A method and device for inspecting carbon dioxide pipeline leakage

Abstract

The application provides a carbon dioxide pipeline leakage inspection method and device, relates to the field of carbon dioxide transportation safety monitoring, and the method comprises the following steps: acquiring monitoring data collected by a plurality of source sensors arranged along a pipeline; performing space-time alignment and preprocessing on the monitoring data based on the respective space-time reference of the plurality of source sensors to form a multi-dimensional data set consistent in space-time; processing the multi-dimensional data set by using a feature extraction and fusion algorithm to determine a plurality of characteristic signals; jointly analyzing the plurality of characteristic signals based on a trained leakage criterion model to determine a leakage event and a leakage range; after determining that a leakage event occurs, fusing the plurality of characteristic signals and pipeline spatial information, and calculating the spatial position coordinates of the leakage point by using an iterative optimization algorithm. The application realizes accurate whole-process inspection of carbon dioxide pipeline leakage and improves data consistency and feature recognition.

Description

Technical Field

[0001] This application relates to the field of carbon dioxide transportation safety monitoring, and more specifically, to a method and apparatus for inspecting carbon dioxide pipeline leaks. Background Technology

[0002] In carbon dioxide transportation scenarios, pipelines are the core transport carriers, and their operational safety directly affects personnel safety, environmental protection, and economic benefits. Carbon dioxide leaks can easily lead to safety hazards such as low-temperature freezing and localized hypoxia, and may also cause resource waste and environmental impact. Therefore, leak inspection is crucial. Existing inspection methods mostly rely on single sensors or manual inspection. Manual inspection is inefficient and has limited coverage, making it difficult to meet the real-time monitoring needs of long-distance pipelines. Single sensors are susceptible to environmental interference, resulting in insufficient monitoring accuracy. Furthermore, multi-source sensor data lacks an effective spatiotemporal alignment mechanism, making it difficult to coordinate the analysis of characteristic signals. This leads to a high rate of misjudgment in leak identification and large deviations in leak point location, failing to meet the actual needs of accurate inspection. Summary of the Invention

[0003] The purpose of this application is to provide a method and apparatus for inspecting carbon dioxide pipeline leaks, which solves the above-mentioned problems existing in the prior art and can provide efficient and reliable inspection support for the safe operation of carbon dioxide pipelines.

[0004] Firstly, a method for inspecting carbon dioxide pipeline leaks is provided, which may include: Acquire monitoring data collected by multi-source sensors deployed along the pipeline; Based on the spatiotemporal references of the multi-source sensors, the monitoring data is spatiotemporally aligned and preprocessed to form a spatiotemporally consistent multidimensional dataset. The multidimensional dataset is processed using feature extraction and fusion algorithms to determine various feature signals characterizing leakage; Based on the trained leakage criterion model, the multiple feature signals are jointly analyzed to determine the leakage event and the leakage range. After confirming that a leak has occurred, the spatial coordinates of the leak point are determined by integrating the various characteristic signals and pipeline spatial information and through an iterative optimization algorithm.

[0005] In one possible implementation, the multi-source sensor includes at least a thermal imaging array for acquiring temperature field data, a flow monitoring unit for acquiring upstream and downstream flow data of the pipeline, and a concentration sensing array for acquiring target gas concentration data.

[0006] In one possible implementation, the multiple characteristic signals characterizing the leak include at least: low-temperature anomaly region characteristics identified based on the temperature field data, flow difference characteristics calculated based on the upstream and downstream flow data of the pipeline, and concentration gradient distribution characteristics generated based on the target gas concentration data.

[0007] In one possible implementation, the joint analysis of the multiple feature signals based on the trained leakage criterion model specifically includes: The low-temperature anomaly region characteristics, the flow difference characteristics, and the concentration gradient distribution characteristics are used as inputs; The spatiotemporal coupling relationship of the various characteristic signals is analyzed using the aforementioned leakage criterion model; A leakage event is determined to have occurred when at least two of the characteristic signals simultaneously exceed a preset threshold and meet the spatiotemporal correlation condition.

[0008] In one possible implementation, the leakage criterion model is a multimodal feature association model built on a deep learning network, used to learn the nonlinear mapping relationship and joint probability distribution between the various feature signals.

[0009] In one possible implementation, determining the leakage range includes: Based on the spatial ranges indicated by the low-temperature anomaly region characteristics and the concentration gradient distribution characteristics, respectively, an inversion calculation is performed using a gas diffusion model, and the confidence interval of the spatial intersection of the two is taken as the leakage range.

