensure a smooth transition in energy consumption.

[0146] The pipeline inspection system ensures the traceability and comparability of inspection results through a standardized report generation process. By employing a reasonable lighting mode switching strategy, it guarantees the continuity of inspection quality while optimizing energy utilization, providing a reliable foundation for subsequent inspection tasks and historical data analysis. For example, after inspecting a section of pipeline, the system-generated report not only includes the location and size of discovered cracks but also predicts crack development trends based on the pipeline's service life and load conditions. Simultaneously, the system pre-adjusts the lighting strategy based on the anticipated conditions of subsequent inspection sections, such as appropriately increasing the reference lighting intensity before entering pipeline bends, ensuring continuous and efficient inspection work.

[0147] In this embodiment, a detection method combining multi-mode collaborative lighting and intelligent analysis is employed, enabling adaptive adjustment of the lighting strategy based on defect characteristics to achieve accurate defect identification and location. Specifically, the system first performs a preliminary scan using an energy-efficient primary lighting mode. When a suspected defect is detected, it automatically switches to a targeted flash mode for detailed imaging. Deep learning models are used to analyze image features, and multi-source data fusion technology is combined for position calibration, effectively solving problems such as unstable image quality, low detection efficiency, and reliance on human experience under traditional fixed lighting methods. The system can also continuously optimize the detection strategy through historical data analysis, establishing a mapping relationship between defect characteristics and optimal lighting parameters, thereby achieving intelligent, standardized, and efficient detection processes. This not only improves detection accuracy and reduces energy consumption but also provides reliable technical support for intelligent operation and maintenance of large-scale pipeline networks.

[0148] The pipeline network detection system in the embodiments of this invention is described below from the perspective of hardware processing. Please refer to [link / reference needed]. Figure 3 This is a schematic diagram of the physical device structure of a pipeline network detection system in an embodiment of this application.

[0149] It should be noted that, Figure 3 The structure of the pipeline inspection system shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0150] like Figure 3 As shown, the pipeline monitoring system includes a CPU 301, which can perform various appropriate actions and processes according to a program stored in ROM 302 or a program loaded from storage section 308 into RAM 303, such as executing the methods described in the above embodiments. RAM 303 also stores various programs and data required for system operation. The CPU 301, ROM 302, and RAM 303 are interconnected via bus 304. I / O interface 305 is also connected to bus 304.

[0151] The following components are connected to I / O interface 305: input section 306 including audio input devices, push-button switches, etc.; output section 307 including liquid crystal display (LCD) and audio output devices, indicator lights, etc.; storage section 308 including hard disks, etc.; and communication section 309 including network interface cards such as LAN (Local Area Network) cards, modems, etc. Communication section 309 performs communication processing via a network such as the Internet. Drive 310 is also connected to I / O interface 305 as needed. Removable media 311, such as disks, optical disks, magneto-optical disks, semiconductor memories, etc., are installed on drive 310 as needed so that computer programs read from them can be installed into storage section 308 as needed.

[0152] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing computer programs for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 309, and / or installed from removable medium 311. When the computer program is executed by CPU 301, it performs the various functions defined in the present invention.

[0153] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. Each block in a flowchart or block diagram may represent a module, program segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those shown in the drawings.

[0154] Specifically, the pipeline network detection system of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, it implements the AI ​​identification method for pipeline defects provided in the above embodiment.

[0155] In another aspect, the present invention also provides a computer-readable storage medium, which may be included in the pipeline inspection system described in the above embodiments; or it may exist independently and not assembled into the pipeline inspection system. The storage medium carries one or more computer programs, which, when executed by a processor of the pipeline inspection system, cause the pipeline inspection system to implement the AI-based pipeline defect identification method provided in the above embodiments.

