Hydrogen-doped natural gas transportation safety risk assessment method and system
By combining distributed fiber optic sensor arrays and historical defect databases, the material performance changes of hydrogen-blended natural gas pipelines are monitored in real time, generating a mixed risk map. This solves the problem of delayed risk assessment caused by hydrogen infiltration in existing technologies, enabling accurate risk assessment and optimized maintenance strategies.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- XIAN UNIV OF SCI & TECH
- Filing Date
- 2026-05-13
- Publication Date
- 2026-06-12
AI Technical Summary
Existing technologies are insufficient for real-time monitoring of material performance degradation caused by hydrogen infiltration in hydrogen-blended natural gas pipelines, and cannot accurately assess the risks under the coupling effect of hydrogen damage and historical defects, resulting in delayed risk assessment and blind spots.
By collecting pipeline strain vibration waveforms using a distributed fiber optic sensor array, low-frequency noise characteristics of hydrogen-induced material property changes are identified. These waveforms are then overlaid and reconstructed with corrosion crack patterns from a historical defect database to generate a hybrid risk profile. Combined with combustible migration rate and cumulative fatigue reduction factor, the risk entropy value is quantified, and differentiated maintenance strategies are formulated.
It enables real-time perception and precise risk location of material degradation processes such as hydrogen embrittlement, improving the accuracy and predictability of risk assessment. It can also target key locations where hydrogen damage and historical defects synergistically affect each other, and optimize maintenance strategies.
Smart Images
Figure CN122198664A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of safety monitoring technology for hydrogen-blended natural gas pipelines, specifically a method and system for assessing the safety risks of hydrogen-blended natural gas transportation. Background Technology
[0002] Current monitoring technologies for natural gas pipeline safety mainly focus on monitoring macroscopic operating parameters such as pressure, flow rate, and temperature, as well as leak detection based on acoustic waves or negative pressure waves. However, existing conventional technologies lack effective online monitoring and early identification methods for the unique risks associated with hydrogen incorporation. Hydrogen infiltration under high pressure into the pipe wall's metallic materials can potentially lead to hydrogen embrittlement, fatigue crack propagation, and other material performance degradation. Conventional pipeline internal inspections or periodic external inspections struggle to capture the microscopic changes in material properties caused by hydrogen in real time, resulting in delays and blind spots in the assessment of the pipeline's safety status under hydrogen-incorporated environments.
[0003] In pipeline risk assessment, existing methods largely rely on statistical analysis of historical leakage or corrosion data, or independent evaluation of detection signals from a single source. Faced with the new risks introduced by hydrogen infiltration, these methods fail to deeply integrate and analyze the inherent historical defect characteristics of pipeline materials with real-time material damage signals induced by hydrogen. This makes it difficult for risk assessment results to accurately reflect the true risk evolution under the coupled effect of hydrogen damage and historical defects, and cannot provide sufficiently refined decision-making basis for the accurate identification and differentiated maintenance of high-risk sections. A method is needed that can sense hydrogen-induced changes in material properties in real time and integrate historical defect information for accurate risk location and assessment. Summary of the Invention
[0004] This invention aims to solve at least one of the technical problems existing in the prior art; Therefore, this invention proposes a method for assessing the safety risks of hydrogen-blended natural gas transportation, comprising: Initialize the safety situation assessment engine at the pipeline control center and inject a pre-configured mixture of gas components into the safety situation assessment engine; By deploying a distributed fiber optic sensor array along the target pipe section, the strain vibration waveform of the pipe body is acquired, and the low-frequency noise characteristics caused by hydrogen-induced changes in material properties are identified in the waveform. Corrosion crack patterns of materials from the same batch as the target pipe section are extracted from the pipeline historical defect database. The corrosion crack patterns are then overlaid and reconstructed with the currently collected low-frequency noise features to generate a hybrid risk pattern. Based on the characteristic intensity of different spatial locations in the hybrid risk map, multiple risk zones are divided; The proportions of the mixed gas components are input into the safety situation assessment engine, which drives the engine to calculate the flammable migration rate of the mixed gas in each risk zone. Extract the service condition records of the target pipe section and calculate the cumulative fatigue reduction factor of the pipe wall material under the mixed gas composition; By combining the flammable migration rate and the cumulative fatigue reduction factor, the current risk entropy value of each risk segment is quantified; Based on the current risk entropy value, the pipeline control center issues differentiated maintenance strategies, including pressure regulation instructions and inspection routes, to the maintenance units belonging to the designated risk sections.
[0005] Preferably, identifying low-frequency noise features in the waveform caused by hydrogen-induced changes in material properties includes: The original strain vibration signal is captured by a distributed fiber optic sensor array, generating a time-domain waveform stream; Adaptive wavelet packet decomposition is performed on the time-domain waveform stream to extract sub-band components within a specified frequency band range; Obtain the reference background noise spectrum of the target pipeline segment when transporting pure natural gas; The residual noise component is obtained by filtering out the components that match the reference background noise spectrum from the sub-band components. Perform a Hilbert transform on the residual noise component to obtain its time-frequency energy distribution; In the time-frequency energy distribution, a continuous region with energy higher than a set threshold and frequency lower than a preset threshold is located, and the center frequency and bandwidth of the continuous region are used as the low-frequency noise characteristics.
[0006] Preferably, the reconstruction analysis of the corrosion crack pattern overlay with the currently acquired low-frequency noise features includes: Identify the main propagation direction and branching density information of the cracks from the corrosion crack pattern; The low-frequency noise features are mapped onto the pipe spatial coordinates to form a noise feature spatial cloud map. In the noise feature space cloud map, directional interpolation and magnification processing is performed along the main direction of crack propagation; Using the bifurcation density information as weights, a convolution enhancement operation is performed on the noise feature space cloud map after interpolation and amplification. The enhanced noise feature spatial cloud map and the original corrosion crack map are fused at the pixel level in the same spatial coordinate system. The pixel-level fusion process adopts a weighted superposition algorithm based on the entropy weight method. The fused image is output as the hybrid risk map.