[0010] In one possible implementation, the spatial coordinates of the leak point are determined through an iterative optimization algorithm, specifically: The leakage range is used as the initial iteration interval; A joint optimization objective function is constructed based on the flow difference characteristics, the concentration gradient distribution characteristics, and the pipeline pressure distribution model. The objective function is iteratively solved in the spatial model of the pipeline using the weighted least squares method, and the spatial coordinates of the leak point are output.

[0011] Secondly, a carbon dioxide pipeline leak detection device is provided, which may include: The acquisition unit is used to acquire monitoring data collected by multi-source sensors deployed along the pipeline; The processing unit is used to perform spatiotemporal alignment and preprocessing on the monitoring data based on the spatiotemporal references of the multi-source sensors to form a spatiotemporally consistent multidimensional dataset. The determining unit is used to process the multidimensional dataset using feature extraction and fusion algorithms to determine various feature signals characterizing leakage; Furthermore, based on the trained leakage criterion model, the various feature signals are jointly analyzed to determine the leakage event and the leakage range; The fusion unit is used to fuse the various feature signals and pipeline spatial information after a leakage event is determined, and to determine the spatial coordinates of the leakage point through an iterative optimization algorithm.

[0012] Thirdly, an electronic device is provided, which includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When a processor executes a program stored in memory, it implements any of the steps described in the first aspect above.

[0013] Fourthly, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when executed by a processor, the computer program implements the steps of any of the methods described in the first aspect above.

[0014] This application provides a method and apparatus for inspecting carbon dioxide pipeline leaks. The method includes: acquiring monitoring data collected by multi-source sensors deployed along the pipeline; performing spatiotemporal alignment and preprocessing on the monitoring data based on the spatiotemporal references of the multi-source sensors to form a spatiotemporally consistent multidimensional dataset; processing the multidimensional dataset using feature extraction and fusion algorithms to determine multiple characteristic signals representing leaks; jointly analyzing the multiple characteristic signals based on a trained leak criterion model to determine the occurrence of a leak event and the leak range; and after determining the occurrence of a leak event, fusing multiple characteristic signals with pipeline spatial information and calculating the spatial coordinates of the leak point through an iterative optimization algorithm. This application achieves precise inspection of carbon dioxide pipeline leaks throughout the entire process, improves data consistency and feature identification, reduces the leak misjudgment rate, accurately locates the leak range and location, and provides high positioning accuracy and fast response, providing reliable support for pipeline safe operation and emergency repairs, and adapting to complex operating conditions. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A schematic flowchart illustrating a carbon dioxide pipeline leak inspection method provided in this application embodiment; Figure 2This is a schematic diagram of the structure of a carbon dioxide pipeline leak inspection device provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0018] In carbon dioxide transportation scenarios, pipelines are the core transport carriers, and their operational safety directly affects personnel safety, environmental protection, and economic benefits. Carbon dioxide leaks can easily lead to safety hazards such as low-temperature freezing and localized hypoxia, and may also cause resource waste and environmental impact. Therefore, leak inspection is crucial. Existing inspection methods mostly rely on single sensors or manual inspection. Manual inspection is inefficient and has limited coverage, making it difficult to meet the real-time monitoring needs of long-distance pipelines. Single sensors are susceptible to environmental interference, resulting in insufficient monitoring accuracy. Furthermore, multi-source sensor data lacks an effective spatiotemporal alignment mechanism, making it difficult to coordinate the analysis of characteristic signals. This leads to a high rate of misjudgment in leak identification and large deviations in leak point location, failing to meet the actual needs of accurate inspection.

[0019] Therefore, this application provides a method for inspecting carbon dioxide pipeline leaks, which solves the above-mentioned problems existing in the prior art and can provide efficient and reliable inspection support for the safe operation of carbon dioxide pipelines.

[0020] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0021] Figure 1 This is a flowchart illustrating a carbon dioxide pipeline leak inspection method provided in an embodiment of this application. Figure 1 As shown, the method may include: Step S110: Obtain monitoring data collected by multi-source sensors deployed along the pipeline.

[0022] The multi-source sensor includes at least a thermal imaging array for acquiring temperature field data, a flow monitoring unit for acquiring upstream and downstream flow data of the pipeline, and a concentration sensing array for acquiring target gas concentration data.