[0156] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0157] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as meaning "if...", "after...", "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if (the stated condition or event) is interpreted as meaning "if determining...", "in response to determining...", "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

Claims

1. An AI identification method for pipeline defects, characterized in that, The method is applied to a pipe network detection system, and the method comprises: controlling a detection robot to travel in a target pipeline in a first illumination mode to obtain first image data; inputting the first image data into a defect identification model to obtain a defect position and a defect type of a suspected defect region; when the defect type is a preset type, calling a preset flash parameter corresponding to the defect type, and obtaining a defect preview image of the defect position while triggering a flash in the preset flash parameter; inputting the defect preview image into a defect verification model corresponding to the defect type to obtain a defect existence probability; when the defect existence probability exceeds a preset probability threshold, determining a supplementary illumination parameter according to the defect type, and controlling the detection robot to obtain second image data in a corresponding second illumination mode based on the supplementary illumination parameter; inputting the first image data and the second image data into the defect identification model to obtain a defect detection result; after generating a defect detection report based on the defect detection result, controlling the detection robot to adjust back to the first illumination mode.

2. The method of claim 1, wherein, The step of inputting the first image data into a defect identification model to obtain a defect position and a defect type of a suspected defect region specifically comprises: performing standardization processing of brightness and contrast on the first image data to generate a standardized image; performing defect feature extraction on the standardized image based on a deep convolutional neural network to determine a suspected defect region and a defect position of the suspected defect region; performing identification classification on the suspected defect region based on a support vector machine to determine a defect type of the suspected defect region.

3. The method of claim 1, wherein, After the step of inputting the first image data and the second image data into the defect identification model to obtain a defect detection result, the method further comprises: extracting a defect type and a geometric appearance parameter of a target defect in the defect detection result; calculating a defect influence coefficient of the target defect according to the defect type and the geometric appearance parameter in combination with a preset pipeline parameter; calculating a defect quantitative score of the target pipeline according to defect influence coefficients of all defects in a pre-detection section of the target pipeline that has been detected.

4. The method of claim 3, wherein, After the step of calculating a defect quantitative score of the target pipeline according to defect influence coefficients of all defects in a pre-detection section of the target pipeline, the method further comprises: when the defect quantitative score is lower than a preset score threshold, calculating a third illumination mode of the detection robot in a post-detection section according to defect types of all defects in the pre-detection section; controlling the detection robot to travel in a target pipeline in the third illumination mode and collect images.

5. The method of claim 1, wherein, The step of, when the defect type is a preset type, calling a preset flash parameter corresponding to the defect type, and obtaining a defect preview image of the defect position while triggering a flash in the preset flash parameter specifically comprises: when the defect type is a preset type, determining reflectivity and a gray value of the suspected defect region, and calculating a minimum illumination energy threshold; determine a flash intensity and a duration according to the minimum illumination energy threshold and the defect type, generate an intensity timing and a trigger timing of flash illumination; trigger a flash with the intensity timing and the trigger timing, and acquire a defect preview image of the defect position.

6. The method of claim 1, wherein, After the step of inputting the first image data and the second image data into the defect identification model to obtain a defect detection result, the method further comprises: acquiring odometer data and an image shooting angle of the detection robot during travel; calculating defect positioning data of a target defect according to the odometer data, the image shooting angle, and a defect position of the target defect in the defect detection result.

7. The method of claim 6, wherein, After the step of calculating defect positioning data of a target defect according to the odometer data, the image shooting angle, and a defect position of the target defect in the defect detection result, the method further comprises: acquiring a pipeline image shot by the detection robot during travel, extracting pipeline structure feature points including a joint and a welding point in the pipeline image, and determining point distribution characteristics of the pipeline structure feature points; acquiring a radius parameter of a pipeline cross section and material change region data of the pipeline along a pipeline length; combining the odometer data, the point distribution characteristics, the radius parameter, and the material change region data to determine positioning reference information of each image position in the pipeline image; correcting the defect positioning data according to the positioning reference information.

8. A pipe network detection system characterised by, The pipeline network detection system comprises one or more processors and a memory; the memory is coupled to the one or more processors, the memory is configured to store computer program code, the computer program code comprises computer instructions, and the one or more processors invoke the computer instructions to enable the pipeline network detection system to perform the method according to any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that, The instructions enable the pipeline network detection system to perform the method according to any one of claims 1-7 when the instructions run on the pipeline network detection system.

10. A computer program product, characterised in that, The computer program product enables the pipeline network detection system to perform the method according to any one of claims 1-7 when the computer program product runs on the pipeline network detection system.

Images