[0007] Preferably, multiple risk zones are defined, including: The hybrid risk map is rasterized to obtain a rasterized risk map containing multiple raster units; Calculate the mean and variance of the feature intensity within each grid cell; Based on the mean and variance of the feature intensity, the local risk clustering index of each grid cell is calculated; Arrange all grid cells in sequence according to pipeline mileage coordinates; Using a dynamic programming algorithm, with the local risk clustering index as input, and with the goal of minimizing the number of segments and minimizing the index differences within segments, the optimal set of segment division boundary points is solved. Based on the optimal set of boundary points for segment division, the continuous pipeline mileage is divided into the multiple risk segments.
[0008] Preferably, the calculation of the combustible migration rate of the gas mixture in each risk zone by the drive engine includes: For each risk segment, obtain the average feature intensity of the risk segment; Based on the component ratio of the mixed gas, the diffusion coefficient and density of the mixed gas are obtained by querying the physical property parameter library; A migration rate calculation model is established that correlates the average feature intensity with the diffusion coefficient and density; The average feature intensity and the queried physical property parameters are input into the migration rate calculation model; The migration rate calculation model outputs the theoretical combustible migration rate of the mixed gas in the corresponding risk zone.
[0009] Preferably, the calculation of the cumulative fatigue reduction factor of the pipe wall material under the mixed gas composition includes: Extract the service condition records of the target pipe section, which include the number of historical pressure cycles, pressure fluctuation amplitude, and the volume fraction of hydrogen in the gas transported in each cycle. Establish the mapping relationship between the hydrogen embrittlement sensitivity of pipe wall materials and the integral number of hydrogen gas and the amplitude of pressure fluctuation; Based on the mapping relationship, calculate the single fatigue damage increment in each pressure cycle, which is jointly affected by the hydrogen gas integral number and the pressure fluctuation amplitude at that time. For all pressure cycles in the service condition record, the corresponding single fatigue damage increment is calculated by summing them up to obtain the total cumulative fatigue damage. Input the total cumulative fatigue damage into the material property degradation curve and output the cumulative fatigue reduction factor.
[0010] Preferably, the current risk entropy value for each risk segment includes: Obtain the combustible migration rate of the mixed gas in the risk zone and the cumulative fatigue reduction factor; A risk entropy quantification function is established, wherein the flammable migration rate is used as the migration risk factor and the reciprocal of the cumulative fatigue reduction coefficient is used as the intensity attenuation factor. Substitute the migration risk factor and the intensity decay factor into the risk entropy quantification function; The risk entropy quantization function performs a logarithmic weighted operation and outputs the current risk entropy value.
[0011] Preferably, the differentiated maintenance strategy, which includes pressure control instructions and inspection routes, is issued to the maintenance unit belonging to the designated risk section. Establish a maintenance strategy mapping table, which defines the mapping relationship between risk entropy values of different ranges and corresponding maintenance instruction templates; Based on the current risk entropy value, query the maintenance strategy mapping table to obtain the corresponding maintenance instruction template; Obtain the start and end mileage coordinates of the target risk zone, the current risk entropy value, and the flammable migration rate; The start and end mileage coordinates, the current risk entropy value, and the combustible migration rate are filled into the matching maintenance instruction template to generate a complete differentiated maintenance strategy message containing pressure control instructions and inspection paths. The differentiated maintenance strategy message is sent to the maintenance unit to which the target risk section belongs through the communication interface of the pipeline control center.
[0012] Preferably, the process includes the following steps before generating the differentiated maintenance strategy message: Receive the available resource status and the current task queue from the maintenance unit; Analyze the estimated resource requirements in the differentiated maintenance strategy message; Compare the estimated resource requirements with the available resource status, and refer to the priority of the currently executing task queue; When resources are sufficient and the task queue allows, confirm the issuance of the differentiated maintenance strategy message; When resources or queues do not meet the requirements, a policy rescheduling is triggered. The rescheduling includes adjusting the time window of the inspection path or modifying the magnitude of the pressure control command.
[0013] Preferably, the present invention also includes a hydrogen-blended natural gas transportation safety risk assessment system, the system including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, it implements the steps of the hydrogen-blended natural gas transportation safety risk assessment method described above.
[0014] Compared with the prior art, the beneficial effects of the present invention are: By acquiring strain vibration waveforms through a distributed fiber optic sensor array deployed along the pipeline, and specifically identifying low-frequency noise characteristics caused by hydrogen-induced changes in material properties, this technology can directly capture specific vibration signals resulting from microstructural changes caused by hydrogen infiltration into the material. This enables online, real-time sensing of material degradation processes such as hydrogen embrittlement. This shifts the focus of safety monitoring from macroscopic operating parameters and post-leak alarms to the early stages of microscopic changes in material properties, gaining valuable time for risk warning.