[0023] Specifically, the multi-source sensors employ a combined distributed and fixed-point deployment approach, adapting to complex laying scenarios of long-distance carbon dioxide pipelines (such as ground laying, buried laying, and crossing sections). Each sensor possesses harsh environmental resistance capabilities (low temperature resistance, moisture resistance, and corrosion resistance), aligning with the characteristics of carbon dioxide leaks that easily trigger localized low temperatures and changes in ambient humidity. The specific deployment and data acquisition logic of each sensor is as follows: The thermal imaging array is deployed along the pipeline axis at intervals of 5-10 meters. Each deployment unit contains two infrared thermal imagers, facing the outer wall of the pipeline and a 1-3 meter range around the pipeline, respectively. It uses a resolution of 160×120 pixels and a sampling frequency of 10Hz to collect temperature field data of the pipeline and its surrounding environment in real time. It can capture local low temperature anomalies caused by phase change heat absorption after carbon dioxide leakage, providing temperature dimension basis for preliminary leakage identification.

[0024] The flow monitoring unit deploys one high-precision electromagnetic flow sensor at each of the key upstream and downstream nodes (with a spacing of no more than 50 kilometers) and branch pipe interfaces. The measurement range is adapted to the actual transport flow of the pipeline (0-1000 m³ / h), and the accuracy level is no less than 0.2. It synchronously collects real-time flow data and flow fluctuation values ​​from upstream and downstream, and captures flow anomalies caused by medium leakage in the pipeline by changes in flow difference.

[0025] The concentration sensor array is deployed at intervals of 8-15 meters around the perimeter of the pipeline. The sensor probe is 0.5-1 meter away from the outer wall of the pipeline. It adopts an infrared absorption gas concentration sensor with a range of 0-5000ppm and a response time of ≤3 seconds. It has the ability to specifically identify carbon dioxide and can effectively eliminate interference factors such as air and water vapor. It can collect target gas concentration data around the pipeline in real time and capture the characteristics of concentration gradient changes.

[0026] The monitoring data collected by each sensor is accompanied by a timestamp and deployment location identification information. It is aggregated to the data acquisition terminal through wired (fiber optic cable) and wireless (LoRa gateway) dual-mode transmission to ensure the stability and real-time performance of data transmission. This lays the foundation for subsequent spatiotemporal alignment preprocessing, helps to achieve rapid detection and accurate location of carbon dioxide pipeline leaks, and provides support for the safe operation of pipelines.

[0027] Step S120: Based on the spatiotemporal references of the multi-source sensors, perform spatiotemporal alignment and preprocessing on the monitoring data to form a spatiotemporally consistent multidimensional dataset.

[0028] Specifically, firstly, a unified time reference (UTC time synchronized to the millisecond level) is established along the pipeline using GPS positioning references. The clock module of the flow monitoring unit serves as the core reference source (its sampling frequency stability is ≤0.01%), and time reference calibration is performed on the thermal imaging array and the concentration sensing array respectively. To address the time asynchrony issue between the thermal imaging array (10Hz acquisition) and the concentration sensing array (response time fluctuates within 3 seconds), interpolation algorithms are used to supplement missing time node data. Based on the GPS coordinates of each sensor's deployment location, the asynchronously acquired data is mapped to a unified time axis, with time alignment errors controlled within ±10ms.

[0029] By combining the pipeline's three-dimensional spatial model with the pipeline's axial mileage markers and radial cross-sectional coordinates as spatial references, the regional temperature field data collected by the thermal imaging array and the point-based concentration data from the concentration sensor array are mapped to the pipeline's three-dimensional coordinate system. To address the spatial deviation between buried and surface-laid sections, a terrain adaptation factor is introduced. Through a coordinate offset correction algorithm, sensor data from different deployment scenarios are calibrated to the same spatial grid (grid accuracy 0.1m × 0.1m), achieving consistent spatial dimension correlation.

[0030] Next, data cleaning was performed. The 3σ criterion was used to remove pulse interference data from the flow monitoring unit, pixel noise data from the thermal imaging array, and environmental interference peak data from the concentration sensing array. Simultaneously, a sliding window filling method (with the window size adapted to the acquisition frequency of each sensor) was used to supplement a small amount of missing data, avoiding feature loss caused by single filtering. Next, data standardization was performed. For the three heterogeneous data types—temperature, flow rate, and concentration—the max-min normalization algorithm was used to map the data to the [0,1] interval, eliminating the impact of dimensional differences on subsequent fusion analysis.

[0031] Finally, through adaptive feature enhancement processing, core leak-related features (such as low temperature anomaly gradient, concentration mutation trend, and flow fluctuation pattern) are retained, while irrelevant interference features such as ambient temperature changes and pipeline vibration are suppressed. A spatiotemporally consistent dataset containing timestamps, spatial coordinates, and standardized multidimensional parameters is generated, providing high-quality data support for subsequent feature extraction and fusion, and leak criterion analysis. Compared with traditional preprocessing methods, this effectively improves data utilization and leak feature identification.

