Pressure vessel safety evaluation method based on big data and multi-parameter coupling

Abstract

The present application belongs to the technical field of industrial data processing, and particularly relates to a pressure vessel safety evaluation method based on big data and multi-parameter coupling. The method steps include: obtaining multi-dimensional monitoring data of the pressure vessel, including internal temperature and pressure data, external image data and equipment basic data; performing time series analysis on the internal temperature and pressure data, and determining the internal safety quality score in combination with the synergistic factor; pre-processing the external image and extracting the scratch feature quantization index and the shape integrity index to obtain the external safety quality score; weighting and summing the internal and external safety quality scores to obtain the safety quality score; correcting the initial safety evaluation threshold value according to the thickness and crack parameters of the pressure vessel, comparing the comprehensive score with the dynamic threshold value, and determining the safety state of the pressure vessel. The present application realizes multi-dimensional comprehensive evaluation, improves the accuracy and real-time performance, and provides reliable support for the safe operation of the pressure vessel.

Description

Technical Field

[0001] This invention belongs to the field of industrial data processing technology, and in particular relates to a pressure vessel safety assessment method based on big data and multi-parameter coupling. Background Technology

[0002] Pressure vessels, as key equipment in core industrial sectors such as petroleum, chemical, and energy, are crucial for ensuring continuous production and preventing safety accidents. Pressure vessels often operate under complex conditions including high temperature, high pressure, and corrosive media. Traditional safety management methods relying on periodic manual inspections or single-parameter monitoring are no longer sufficient to capture potential safety hazards in real time and comprehensively, failing to meet the accuracy and dynamic requirements of modern industrial safety assessments. Existing pressure vessel safety assessment technologies have significant shortcomings: First, limited monitoring dimensions, with most technologies focusing only on single parameters such as internal temperature and pressure, leading to incomplete assessment information; second, rigid threshold settings, with fixed initial safety assessment thresholds that do not consider influencing parameters such as equipment thickness and internal wall cracks, making it impossible to adapt to safety standards at different stages of the equipment's lifecycle; this easily leads to assessment biases or missed safety hazards, failing to provide reliable technical support for the safe operation of pressure vessels. Summary of the Invention

[0003] To address the technical problems existing in the background art described above, this invention proposes a pressure vessel safety assessment method based on big data and multi-parameter coupling.

[0004] To achieve the above objectives, the technical solution adopted by the present invention includes the following steps:

[0005] Acquire multi-dimensional monitoring data of the pressure vessel, including internal monitoring data, external monitoring data, and equipment basic data; wherein the internal monitoring data includes internal temperature data and pressure data; the external monitoring data includes image data of the external pressure vessel; and the equipment basic data includes pressure vessel thickness and crack parameter data.

[0006] Based on the internal temperature and internal pressure data at all acquisition times, the internal safety quality score of the pressure vessel is determined through time-series data analysis and data standardization scores.

[0007] The image data in the external monitoring data is preprocessed and features are extracted to obtain the shape integrity index and scratch feature quantification index of the pressure vessel. Based on the shape integrity index and scratch feature quantification index, the external safety quality score of the pressure vessel is determined.

[0008] The safety quality score of the pressure vessel is obtained by weighted summation of the internal safety quality score and the external safety quality score.

[0009] Based on the pressure vessel thickness and crack parameter data in the equipment's basic data, the preset initial safety assessment threshold is dynamically corrected to obtain the dynamic safety assessment threshold.

[0010] The safety quality score is compared with the dynamic safety assessment threshold. If the safety quality score is lower than the dynamic safety assessment threshold, the pressure vessel is determined to be in an unsafe state; if the safety quality score is greater than or equal to the dynamic safety assessment threshold, the pressure vessel is determined to be in a safe state.

[0011] Preferably, the specific implementation of determining the internal safety quality score of the pressure vessel based on internal temperature and internal pressure data at all acquisition times, through time-series data analysis and combined with data standardization scores, includes:

[0012] The collected time-series temperature and pressure data sequences were reconstructed using nonlinear fitting algorithms to obtain temperature fitting curves in the continuous time dimension. and pressure fitting curve , where t is the time index;

[0013] Extract a peak and a trough from the temperature time-series fitting curve, and record the time period from the corresponding trough acquisition time to the adjacent peak acquisition time as a complete operating condition fluctuation period of the pressure vessel; at the same time, collect the pressure fitting curve of the same complete operating condition fluctuation period.