[0015] Corrosion crack patterns of materials from the same batch are extracted from a historical defect database and overlaid with real-time acquired hydrogen-induced low-frequency noise features for reconstruction analysis, generating a hybrid risk map. This process is not a simple data juxtaposition, but rather a fusion and enhancement of risk characteristics from two different causes and periods in terms of features or spatial dimensions. This method overcomes the limitations of relying solely on historical data or current detection signals, ensuring that the final risk zones simultaneously carry the material's "historical memory" and "current state," improving the accuracy of risk location and the predictability of risk assessment. Risk zone division based on this map allows subsequent maintenance strategies to be more targeted at key locations where hydrogen damage and historical defects may have a synergistic effect. Attached Figure Description
[0016] Figure 1 This is a flowchart illustrating the steps of the hydrogen-blended natural gas transportation safety risk assessment method described in this invention. Figure 2 A flowchart for identifying the characteristics of hydrogen-induced low-frequency noise; Figure 3 A flowchart for generating a hybrid risk map based on overlay and reconstruction analysis; Figure 4 A bar chart comparing the average characteristic intensity of different risk sections of a hydrogen-blended natural gas pipeline; Figure 5 A comparison chart of strategic scheduling priorities and estimated resource requirements for hydrogen-blended natural gas pipeline maintenance tasks. Detailed Implementation
[0017] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0018] See Figure 1The software instance of the safety situation assessment engine is initialized in the server cluster of the pipeline control center, and the pre-configured mixed gas component ratio parameters, configured through a human-machine interface, are injected into the engine's configuration module. A distributed fiber optic sensor array, densely deployed along the target pipe section, collects raw data of the pipeline's strain and vibration waveforms during operation. A built-in signal processing algorithm identifies low-frequency noise features in the waveform data caused by hydrogen-induced changes in material properties. Historical corrosion crack detection maps of materials from the same batch as the target pipe section are extracted from the historical defect database connected to the pipeline control center. These corrosion crack maps, along with the currently collected low-frequency noise feature data, are input into the image analysis module for overlay and reconstruction analysis, generating a hybrid risk map that integrates historical defects and real-time features. Based on the feature intensity values of different spatial pixel locations in the hybrid risk map, a clustering segmentation algorithm divides the map into multiple continuous risk zones with varying risk levels. The injected mixed gas component ratio parameters are input into the core computing unit of the safety situation assessment engine, driving the unit to call the built-in model to calculate the flammable migration rate of the mixed gas within each divided risk zone. Simultaneously, the complete service condition record of the target pipe section is extracted from the pipeline control center's database, and the material performance analysis module is invoked to calculate the cumulative fatigue reduction factor of the pipe wall material under the current mixed gas composition. Combining the combustible migration rate and the cumulative fatigue reduction factor, the current risk entropy value of each risk section is calculated and quantified using a preset risk entropy quantification function. Finally, based on the calculated current risk entropy value, the pipeline control center's communication scheduling module automatically issues differentiated maintenance strategies, including specific pressure control instructions and recommended inspection routes, to the maintenance units belonging to the designated risk sections.
[0019] In one embodiment of the present invention, see [reference] Figure 2 A distributed fiber optic sensor array captures strain and vibration signals from the pipeline body, generating continuous time-domain waveform stream data. The signal processing unit performs adaptive wavelet packet decomposition on the time-domain waveform stream, extracting multiple sub-band components within a specified frequency range. The system acquires the baseline background noise spectrum recorded for the target pipeline section under pure natural gas transport conditions. From the decomposed sub-band components, frequency components matching the baseline background noise spectrum are filtered out using a spectral subtraction algorithm to obtain residual noise components. A Hilbert transform is performed on the residual noise components to calculate their time-frequency energy distribution. In the obtained time-frequency energy distribution map, continuous regions with energy above a set threshold and frequency below a preset threshold are located, and the center frequency and bandwidth values of these continuous regions are output as low-frequency noise features for identification.
[0020] In practical implementation, a 10-kilometer-long target pipeline section is used as an example. This section transports hydrogen-blended natural gas with a hydrogen gas integral of 10%. A distributed fiber optic sensor array is deployed on the outer wall of the target pipeline section at 1-meter intervals. The distributed fiber optic sensor array captures the strain and vibration signals of the pipeline body, generating continuous time-domain waveform stream data with a sampling frequency of 10 kHz. This time-domain waveform stream data contains real-time strain and vibration information of the pipeline during operation. The signal processing unit performs adaptive wavelet packet decomposition on the time-domain waveform stream, using a 6-level decomposition depth to extract multiple sub-band components within the frequency range of 0 Hz to 500 Hz. The system obtains the baseline background noise spectrum of the target pipeline section under pure natural gas transportation conditions from a historical database. The baseline background noise spectrum is generated by statistically analyzing 24 hours of continuous vibration data under stable operating conditions during pure natural gas transportation. From the decomposed sub-band components, a spectral subtraction algorithm is used to filter out frequency components that match the baseline background noise spectrum. The spectral subtraction algorithm subtracts the power spectrum of the sub-band component from the power spectrum of the baseline background noise spectrum point by point to obtain the residual noise component. A Hilbert transform is applied to the residual noise component to calculate its time-frequency energy distribution, which is presented as energy density in the time-frequency plane. In practice, a continuous region with energy above a set threshold and frequency below a preset threshold is located. The set threshold is three times the global average energy of the time-frequency energy distribution, and the preset threshold is 50 Hz. The center frequency and bandwidth of the continuous region are output as the low-frequency noise features for identification.
[0021] It is understandable that the sampling frequency of the time-domain waveform stream data needs to be configured based on the pipe diameter and the expected hydrogen-induced noise frequency range to ensure coverage of the possible frequency bands of low-frequency noise characteristics. It is also understandable that the spectral subtraction coefficient in the spectral subtraction algorithm is set to 0.8 to balance background noise suppression with the risk of signal distortion. Optionally, the number of layers in the adaptive wavelet packet decomposition can be adjusted to 5 or 7 layers depending on the pipe material characteristics and sensor accuracy. Optionally, the threshold and preset threshold can be manually calibrated through the human-machine interface based on historical recognition results.
[0022] In practical implementation, to quantify the energy distribution of subband components, a formula is introduced to calculate the normalized energy of subband components: in: This represents the normalized energy of the b-th subband component. This represents the coefficient value of the b-th sub-band component at the n-th sampling point, where N represents the total number of sampling points for that sub-band component. This formula is used to evaluate the energy contribution of each sub-band after decomposition, assisting in the selection of a specified frequency band range. Data comparison shows that when transporting pure natural gas, the normalized average energy of the sub-band components in the 0 Hz to 50 Hz frequency band is less than 0.01; after transporting hydrogen-blended natural gas, the normalized average energy of the sub-band components in the same frequency band rises to above 0.05, and the residual noise component shows a significant continuous energy accumulation from 10 Hz to 30 Hz in the time-frequency energy distribution. In specific implementation, the time-frequency energy distribution matrix obtained after Hilbert transform is used for scanning and positioning, with a scanning step size of 0.1 Hz. The identified continuous region must meet the conditions of a duration greater than 0.1 seconds and a bandwidth greater than 5 Hz, and the final output is a low-frequency noise characteristic parameter such as a center frequency of 20 Hz and a bandwidth of 8 Hz.