[0032] The monitoring data collected by each sensor is accompanied by a timestamp and deployment location identification information. It is aggregated to the data acquisition terminal through wired (fiber optic cable) and wireless (LoRa gateway) dual-mode transmission to ensure the stability and real-time performance of data transmission. This lays the foundation for subsequent spatiotemporal alignment preprocessing, helps to achieve rapid detection and accurate location of carbon dioxide pipeline leaks, and provides support for the safe operation of pipelines.

[0033] Step S130: Use feature extraction and fusion algorithms to process the multidimensional dataset and determine various feature signals that characterize the leakage.

[0034] Among them, the various characteristic signals characterizing the leak include at least: the low-temperature anomaly region characteristics identified based on temperature field data, the flow difference characteristics calculated based on upstream and downstream flow data of the pipeline, and the concentration gradient distribution characteristics generated based on target gas concentration data.

[0035] Specifically, the feature extraction process employs differentiated algorithms adapted to the characteristics of data across different dimensions, specifically capturing various leaked core features to avoid feature distortion or loss of key information caused by general-purpose algorithms, as detailed below: For feature extraction of low-temperature anomaly regions from temperature field data: An improved U-Net semantic segmentation algorithm is employed, combined with optimization of the network loss function based on the low-temperature distribution patterns of carbon dioxide leakage. A spatial attention mechanism is introduced to focus on key areas within 1-3 meters around the pipeline, segmenting and identifying the standardized temperature field grid data. By setting a dynamic low-temperature threshold (based on a 5-8℃ reduction from the average normal operating temperature of the pipeline, and dynamically corrected by ambient temperature), the boundaries, area, and temperature gradient change trends of low-temperature anomaly regions are accurately delineated. Simultaneously, pseudo-low-temperature features caused by environmental factors such as cloud cover and diurnal temperature variations are eliminated, outputting low-temperature anomaly region features with spatial location attributes and temperature gradient information.

[0036] For the extraction of flow difference features from upstream and downstream pipeline flow data: an algorithm combining sliding window statistics and trend fitting is adopted. A 10-second sliding window adapted to the sampling frequency of the flow monitoring unit is set to calculate the upstream and downstream flow difference and the rate of change of flow difference in real time. The flow difference change curve is fitted by a first-order exponential smoothing algorithm to eliminate the interference of instantaneous flow difference caused by pipeline medium pulsation and pressure fluctuation. When the slope of the fitted curve exceeds the preset threshold (set to ±0.02 m³ / (h·s) based on the normal flow fluctuation range of the pipeline), it is judged as an abnormal flow difference feature, and key parameters such as the peak value, duration and trend of flow difference are output simultaneously.

[0037] For extracting concentration gradient distribution features from target gas concentration data: The K-nearest neighbor interpolation algorithm is used to complete the point-based concentration data into a continuous concentration field. A concentration gradient calculation model is constructed based on the pipeline's three-dimensional coordinate system to solve for the gradient vectors of the concentration field in the axial and radial directions. Adaptive threshold filtering (based on a threshold three times the ambient background carbon dioxide concentration) identifies regions of abrupt changes in concentration gradient. Simultaneously, combined with a gas diffusion dynamics model, concentration distribution distortions caused by environmental factors such as wind and humidity are eliminated. Core features such as the location of high-concentration areas, the magnitude of gradient changes, and the diffusion direction are output to form the concentration gradient distribution characteristics.

[0038] The feature fusion stage employs a multimodal attention fusion algorithm to construct a feature correlation matrix, enabling intelligent fusion of three types of feature signals. First, the three types of features undergo dimensionality normalization, mapping parameters such as the area of ​​the low-temperature anomaly region, temperature gradient, peak value and rate of change of the flow difference, and amplitude and diffusion direction of the concentration gradient to the same feature space. Then, an attention mechanism dynamically allocates feature weights, assigning higher weights (0.35-0.4) to features highly sensitive to leakage (such as abrupt changes in concentration gradient and low-temperature region gradient) and moderate weights (0.2-0.3) to features highly sensitive to interference (such as the rate of change of flow difference). Through weighted fusion and feature dimensionality reduction, redundant feature information is eliminated, ultimately outputting a set of multiple feature signals that characterize the leakage, possess strong correlation and anti-interference capabilities, providing accurate input for subsequent joint analysis of the leakage criterion model.