[0014] The temperature change rate curve is obtained by taking the first derivative of the temperature fitting curve for a complete operating condition fluctuation period. The curve of the degree of temperature fluctuation is obtained by taking the second derivative. The pressure change rate curve was obtained in the same way. And the curve of the intensity of pressure fluctuation ;

[0015] Calculate temperature rate deviation ,in, These represent the start and end times of a complete operating condition fluctuation segment. The maximum and minimum values ​​of the pre-set safety threshold for the rate of temperature change are determined; the temperature fluctuation deviation is also calculated. ,in The maximum and minimum values ​​of the pre-defined temperature fluctuation deviation threshold are used; and the pressure rate deviation is obtained using the same logic. Deviation from pressure fluctuation ;

[0016] Define rate co-factor ,in It is an extremely small constant; the volatility co-factor is defined in the same way. Then, the rate synergy factor and the fluctuation synergy factor are fused by weighted summation to obtain the comprehensive synergy deviation coefficient C;

[0017] The internal safety quality score is obtained by using an exponential function. .

[0018] Preferably, the specific implementation of preprocessing and feature extraction of image data from external monitoring data to obtain the shape integrity index and scratch feature quantification index of the pressure vessel is as follows:

[0019] Synchronously acquire RGB images of the pressure vessel and polarization image The spatial transformation matrix is ​​calculated based on the SIFT feature point matching algorithm, and the polarization degree image is registered to the RGB image coordinate system to obtain the registered polarization degree map. ;

[0020] right The image is decomposed into illumination and reflection components. The light component is ,in, Gaussian kernels of 3 scales, It is a very small constant. This is the polarization suppression coefficient;

[0021] Using MobileNetV2 network Semantic segmentation is performed to obtain the pressure vessel region mask. ,in This is indicated as the pressure vessel area. Indicated as background area;

[0022] Regional adaptive correction of the illumination component is obtained ,in For the corrected illumination components, This represents the average value of the illumination component within the pressure vessel area. For dynamic adjustment coefficients;

[0023] The reconstructed image after illumination correction yields the corrected image. ;

[0024] YOLOv5 is used to identify scratch areas in the corrected image and semantic segmentation is performed to calculate the scratch area. The scratch area is compared with the surface area of ​​the pressure vessel and normalized to obtain a scratch feature quantification index. At the same time, semantic segmentation is performed to select n pressure vessel contour coordinate points in the corrected image. The Euclidean distance of each corresponding point is calculated with the corresponding coordinates of the standard template, the average distance value is calculated, and then normalized to obtain the shape integrity index.

[0025] Preferably, the specific implementation of determining the external safety quality score of the pressure vessel based on the shape integrity index and the scratch feature quantification index includes:

[0026] The external feature coefficient is obtained by weighted summation and fusion of the shape integrity index and the scratch feature quantification index. ,in, These are the shape integrity index and the scratch feature quantification index, respectively. The corresponding weighting coefficients are used; the external safety quality score is obtained through an exponential function. .

[0027] As a preferred embodiment, based on the pressure vessel thickness and crack parameter data in the equipment's basic data, a preset initial safety assessment threshold is dynamically corrected to obtain the dynamic safety assessment threshold. Specific implementation methods include:

[0028] First, obtain the pressure vessel thickness and crack parameter data of the equipment, and then use ultrasonic technology to identify the pressure vessel thickness, crack length, crack depth, and crack opening width.

[0029] Calculate the thickness coefficient of a pressure vessel ,in The actual thickness of the pressure vessel. These represent the maximum and minimum allowable thicknesses for the pressure vessel, respectively. Simultaneously, for crack length, crack depth, and crack opening width, normalized values ​​are calculated based on their maximum allowable values ​​according to safety standards. Weighted summations are then performed using empirically assigned weights to obtain the crack parameter coefficients. ;

[0030] The preset initial safety assessment threshold is dynamically corrected by using the pressure vessel thickness coefficient and crack parameter coefficient, resulting in the corrected dynamic safety assessment threshold. ,in These are the weighting coefficients. , which is the preset dynamic security assessment threshold These are the pressure vessel thickness coefficient and the crack parameter coefficient, respectively.

[0031] Compared with existing technologies, the advantages and positive effects of this invention are as follows: it overcomes the shortcomings of single-parameter monitoring and manual inspection, integrating internal temperature and pressure data, external image data, and equipment age and fatigue parameters to comprehensively cover the equipment's operating status and aging characteristics, avoiding missed diagnoses due to single-parameter information. Internal assessment deepens time-series analysis, reconstructing temperature and pressure curves through nonlinear fitting, and quantifying the temperature-pressure coupling relationship using multi-order derivatives and synergistic factors to accurately characterize the thermodynamic response law, which is more suitable for complex operating conditions than single-parameter judgment. External assessment achieves intelligentization, employing RGB and polarization image registration, MobileNetV2 semantic segmentation, and YOLOV5 scratch recognition to automatically quantify the scratch proportion and shape integrity, replacing manual inspection and improving real-time performance and accuracy. An innovative dynamic threshold mechanism corrects the initial threshold based on the pressure vessel thickness and crack parameters, adapting to different safety standards throughout the equipment's life cycle and avoiding misjudgments based on fixed thresholds. Attached Figure Description

[0032] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a schematic diagram of the structural process of a pressure vessel safety assessment method based on big data and multi-parameter coupling. Detailed Implementation

[0034] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described below in conjunction with the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0035] Numerous specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways than those described herein, and therefore the invention is not limited to the specific embodiments disclosed in the following specification.