[0023] In one embodiment of the present invention, see [reference] Figure 3 From corrosion crack maps, image morphology analysis is used to identify the main direction of crack propagation and the bifurcation density per unit area. Low-frequency noise feature data is mapped to the sensor's spatial coordinates at the time of acquisition, forming a noise feature spatial cloud map in the pipeline's spatial coordinates. In this noise feature spatial cloud map, directional interpolation amplification is performed along the main direction of crack propagation. Using crack bifurcation density information as weighting coefficients, convolution enhancement is performed on the interpolated and amplified noise feature spatial cloud map. The enhanced noise feature spatial cloud map and the original corrosion crack map are then fused pixel-by-pixel in the same pipeline spatial coordinate system using a weighted superposition algorithm based on entropy weighting. The image analysis module outputs the fused image as a hybrid risk map.
[0024] In this implementation, a five-kilometer-long X70 steel target pipe section is used as an example. The corrosion crack pattern is derived from ultrasonic scan images of the same batch of material from the previous inspection cycle stored in the historical defect database. The low-frequency noise characteristics are derived from the center frequency and bandwidth data sequence output by the example. The main direction of crack propagation and bifurcation density information are identified from the corrosion crack pattern. Image morphological analysis uses Hough transform to detect the inclination angle of the main crack line, and the peak value of the inclination angle distribution histogram is used as the main direction of propagation. The bifurcation density information is obtained by calculating the number of branch nodes per unit area after skeletonization analysis. The low-frequency noise characteristics are mapped onto the pipe spatial coordinates to form a noise characteristic spatial cloud map. The mapping process is based on the physical location coordinates of the distributed fiber optic sensor array. The center frequency value and bandwidth value identified by each sensor are assigned to the corresponding pipe mileage coordinate point, and a continuous two-dimensional spatial cloud map is generated by Kriging space interpolation. In the noise feature space cloud map, directional interpolation amplification is performed along the main crack propagation direction. Cubic spline interpolation is used to increase the spatial resolution of the original cloud map by five times along the main propagation direction, while maintaining the original resolution in the vertical direction. Using bifurcation density information as weights, convolution enhancement is performed on the amplified noise feature space cloud map. A 3x3 convolution kernel is used, and the center weight coefficient of the kernel is determined by normalizing the bifurcation density values at the corresponding positions. The enhanced noise feature space cloud map and the original corrosion crack map are then fused pixel-wise in the same spatial coordinate system. The fusion process uses a weighted superposition algorithm based on entropy weighting. This algorithm calculates the information entropy of both the noise feature space cloud map and the original corrosion crack map, and uses this to determine their respective fusion weight coefficients. The fused image is output as a hybrid risk map.
[0025] It is understandable that the variogram model chosen in the Kriging space interpolation method is a spherical model, and the interpolation search radius is the distance between three adjacent sensors. It is also understandable that in the weighted superposition algorithm based on the entropy weight method, the information entropy is calculated based on the image grayscale histogram. Optionally, in the directional interpolation amplification process, if the main expansion direction is not unique, interpolation amplification is performed separately for each of the identified main directions, and the results are then superimposed and averaged. Optionally, the convolution kernel size for the convolution enhancement operation can be adjusted to 5x5 according to the overall image resolution.
[0026] In practical implementation, to quantify the influence of bifurcation density information on the weights of the convolution enhancement process, a method is introduced to calculate the weight coefficients of the convolution kernel center. The formula: in: Represents the center weight coefficient of the convolution kernel. This represents the bifurcation density value of the local region corresponding to the current center position of the convolution kernel. This represents the maximum bifurcation density value in the entire corrosion crack pattern. This represents the minimum bifurcation density value in the entire corrosion crack map. The formula linearly maps the local density to a range of 0.1 to 1.0 as the core weight. Data comparison shows that before using bifurcation density weights for convolutional enhancement, the average gray-level gradient of the noise feature space map in the crack-dense region is 15; after applying bifurcation density weights for convolutional enhancement, the average gray-level gradient in the same region increases to 28. In specific implementation, the weighted superposition algorithm based on entropy weighting calculates the information entropy of the original corrosion crack map to be 5.2, and the information entropy of the enhanced noise feature space map to be 4.8, thus determining the fusion weight coefficients to be 0.52 and 0.48, respectively. Pixel-level fusion weights each corresponding pixel of the two images according to the weight coefficients, generating a new grayscale image as the final hybrid risk map.
[0027] In one embodiment of the present invention, the hybrid risk map is rasterized, dividing it into multiple regularly arranged raster units to obtain a rasterized risk map. The mean and variance of the feature intensity of all pixels within each raster unit are calculated. Based on the mean and variance of the feature intensity of each raster unit, the local risk clustering index of that unit is calculated. All raster units are arranged in spatial sequence according to the pipeline mileage coordinates. A dynamic programming algorithm is used, taking the local risk clustering index sequence as input, with the optimization objective of minimizing the number of segments and minimizing the index differences within segments, to solve for the optimal set of segment division boundary points. Based on the solved optimal set of segment division boundary points, the continuous pipeline mileage is divided into multiple risk segments.