[0039] Step S140: Based on the trained leakage criterion model, perform joint analysis on multiple feature signals to determine the leakage event and leakage range.

[0040] Among them, a joint analysis of multiple feature signals is performed based on a trained leakage criterion model, including: The leakage criterion model is a multimodal feature association model built on a deep learning network, used to learn the nonlinear mapping relationship and joint probability distribution between various feature signals.

[0041] Furthermore, the leakage criterion model is a multimodal feature association model based on an improved Transformer architecture. Unlike traditional deep learning models that suffer from isolated single-modal learning and insufficient nonlinear association capture, it introduces a cross-modal interactive attention module and a joint probability distribution learning unit. It is specifically adapted to the heterogeneous attributes of low-temperature anomaly region features, flow difference features, and concentration gradient distribution features. It can efficiently learn the nonlinear mapping relationship and spatiotemporal joint probability distribution between the three types of features, and at the same time has dynamic anti-interference capabilities, effectively filtering out the influence of irrelevant factors such as environmental noise and pipeline vibration.

[0042] The model training employs a hybrid training strategy using both simulated and experimental datasets to ensure generalization ability: the simulated dataset is generated through a pipeline leakage simulation platform, covering feature combinations of different leakage rates (0.1-10 m³ / h), environmental scenarios (high temperature, low temperature, windy and rainy weather), and laying types (ground and buried); the experimental dataset comes from leakage test data of pilot pipelines in the field, supplementing feature deviation information under real working conditions. During training, data augmentation techniques such as random pruning, noise addition, and spatiotemporal misalignment perturbation are used to expand the sample size. The cross-entropy loss function and joint probability loss function are used for collaborative optimization. The final model achieves a leakage detection accuracy of ≥98.5%, a false positive rate of ≤1%, and a spatiotemporal correlation recognition latency of ≤2 seconds, meeting the requirements of real-time inspection.

[0043] Specifically, in step 1, the features of the low-temperature anomaly region, the flow difference features, and the concentration gradient distribution features are used as inputs. Further, the features of the low-temperature anomaly region (boundary coordinates, temperature gradient, area), the flow difference features (peak value, rate of change, duration), and the concentration gradient distribution features (gradient vector, location of high-value areas, diffusion direction) are unified in dimensionality (converted into a 256-dimensional adaptation vector), and the spatiotemporal identifiers (millisecond-level timestamps, three-dimensional coordinates of the pipeline) corresponding to each feature are retained simultaneously to ensure the integrity of the input data.

[0044] Step 2: Analyze the spatiotemporal coupling relationship of various feature signals using a leakage criterion model; further, the model dynamically mines the spatiotemporal correlation of the three types of features through a cross-modal attention module, quantifies the coupling coefficient between features, such as the spatial overlap between low temperature anomaly areas and high concentration areas, and the temporal synchronization between abrupt changes in flow rate difference and the increase in concentration gradient, to avoid false triggering of judgment by a single feature.

[0045] Step 3: When at least two of the feature signals simultaneously exceed the preset threshold and meet the spatiotemporal correlation condition, a leakage event is determined to have occurred. Furthermore, the model has a built-in adaptive threshold system (statistically optimized based on the training dataset and dynamically adjusted according to pipeline operating parameters). When at least two types of feature signals simultaneously exceed their corresponding thresholds, and the spatiotemporal coupling coefficient meets the preset condition (≥0.75, i.e., significant spatiotemporal correlation of the features), a leakage event is determined to have occurred. If only a single feature exceeds the threshold or the spatiotemporal correlation is insufficient, it is determined to be an interference signal, and no leakage alarm is triggered, minimizing the risk of misjudgment.

[0046] In some embodiments, determining the leakage range includes: Based on the spatial range indicated by the characteristics of the low temperature anomaly region and the concentration gradient distribution, respectively, the gas diffusion model is used for inversion calculation, and the confidence interval of the spatial intersection of the two is taken as the leakage range.

[0047] The above process can be understood as follows: After determining that a leakage event has occurred, the leakage range is defined by using dual-feature basis, diffusion model inversion and confidence interval optimization, breaking through the accuracy bottleneck of traditional single feature range direct superposition: First, the spatial range corresponding to the output low temperature anomaly region features (based on the boundary coordinates of the improved U-Net segmentation, with the error controlled within ±0.2 meters) and the spatial range corresponding to the concentration gradient distribution features (based on the boundary delineation of the concentration gradient abrupt change, covering the concentration exceeding the threshold region) are extracted as the initial range basis.