[0036] In practice, existing pressure vessel safety assessment methods mainly include periodic manual inspections and single-parameter monitoring. Manual inspections rely on periodic shutdowns and personnel experience, making real-time safety status awareness impossible. Single-parameter monitoring focuses only on local indicators such as temperature or pressure, resulting in a one-sided assessment that fails to reflect the combined effects of material fatigue accumulation and external structural damage. Furthermore, existing assessment methods typically rely on fixed safety thresholds, neglecting the dynamic evolution of pressure vessel thickness and crack parameters. This leads to overestimation of the safety of older equipment and underestimation of the reliability of newer equipment, thus reducing the accuracy and timeliness of the assessment.

[0037] To overcome the aforementioned problems, a pressure vessel safety assessment method based on big data and multi-parameter coupling is proposed. By integrating internal operational data, external structural images, and historical equipment information, a multi-source information comprehensive assessment model is established to achieve intelligent, dynamic, and interpretable judgment of the pressure vessel's safety status. Through multi-dimensional analysis of temperature and pressure temporal variation patterns, external morphological integrity, scratch damage severity, and material fatigue parameters, a quantitative assessment of equipment health can be achieved, providing a theoretical basis and technical support for proactive safety management of industrial equipment. The specific implementation process is as follows: Figure 1 As shown.

[0038] Firstly, to achieve comprehensive awareness of the pressure vessel's safety status, multi-dimensional monitoring data is acquired. This multi-dimensional monitoring data includes internal monitoring data, external monitoring data, and basic equipment data. Internal monitoring data includes internal temperature and pressure data; external monitoring data includes image data of the external pressure vessel; and basic equipment data includes pressure vessel thickness and crack parameter data. To ensure data accuracy and synchronization, a distributed acquisition system is employed to align timestamps from different sensor sources and fuse multi-channel data. This ensures that internal physical quantities and external images correspond to the same operating state at the same time, providing reliable data support for subsequent comprehensive evaluation.

[0039] To quantify the safety of the internal operating state of a pressure vessel, in-depth analysis of temperature and pressure time-series data is performed. Based on internal temperature and pressure data at all acquisition times, the internal safety quality score of the pressure vessel is determined through time-series data analysis and data standardization scoring. First, a nonlinear curve fitting algorithm is used to smoothly reconstruct discrete sampling points, obtaining temperature and pressure curves in the continuous time dimension to remove sensor noise and restore the true trend of operating condition changes. Subsequently, within each operating condition cycle, the time period between a peak and an adjacent trough of the temperature curve is extracted, defined as a complete operating condition fluctuation cycle, and the pressure curve is simultaneously extracted within this cycle. Based on this, the rate of change of temperature and pressure, i.e., the first-order rate of change of temperature or pressure over time, is calculated to characterize the speed of their rise or fall; simultaneously, the second-order rate of change is calculated to reflect the drasticness of the change. When the rate or drasticness exceeds the preset safety range, it indicates that the equipment may have abnormal load or overheating risk. The system standardizes the rate deviation and fluctuation deviation of each cycle to eliminate scale differences between different operating conditions. Then, a co-factor between temperature and pressure is defined to describe the consistency of their changing trends. When temperature and pressure increases are highly synchronized, it indicates a high degree of coupling between container pressure and thermal stress. If this synergy exceeds a set threshold, it reflects a potential dangerous trend. A comprehensive synergy deviation coefficient is obtained by weighting and fusing rate synergy factors and fluctuation synergy factors. Finally, an exponential scoring function is used to calculate the internal safety quality score based on this coefficient. Specifically, nonlinear fitting algorithms are used to reconstruct curves from the collected time-series temperature and pressure data sequences, resulting in continuous-time temperature fitting curves. and pressure fitting curve Where t is the time index; extract a peak and a trough from the temperature time-series fitting curve, and denote the time interval from the acquisition time of the corresponding trough to the acquisition time of the adjacent peak as a complete operating condition fluctuation period of the pressure vessel; simultaneously acquire the pressure fitting curve of the same complete operating condition fluctuation period; obtain the temperature change rate curve by taking the first derivative of the temperature fitting curve of a complete operating condition fluctuation period. The curve of the degree of temperature fluctuation is obtained by taking the second derivative. The pressure change rate curve was obtained in the same way. And the curve of the intensity of pressure fluctuation ; Calculate the temperature rate deviation ,in, These represent the start and end times of a complete operating condition fluctuation segment. The maximum and minimum values ​​of the pre-set safety threshold for the rate of temperature change are determined; the temperature fluctuation deviation is also calculated. ,in The maximum and minimum values ​​of the pre-defined temperature fluctuation deviation threshold are used; and the pressure rate deviation is obtained using the same logic. Deviation from pressure fluctuation Define the rate co-factor ,in It is an extremely small constant; the volatility co-factor is defined in the same way. Then, the rate coordination factor and the fluctuation coordination factor are fused by weighted summation to obtain the comprehensive coordination deviation coefficient C; finally, the internal safety quality score is obtained by exponential function. Traditional pressure vessel safety assessments in this step typically rely solely on a single parameter, temperature or pressure, failing to reflect the dynamic coupling relationship between the two, leading to biased and delayed safety status judgments. To address this issue, this invention employs nonlinear reconstruction and multi-derivative analysis of temperature and pressure time-series data. This not only captures the rate characteristics of operating condition changes but also quantifies the severity of those changes, thereby accurately characterizing the thermodynamic response behavior of the equipment under different load conditions. Simultaneously, a cooperative deviation coefficient is introduced to measure the cooperative volatility between temperature and pressure, enabling the assessment results to reflect the true intensity of coupled stress. Through an exponential scoring mechanism, complex time-series characteristics are transformed into interpretable safety quality scores, achieving an intelligent shift from single-point anomaly detection to multi-parameter comprehensive judgment. This scheme significantly improves the accuracy, sensitivity, and traceability of internal pressure vessel safety assessments, providing a reliable quantitative basis for overall safety assessments.