[0028] In practice, the hybrid risk map is a grayscale image with a resolution of 1000 pixels × 1000 pixels, where each pixel corresponds to a 0.01-meter by 0.01-meter pipe surface area. The hybrid risk map is rasterized, dividing the image spatially into a regular grid of 100 rows and 1000 columns, resulting in a rasterized risk map containing 100,000 raster cells, each representing a 10-meter pipe segment. The mean and variance of the feature intensity within each raster cell are calculated. The mean feature intensity is obtained by taking the arithmetic mean of the grayscale values of all pixels within the raster cell, and the variance is obtained by calculating the dispersion of these grayscale values relative to the mean feature intensity. Based on the mean and variance of the feature intensity, a local risk clustering index is calculated for each raster cell. According to the pipe mileage coordinates, all raster cells are arranged from left to right and from top to bottom, forming a one-dimensional raster cell sequence. A dynamic programming algorithm is employed, using a sequence of local risk clustering indicators as input, with the objective of minimizing the number of segments and the differences in indicators within each segment, to find the optimal set of segment division boundary points. Based on this optimal set of boundary points, a continuous pipeline mileage is divided into multiple risk segments.
[0029] It is understandable that the feature intensity variance is calculated using the sample variance formula, with the denominator being the number of pixels in the raster cell minus one. It is also understandable that the state transition equation of the dynamic programming algorithm includes a penalty coefficient to balance the number of segments and the consistency within each segment. Optionally, when the pipeline has a loop or branching structure, the hybrid risk map needs to be expanded into a linear representation before rasterization and sorting. Optionally, the optimal segment division boundary point set obtained from the solution will be mapped back to the original two-dimensional hybrid risk map coordinates to indicate the actual spatial extent of each risk segment.
[0030] In practical implementation, to quantify the local risk clustering index Introducing the formula: in: This represents the local risk clustering index of the j-th grid cell. This represents the mean feature intensity of the j-th grid cell. This represents the variance of the feature intensity of the j-th grid cell. This is a preset balancing parameter used to adjust the weighting of variance in the indicator. The formula combines the average risk level with the volatility of the risk distribution. Data comparison shows that a raster cell with a mean feature strength of 85 and a variance of 120... When set to 0.1, its local risk clustering index is 97; another raster cell with a mean feature intensity of 88 and a variance of 400, under the same... At this value, the local risk clustering index is 128, and the higher variance leads to a significant increase in the index. In specific implementation, a balance parameter is set. The value is 0.05, and the penalty coefficient for the dynamic programming algorithm is set to 0.5. The solution process traverses all possible partitioning schemes and finally outputs a set of partitioning boundary points. For example, the boundaries are set at the 25001st, 58000th, and 87000th cells in a sequence of 100000 grid cells, thereby dividing the 10-kilometer pipeline into four consecutive risk sections with lengths of 2.5 km, 3.3 km, 2.9 km, and 1.3 km, respectively.
[0031] In one embodiment of the present invention, for each defined risk segment, the average value of the characteristic intensity of all grid cells within that segment is obtained as the average characteristic intensity of that segment. Based on the configured proportion of the mixed gas components, a pre-set physical property parameter library is consulted to obtain the diffusion coefficient and density parameters of the mixed gas at that specific proportion. The migration rate calculation model is a correlation model fitted using historical data, which correlates the average characteristic intensity with the diffusion coefficient and density. The average characteristic intensity of the risk segment and the queried physical property parameters are input into the migration rate calculation model, and the model outputs the theoretical flammable migration rate of the mixed gas in the corresponding risk segment. Service condition records of the target pipe segment are extracted, including the number of historical pressure cycles, the pressure fluctuation amplitude of each cycle, and the volume fraction of hydrogen in the transported gas each time. The material performance analysis module establishes a mapping relationship between the hydrogen embrittlement sensitivity of the pipe wall material and the hydrogen gas integral and pressure fluctuation amplitude. Based on this mapping relationship, the single fatigue damage increment under the combined effect of the hydrogen gas integral and pressure fluctuation amplitude in each historical pressure cycle is calculated. For all pressure cycles recorded in the service condition data, the corresponding single-cycle fatigue damage increment is calculated by summing them up to obtain the total cumulative fatigue damage. The total cumulative fatigue damage is then input into the material property degradation curve, and the cumulative fatigue reduction factor of the pipe wall material is output. The risk entropy quantification function uses the combustible migration rate as the migration risk factor and the reciprocal of the cumulative fatigue reduction factor as the strength attenuation factor. Substituting the migration risk factor and the strength attenuation factor into the risk entropy quantification function, this function performs a logarithmic weighted calculation and outputs the current risk entropy value.
[0032] In this specific implementation, a 10-kilometer-long X70 steel pipeline transporting hydrogen-blended natural gas with a hydrogen gas fraction of 10% is used as an example. This pipeline has been divided into four consecutive risk sections according to the embodiment, with lengths of 2.5 km, 3.3 km, 2.9 km, and 1.3 km, respectively. For each risk section, the average of the characteristic intensity of all grid cells within that section is obtained as the average characteristic intensity of the risk section. The calculated average characteristic intensities for the four risk sections are 72, 115, 88, and 60, respectively. Based on the mixed gas component ratio injected into the safety situation assessment engine, a pre-set physical property parameter library is consulted. This library stores the diffusion coefficient and density parameters of mixed gases with different hydrogen gas fractions. For hydrogen-blended natural gas with a hydrogen gas fraction of 10%, the diffusion coefficient under standard conditions is found to be 0.15 cm² / s and the density to be 0.65 kg / m³. The migration rate calculation model is a multinomial regression model obtained by fitting historical experimental data. The migration rate calculation model correlates the average characteristic intensity with the diffusion coefficient and density. The average characteristic intensity of each risk zone, along with the queried diffusion coefficient and density, is input into the migration rate calculation model. The migration rate calculation model outputs the theoretical combustible migration rate of the mixed gas in the corresponding risk zone. The calculation results are 0.22 m / s, 0.38 m / s, 0.28 m / s, and 0.18 m / s, respectively.