[0048] Subsequently, an improved Gaussian diffusion model was introduced for inversion calculations. This model, which takes into account the endothermic phase transition and greater density of carbon dioxide leakage than air, incorporates a temperature correction factor and a terrain adaptation coefficient, replacing the traditional general diffusion model. It can accurately simulate the diffusion trajectory and concentration distribution of carbon dioxide under different environmental conditions. The initial range data of the low-temperature anomaly area and the concentration gradient distribution were input into the model. Combined with real-time environmental parameters (wind speed, wind direction, temperature, humidity) and pipeline laying terrain data, the possible spatial areas of the leakage source corresponding to the two types of characteristics were inverted and calculated, correcting the deviation and distortion of the initial range.

[0049] Finally, the range results were optimized through confidence interval analysis: the spatial intersection of the two types of inverted regions was calculated, and a confidence interval (95% confidence level) was set based on the feature confidence level of the model output (low temperature feature confidence level ≥ 0.9, concentration feature confidence level ≥ 0.92). Edge points with low confidence in the intersection region were removed, and the final output leakage range boundary error was ≤ 0.5m. This can accurately locate the core leakage area while avoiding the reduction of subsequent positioning efficiency due to the excessive range, thus balancing accuracy and practicality.

[0050] This step breaks through the limitations of traditional leak detection methods, which rely on single feature thresholds and ignore spatiotemporal correlations. By constructing a multimodal deep learning criterion model, it achieves cross-dimensional coupled analysis of the fused feature signals. This not only improves the accuracy of leak event detection and reduces the false positive rate, but also accurately defines the leak range, providing a reliable boundary basis for subsequent leak point location. It combines theoretical innovation with engineering feasibility.

[0051] Step S150: After confirming that a leakage event has occurred, the spatial coordinates of the leakage point are determined by integrating multiple feature signals and pipeline spatial information and using an iterative optimization algorithm.

[0052] Specifically, the leakage range is used as the initial iteration interval; furthermore, the initial iteration interval and spatial constraints are established. The output leakage range (the intersection area with 95% confidence) is used as the initial iteration interval. Constraint boundaries are set based on the three-dimensional spatial model of the pipeline, limiting the iterative solution to only the pipeline body and its surrounding 0.5m range, excluding invalid solutions in non-pipeline areas, significantly improving iteration efficiency. Simultaneously, pipeline spatial topology information (such as the coordinates of bends and joints) is imported as a basis for verifying the rationality of subsequent positioning results, avoiding deviations of the positioning point from the pipeline body due to algorithm iteration.

[0053] A joint optimization objective function is constructed using flow difference characteristics, concentration gradient distribution characteristics, and pipeline pressure distribution model. Furthermore, a multi-factor coupled joint optimization objective function is constructed to overcome the limitations of traditional single objective functions. The objective function uses the leak point coordinates as variables, integrating three core elements: flow difference characteristics, concentration gradient distribution characteristics, and pipeline pressure distribution model. Dynamic weight allocation achieves synergistic constraints among these factors. The specific expression is as follows:

[0054] Where (x, y, z) are the three-dimensional coordinates of the leak point. , , Dynamic weights (adjusted dynamically based on feature credibility) The sum is 1). Based on the peak value and rate of change of the output flow difference, and combined with the pipe cross-sectional area and medium velocity parameters, a correlation model between leakage and flow difference is established for the corresponding flow difference term. Quantify the impact of leakage at different coordinate points on the difference in upstream and downstream flow rates; The corresponding concentration gradient components, based on the concentration gradient vector and diffusion direction, combined with an improved Gaussian diffusion model, construct the theoretical distribution of the concentration gradient. The positioning accuracy is constrained by the deviation from the measured gradient; For the corresponding pressure component, a pipeline pressure distribution model is introduced. Based on real-time pressure data from upstream and downstream of the pipeline, the pressure loss distribution after leakage at different coordinate points is calculated. This fills the positioning blind spot in buried pipeline scenarios where single flow and concentration characteristics are insufficient.

[0055] The objective function is iteratively solved in the spatial model of the pipeline using weighted least squares, outputting the spatial coordinates of the leak point. Furthermore, an improved weighted least squares method is employed for iterative solution, enhancing positioning accuracy and convergence speed. Unlike the fixed weights of traditional weighted least squares, this step introduces an iterative adaptive weight adjustment mechanism, dynamically adjusting the weights after each iteration based on the deviation contribution of each feature component. , , Items with smaller deviations and higher reliability are assigned higher weights to strengthen effective constraints. During the iteration process, the center point of the initial leakage range is used as the initial value for iteration. A convergence threshold (coordinate deviation ≤ 0.1 meters) and a maximum number of iterations (50 times) are set to gradually approximate the optimal solution in the three-dimensional spatial model of the pipeline. 1. Substitute the initial coordinate values ​​to calculate the objective function value F0, and assign initial weights based on the deviations of each component; 2. Update the coordinates of the leak point using the gradient descent method, and calculate the objective function value F corresponding to the new coordinates. n 3. Comparison and If the difference is less than the convergence threshold, the iteration stops and the current coordinates are output; if the threshold is not met, the weights are adjusted and the next iteration begins, until the convergence condition or the maximum number of iterations is reached.