[0033] Service condition records for the target pipe section are extracted from the pipeline control center's historical monitoring database, containing records of every pressure cycle over the past five years. Each record includes the number of historical pressure cycles, pressure fluctuation amplitude, and the volume fraction of hydrogen in the transported gas for each cycle. The material performance analysis module establishes a mapping relationship between the pipe wall material's hydrogen embrittlement sensitivity and the hydrogen gas integral and pressure fluctuation amplitude, stored as a 3D surface plot. Based on this mapping relationship, the incremental fatigue damage per cycle, influenced by the combined effects of the hydrogen gas integral and pressure fluctuation amplitude at that time, is calculated for each historical pressure cycle. For all pressure cycles in the service condition records, the corresponding incremental fatigue damage per cycle is summed to obtain the total cumulative fatigue damage. This total cumulative fatigue damage is input into the material performance degradation curve, which describes the relationship between the fatigue damage of X70 steel and the residual strength coefficient, outputting the cumulative fatigue reduction factor for the pipe wall material. In some embodiments, the query process of the physical property parameter library considers the current average temperature and pressure conditions of the pipeline, correcting for the diffusion coefficient and density. In some embodiments, if data is missing from the historical pressure cycle records, the average value of adjacent time periods is used for interpolation.
[0034] It is understandable that the migration rate calculation model is in the form of a multivariate quadratic polynomial, with its coefficients determined by fitting a large amount of experimental data using the least squares method. It is also understandable that the calculation of the incremental fatigue damage per instance uses a modified form of the Paris formula based on the stress intensity factor amplitude. Optionally, the material property degradation curve can be fine-tuned according to the batch and heat treatment history of the pipe material. Optionally, the calculation of the total cumulative fatigue damage uses the linear cumulative damage rule.
[0035] Referring to Table 1, the risk entropy quantification function uses the flammable migration rate as the migration risk factor and the reciprocal of the cumulative fatigue reduction coefficient as the intensity attenuation factor. The migration risk factor and intensity attenuation factor are substituted into the risk entropy quantification function. The risk entropy quantification function performs a logarithmic weighted operation and outputs the current risk entropy value. The calculated cumulative fatigue reduction coefficients for the four risk zones are 0.92, 0.85, 0.89, and 0.94, respectively, and the calculated current risk entropy values are 1.84, 3.15, 2.33, and 1.41, respectively.
[0036] Table 1: Comparison of Safety Parameter Calculation Results for Risk Sections In practical implementation, to perform the logarithmic weighted operation in the risk entropy quantification function, the following formula is introduced: in: This represents the current risk entropy value. This represents the migration risk factor, namely the flammable migration rate. This represents the cumulative fatigue reduction factor. and These are preset weighting coefficients used to balance the contributions of the two factors, set to 2.0 and 1.5 respectively. This formula ensures that when the flammable migration rate increases or the cumulative fatigue reduction factor decreases, the current risk entropy value remains constant. Monotonically increasing. The data comparison shows that for segment 2, [the following is a more detailed description:] and Substituting into the formula, we can calculate... For section 4, and Substituting into the formula, we can calculate... The risk entropy value of segment 2 is significantly higher than that of segment 4.
[0037] See Figure 4This is a bar chart comparing the average characteristic intensity of different risk sections in a hydrogen-blended natural gas pipeline. It visually demonstrates the differences in core risk indicators across the four risk sections, serving as a crucial basis for subsequent risk assessment and maintenance decisions. Section 2 is the longest (3.3 km) and has the highest risk, while the shortest section 4 (1.3 km) has the lowest risk, indicating that the degree of risk is primarily determined by the pipeline's material condition and damage, rather than simply its length. This chart directly provides the foundational data for subsequent calculations of the combustible migration rate, cumulative fatigue reduction factor, and final risk entropy value, serving as the starting point for the entire risk assessment process. Clear risk level stratification allows for precise implementation of maintenance strategies. For example, high-risk section 2 should be prioritized for inspections and pressure control, while low-risk section 4 can have its maintenance frequency appropriately reduced, thereby optimizing resource allocation.
[0038] In one embodiment of the present invention, a maintenance strategy mapping table defines the mapping relationship between risk entropy values of different numerical ranges and corresponding maintenance instruction templates. Based on the calculated current risk entropy value, the maintenance strategy mapping table is queried to obtain the corresponding maintenance instruction template. The system obtains the start and end mileage coordinates, current risk entropy value, and combustible migration rate of the target risk section. The start and end mileage coordinates, current risk entropy value, and combustible migration rate are used as variables to fill the matched maintenance instruction template, generating a complete differentiated maintenance strategy message containing pressure control instructions and inspection paths. After generating the differentiated maintenance strategy message, the system receives the available resource status information and the current task queue from the target maintenance unit. The estimated resource demand type and quantity in the differentiated maintenance strategy message are parsed. The estimated resource demand is compared with the available resource status reported by the maintenance unit, and the priority setting of the current task queue is referenced. When resources are sufficient and the task queue allows the insertion of new tasks, the differentiated maintenance strategy message is sent to the maintenance unit to which the target risk section belongs via the pipeline control center's communication interface. When resources or task queues do not meet requirements, a policy rescheduling process is triggered. Rescheduling includes adjusting the time window of the inspection path or modifying the magnitude of the pressure control command.
[0039] In the specific implementation, the current risk entropy values of the four calculated risk sections are used as an example. Risk section 2, with a current risk entropy value of 3.15, is identified as a target risk section requiring priority handling. A maintenance strategy mapping table is established and stored in the pipeline control center's strategy database in spreadsheet format. This table defines the mapping relationship between risk entropy values of different ranges and corresponding maintenance instruction templates. Based on the current risk entropy value of 3.15 for target risk section 2, the maintenance strategy mapping table is queried to match the maintenance instruction templates corresponding to risk entropy values ranging from 2.5 to 4.0. These templates include the framework of pressure control instructions and inspection paths. The start and end mileage coordinates, current risk entropy value, and combustible migration rate of target risk section 2 are obtained. The start and end mileage coordinates are from 25.0 km to 28.3 km, and the combustible migration rate is 0.38 m / s. The start and end mileage coordinates, current risk entropy value, and combustible migration rate are filled into the matching maintenance instruction template to generate a complete differentiated maintenance strategy message containing pressure control instructions and inspection paths. The pressure control instruction is "reduce the pressure of the upstream throttle valve by 0.5 MPa in the 25.0-28.3 km section", and the inspection path is "the UAV inspection path plan covers the 25.0-28.3 km section, focusing on scanning welds and anti-corrosion layers".
[0040] After generating the differentiated maintenance strategy message, the pipeline control center automatically sends a resource status query request to the third maintenance unit responsible for target risk section 2. It receives feedback from the third maintenance unit regarding available resource status and the current task queue. The available resource status shows that two inspection vehicles are on standby, one drone is available, and the current task queue includes a routine maintenance task with a priority of "medium." The center parses the estimated resource requirements from the differentiated maintenance strategy message, which are estimated to require one drone to operate continuously for two hours and one inspection vehicle for ground support. It compares the estimated resource requirements with the available resource status reported by the third maintenance unit, and considers the priority of the current task queue. When resources are sufficient and the task queue allows, the center confirms the issuance of the differentiated maintenance strategy message to the third maintenance unit to which target risk section 2 belongs via the pipeline control center's communication interface. In some embodiments, the maintenance strategy mapping table is periodically reviewed and updated according to the pipeline company's safety procedures. In some embodiments, the resource status query request and feedback are transmitted via a dedicated secure wireless communication protocol.
[0041] It is understandable that the parsing of estimated resource requirements is based on predefined resource tags in the maintenance instruction template. It is also understandable that the priority determination of the task queue is based on a preset priority encoding rule. Optionally, when multiple risk segments need to issue policies to the same maintenance unit, the pipeline control center will queue the policy messages. Optionally, the current execution task queue information fed back by the maintenance unit includes the estimated completion time of each task.
[0042] When resources or queues are insufficient, a policy rescheduling process is triggered. For example, if the third maintenance unit reports that there is a high-priority emergency repair task in the current task queue, the system determines that the task queue does not allow the immediate insertion of a new inspection task. Policy rescheduling includes adjusting the time window of the inspection path or modifying the magnitude of the pressure control command. Adjusting the time window of the inspection path may postpone an inspection originally planned to be executed immediately to four hours later, and modifying the magnitude of the pressure control command may change the pressure reduction value from 0.5 MPa to 0.3 MPa as a temporary transitional measure.
[0043] In practical implementation, to quantify the judgment of task queue permissibility, a method for calculating policy scheduling priority is introduced. The formula: in: Indicates the policy scheduling priority, This represents the current risk entropy value. This indicates the estimated time required for a task to wait for a higher-priority task in the queue to complete if the task is delayed. This formula is used to compare the urgency of different pending policy messages when resource conflicts occur. Data comparison shows that a task with a current risk entropy value of 3.15 and a waiting time of 2 hours has different policy scheduling priorities. The calculated value is 1.05; another task with a current risk entropy value of 2.33, which does not require waiting, has a policy scheduling priority. The calculated value is 2.33. Although the former has a higher risk entropy value, its calculated scheduling priority is actually lower than the latter due to the waiting time. This can guide the order decision during rescheduling. After a rescheduling is triggered, the system will attempt to prioritize the policy scheduling. Tasks with higher computational values will be prioritized for resource allocation or instruction adjustment.
[0044] See Figure 5This is a chart comparing the strategic scheduling priority and estimated resource requirements of hydrogen-blended natural gas pipeline maintenance tasks. Using a dual Y-axis design, it clearly shows the correspondence between the scheduling priority and resource consumption of the five maintenance tasks. This chart provides crucial information for the dynamic scheduling of maintenance strategies, revealing the trade-off between the "value" and "cost" of each task. Task 1, a high-value, medium-cost task, should be executed immediately. For high-cost tasks like Task 2 in risky areas, the decision to execute immediately or postpone it depends on the available resources and task queue priority. The chart clearly quantifies the "strategic scheduling priority" and "estimated resource requirements" of each maintenance task, intuitively revealing the contradiction between the two.
[0045] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.
Claims
1. A method for assessing the safety risks of hydrogen-blended natural gas transportation, characterized in that, include: Initialize the safety situation assessment engine at the pipeline control center and inject a pre-configured mixture of gas components into the safety situation assessment engine; By deploying a distributed fiber optic sensor array along the target pipe section, the strain vibration waveform of the pipe body is acquired, and the low-frequency noise characteristics caused by hydrogen-induced changes in material properties are identified in the waveform. Corrosion crack patterns of materials from the same batch as the target pipe section are extracted from the pipeline historical defect database. The corrosion crack patterns are then overlaid and reconstructed with the currently collected low-frequency noise features to generate a hybrid risk pattern. Based on the characteristic intensity of different spatial locations in the hybrid risk map, multiple risk zones are divided; The proportions of the mixed gas components are input into the safety situation assessment engine, which drives the engine to calculate the flammable migration rate of the mixed gas in each risk zone. Extract the service condition records of the target pipe section and calculate the cumulative fatigue reduction factor of the pipe wall material under the mixed gas composition; By combining the flammable migration rate and the cumulative fatigue reduction factor, the current risk entropy value of each risk segment is quantified; Based on the current risk entropy value, the pipeline control center issues differentiated maintenance strategies, including pressure regulation instructions and inspection routes, to the maintenance units belonging to the designated risk sections.
2. The method for assessing the safety risks of hydrogen-blended natural gas transportation as described in claim 1, characterized in that, The low-frequency noise characteristics in the waveform caused by hydrogen-induced changes in material properties include: The original strain vibration signal is captured by a distributed fiber optic sensor array, generating a time-domain waveform stream; Adaptive wavelet packet decomposition is performed on the time-domain waveform stream to extract sub-band components within a specified frequency band range; Obtain the reference background noise spectrum of the target pipeline segment when transporting pure natural gas; The residual noise component is obtained by filtering out the components that match the reference background noise spectrum from the sub-band components. Perform a Hilbert transform on the residual noise component to obtain its time-frequency energy distribution; In the time-frequency energy distribution, a continuous region with energy higher than a set threshold and frequency lower than a preset threshold is located, and the center frequency and bandwidth of the continuous region are used as the low-frequency noise characteristics.