[0056] After the iterative solution is completed, the results are verified by combining the pipeline spatial topology information. If the location point falls within a reasonable range of the pipeline body and its surroundings, the final three-dimensional spatial coordinates of the leak point are output (accuracy ≤ ±0.3 meters). If it deviates, a second iteration is triggered to ensure the reliability of the location results. This scheme effectively overcomes the influence of environmental interference, pipeline laying terrain, and other factors through multi-feature collaborative constraints and iterative optimization. Compared with traditional positioning methods, the accuracy is improved by more than 40%, and it can quickly provide accurate location guidance for emergency repair operations.

[0057] This application provides a method for inspecting carbon dioxide pipeline leaks. The method includes: acquiring monitoring data collected by multi-source sensors deployed along the pipeline; performing spatiotemporal alignment and preprocessing on the monitoring data based on the spatiotemporal references of each multi-source sensor to form a spatiotemporally consistent multidimensional dataset; using feature extraction and fusion algorithms to process the multidimensional dataset and determine various characteristic signals representing the leak; jointly analyzing the various characteristic signals based on a trained leak criterion model to determine the occurrence of a leak event and its range; and after determining the occurrence of a leak event, fusing the various characteristic signals with pipeline spatial information and calculating the spatial coordinates of the leak point through an iterative optimization algorithm. This application achieves precise inspection of carbon dioxide pipeline leaks throughout the entire process, improves data consistency and feature identification, reduces the false positive rate, accurately locates the leak range and position, and offers high positioning accuracy and fast response, providing reliable support for pipeline safe operation and emergency repairs, and adapting to complex operating conditions.

[0058] Corresponding to the above method, this application also provides a carbon dioxide pipeline leak inspection device, such as... Figure 2 As shown, the device includes: Acquisition unit 210 is used to acquire monitoring data collected by multi-source sensors deployed along the pipeline; Processing unit 220 is used to perform spatiotemporal alignment and preprocessing on the monitoring data based on the spatiotemporal references of the multi-source sensors to form a spatiotemporally consistent multidimensional dataset. The determining unit 230 is used to process the multidimensional dataset using feature extraction and fusion algorithms to determine multiple feature signals characterizing leakage; Furthermore, based on the trained leakage criterion model, the various feature signals are jointly analyzed to determine the leakage event and the leakage range; The fusion unit 240 is used to fuse the various feature signals and pipeline spatial information after determining that a leakage event has occurred, and to calculate the spatial coordinates of the leakage point through an iterative optimization algorithm.

[0059] The functions of each functional unit of the carbon dioxide pipeline leak inspection device provided in the above embodiments of this application can be realized through the above methods and steps. Therefore, the specific working process and beneficial effects of each unit in the carbon dioxide pipeline leak inspection device provided in the embodiments of this application will not be repeated here.

[0060] This application also provides an electronic device, such as... Figure 3 As shown, it includes a processor 310, a communication interface 320, a memory 330, and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other through the communication bus 340.

[0061] Memory 330 is used to store computer programs; When the processor 310 executes the program stored in the memory 330, it performs the following steps: Acquire monitoring data collected by multi-source sensors deployed along the pipeline; Based on the spatiotemporal references of the multi-source sensors, the monitoring data is spatiotemporally aligned and preprocessed to form a spatiotemporally consistent multidimensional dataset. The multidimensional dataset is processed using feature extraction and fusion algorithms to determine various feature signals characterizing leakage; Based on the trained leakage criterion model, the multiple feature signals are jointly analyzed to determine the leakage event and the leakage range. After confirming that a leak has occurred, the spatial coordinates of the leak point are determined by integrating the various characteristic signals and pipeline spatial information and through an iterative optimization algorithm.

[0062] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0063] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0064] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0065] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0066] The implementation methods and beneficial effects of the various components of the electronic device in the above embodiments for solving the problem can be found in [reference needed]. Figure 1 The steps in the illustrated embodiments are used to implement the electronic device. Therefore, the specific working process and beneficial effects of the electronic device provided in this application will not be repeated here.