3. The method for assessing the safety risks of hydrogen-blended natural gas transportation as described in claim 1, characterized in that, The reconstruction analysis by overlaying the corrosion crack pattern with the currently acquired low-frequency noise features includes: Identify the main propagation direction and branching density information of the cracks from the corrosion crack pattern; The low-frequency noise features are mapped onto the pipe spatial coordinates to form a noise feature spatial cloud map. In the noise feature space cloud map, directional interpolation and magnification processing is performed along the main direction of crack propagation; Using the bifurcation density information as weights, a convolution enhancement operation is performed on the noise feature space cloud map after interpolation and amplification. The enhanced noise feature spatial cloud map and the original corrosion crack map are fused at the pixel level in the same spatial coordinate system. The pixel-level fusion process adopts a weighted superposition algorithm based on the entropy weight method. The fused image is output as the hybrid risk map.
4. The method for assessing the safety risks of hydrogen-blended natural gas transportation as described in claim 1, characterized in that, Multiple risk zones were identified, including: The hybrid risk map is rasterized to obtain a rasterized risk map containing multiple raster units; Calculate the mean and variance of the feature intensity within each grid cell; Based on the mean and variance of the feature intensity, the local risk clustering index of each grid cell is calculated; Arrange all grid cells in sequence according to pipeline mileage coordinates; Using a dynamic programming algorithm, with the local risk clustering index as input, and with the goal of minimizing the number of segments and minimizing the index differences within segments, the optimal set of segment division boundary points is solved. Based on the optimal set of boundary points for segment division, the continuous pipeline mileage is divided into the multiple risk segments.
5. The method for assessing the safety risks of hydrogen-blended natural gas transportation as described in claim 4, characterized in that, The engine calculates the flammable migration rate of the gas mixture in each risk zone, including: For each risk segment, obtain the average feature intensity of the risk segment; Based on the component ratio of the mixed gas, the diffusion coefficient and density of the mixed gas are obtained by querying the physical property parameter library; A migration rate calculation model is established that correlates the average feature intensity with the diffusion coefficient and density; The average feature intensity and the queried physical property parameters are input into the migration rate calculation model; The migration rate calculation model outputs the theoretical combustible migration rate of the mixed gas in the corresponding risk zone.
6. The method for assessing the safety risks of hydrogen-blended natural gas transportation as described in claim 5, characterized in that, The calculation of the cumulative fatigue reduction factor for the pipe wall material under the aforementioned gas mixture includes: Extract the service condition records of the target pipe section, which include the number of historical pressure cycles, pressure fluctuation amplitude, and the volume fraction of hydrogen in the gas transported in each cycle. Establish the mapping relationship between the hydrogen embrittlement sensitivity of pipe wall materials and the integral number of hydrogen gas and the amplitude of pressure fluctuation; Based on the mapping relationship, calculate the single fatigue damage increment in each pressure cycle, which is jointly affected by the hydrogen gas integral number and the pressure fluctuation amplitude at that time. For all pressure cycles in the service condition record, the corresponding single fatigue damage increment is calculated by summing them up to obtain the total cumulative fatigue damage. Input the total cumulative fatigue damage into the material property degradation curve and output the cumulative fatigue reduction factor.
7. The method for assessing the safety risks of hydrogen-blended natural gas transportation as described in claim 6, characterized in that, The current risk entropy value for each risk segment is quantified as follows: Obtain the combustible migration rate of the mixed gas in the risk zone and the cumulative fatigue reduction factor; A risk entropy quantification function is established, wherein the flammable migration rate is used as the migration risk factor and the reciprocal of the cumulative fatigue reduction coefficient is used as the intensity attenuation factor. Substitute the migration risk factor and the intensity decay factor into the risk entropy quantification function; The risk entropy quantization function performs a logarithmic weighted operation and outputs the current risk entropy value.
8. The method for assessing the safety risks of hydrogen-blended natural gas transportation as described in claim 7, characterized in that, Differentiated maintenance strategies, including pressure control instructions and inspection routes, are issued to the maintenance units belonging to designated risk areas. Establish a maintenance strategy mapping table, which defines the mapping relationship between risk entropy values of different ranges and corresponding maintenance instruction templates; Based on the current risk entropy value, query the maintenance strategy mapping table to obtain the corresponding maintenance instruction template; Obtain the start and end mileage coordinates of the target risk zone, the current risk entropy value, and the flammable migration rate; The start and end mileage coordinates, the current risk entropy value, and the combustible migration rate are filled into the matching maintenance instruction template to generate a complete differentiated maintenance strategy message containing pressure control instructions and inspection paths. The differentiated maintenance strategy message is sent to the maintenance unit to which the target risk section belongs through the communication interface of the pipeline control center.
9. The method for assessing the safety risks of hydrogen-blended natural gas transportation as described in claim 8, characterized in that, The process before generating the differentiated maintenance policy message also includes: Receive the available resource status and the current task queue from the maintenance unit; Analyze the estimated resource requirements in the differentiated maintenance strategy message; Compare the estimated resource requirements with the available resource status, and refer to the priority of the currently executing task queue; When resources are sufficient and the task queue allows, confirm the issuance of the differentiated maintenance strategy message; When resources or queues do not meet the requirements, a policy rescheduling is triggered. The rescheduling includes adjusting the time window of the inspection path or modifying the magnitude of the pressure control command.
10. A safety risk assessment system for the transportation of hydrogen-blended natural gas, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for assessing the safety risks of hydrogen-blended natural gas transportation as described in any one of claims 1 to 9.