[0067] In another embodiment provided in this application, a computer-readable storage medium is also provided, which stores instructions that, when executed on a computer, cause the computer to perform a carbon dioxide pipeline leak inspection method as described in any of the above embodiments.

[0068] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute a carbon dioxide pipeline leak inspection method as described in any of the above embodiments.

[0069] Those skilled in the art will understand that the embodiments in this application can be provided as methods, systems, or computer program products. Therefore, the embodiments in this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments in this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0070] This application describes embodiments of methods, apparatus (systems), and computer program products according to embodiments of this application with reference to flowchart illustrations and / or block diagrams. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0071] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0072] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0073] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected," "coupled," or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0074] Although preferred embodiments have been described in this application, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the embodiments in this application are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments in this application.

[0075] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of the embodiments of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the embodiments of this application and their equivalents, then these modifications and variations are also intended to be included in the embodiments of this application.

Claims

1. A method for inspecting carbon dioxide pipeline leaks, characterized in that, The method includes: Acquire monitoring data collected by multi-source sensors deployed along the pipeline; Based on the spatiotemporal references of the multi-source sensors, the monitoring data is spatiotemporally aligned and preprocessed to form a spatiotemporally consistent multidimensional dataset. The multidimensional dataset is processed using feature extraction and fusion algorithms to determine various feature signals characterizing leakage; Based on the trained leakage criterion model, the multiple feature signals are jointly analyzed to determine the leakage event and the leakage range. After confirming that a leak has occurred, the spatial coordinates of the leak point are determined by integrating the various characteristic signals and pipeline spatial information and through an iterative optimization algorithm.

2. The method as described in claim 1, characterized in that, The multi-source sensor includes at least a thermal imaging array for acquiring temperature field data, a flow monitoring unit for acquiring upstream and downstream flow data of the pipeline, and a concentration sensing array for acquiring target gas concentration data.

3. The method as described in claim 2, characterized in that, The multiple characteristic signals characterizing the leak include at least: low-temperature anomaly region characteristics identified based on the temperature field data, flow difference characteristics calculated based on the upstream and downstream flow data of the pipeline, and concentration gradient distribution characteristics generated based on the target gas concentration data.

4. The method as described in claim 3, characterized in that, The joint analysis of the multiple feature signals based on the trained leakage criterion model specifically includes: The low-temperature anomaly region characteristics, the flow difference characteristics, and the concentration gradient distribution characteristics are used as inputs; The spatiotemporal coupling relationship of the various characteristic signals is analyzed using the aforementioned leakage criterion model; A leakage event is determined to have occurred when at least two of the characteristic signals simultaneously exceed a preset threshold and meet the spatiotemporal correlation condition.

5. The method as described in claim 1, characterized in that, The leakage criterion model is a multimodal feature association model built on a deep learning network, used to learn the nonlinear mapping relationship and joint probability distribution between the various feature signals.

6. The method as described in claim 4, characterized in that, Determining the extent of the leak includes: Based on the spatial ranges indicated by the low-temperature anomaly region characteristics and the concentration gradient distribution characteristics, respectively, an inversion calculation is performed using a gas diffusion model, and the confidence interval of the spatial intersection of the two is taken as the leakage range.

7. The method as described in claim 4, characterized in that, The spatial coordinates of the leak point were determined through iterative optimization algorithms, specifically: The leakage range is used as the initial iteration interval; A joint optimization objective function is constructed based on the flow difference characteristics, the concentration gradient distribution characteristics, and the pipeline pressure distribution model. The objective function is iteratively solved in the spatial model of the pipeline using the weighted least squares method, and the spatial coordinates of the leak point are output.

8. A carbon dioxide pipeline leak inspection device, characterized in that, The device includes: The acquisition unit is used to acquire monitoring data collected by multi-source sensors deployed along the pipeline; The processing unit is used to perform spatiotemporal alignment and preprocessing on the monitoring data based on the spatiotemporal references of the multi-source sensors to form a spatiotemporally consistent multidimensional dataset. The determining unit is used to process the multidimensional dataset using feature extraction and fusion algorithms to determine various feature signals characterizing leakage; Furthermore, based on the trained leakage criterion model, the various feature signals are jointly analyzed to determine the leakage event and the leakage range; The fusion unit is used to fuse the various feature signals and pipeline spatial information after a leakage event is determined, and to determine the spatial coordinates of the leakage point through an iterative optimization algorithm.

9. An electronic device, characterized in that, The electronic device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method described in any one of claims 1-7.

Images