[0040] To achieve a quantitative assessment of the external structural condition of pressure vessels, an appearance safety assessment method based on image intelligent recognition is adopted. Image data from external monitoring data undergoes preprocessing and feature extraction to obtain shape integrity and scratch feature quantification indices for the pressure vessel. Based on these indices, the external safety quality score of the pressure vessel is determined. First, visible light and polarized light images of the pressure vessel are simultaneously acquired. A feature point matching algorithm is used to calculate the spatial transformation relationship between the two images, accurately registering the polarized image to the visible light image coordinate system to ensure spatial consistency in subsequent feature extraction. Then, the image undergoes illumination decomposition, dividing it into illumination and reflection components. The illumination component primarily reflects the light intensity distribution, while the reflection component reflects surface material properties. For the illumination component, a region adaptive correction algorithm dynamically balances local brightness, eliminating shadows and reflection interference caused by ambient light differences, thus obtaining a corrected illumination image. In the image segmentation stage, a lightweight deep neural network model is used for semantic segmentation of the image, extracting the pressure vessel body region mask and eliminating background interference. For the corrected image, a target detection algorithm is used to automatically identify surface scratches, cracks, and corrosion areas, and the scratch area ratio is calculated. This ratio is normalized to obtain a scratch feature quantification index. Simultaneously, based on the segmented container outline, multiple feature points are selected and their corresponding points in the standard model are compared to calculate the average Euclidean distance, yielding a shape integrity index. These two indices are weighted and fused to form external feature coefficients, which are then converted into an external safety quality score using an exponential decay function. The lower the score, the more severe the external structural deformation or damage. Specifically, RGB images of the pressure vessel are acquired simultaneously. and polarization image The spatial transformation matrix is ​​calculated based on the SIFT feature point matching algorithm, and the polarization degree image is registered to the RGB image coordinate system to obtain the registered polarization degree map. ;right The image is decomposed into illumination and reflection components. The light component is ,in, Gaussian kernels of 3 scales, It is a very small constant. The polarization suppression coefficient is used; the MobileNetV2 network is employed for... Semantic segmentation is performed to obtain the pressure vessel region mask. ,in This is indicated as the pressure vessel area. Represented as the background region; region-adaptive correction is performed on the illumination components to obtain... ,in For the corrected illumination components, This represents the average value of the illumination component within the pressure vessel area. To dynamically adjust the coefficients; reconstruct the illumination-corrected image to obtain the corrected image. For the corrected image, YOLOv5 is used to identify scratch regions and perform semantic segmentation to calculate the scratch area. This area is then compared to the surface area of ​​the pressure vessel and normalized to obtain a scratch feature quantification index. Simultaneously, semantic segmentation is performed to select n pressure vessel contour coordinate points in the corrected image. The Euclidean distance between each corresponding point and the coordinates of the corresponding points on the standard template is calculated, and the average distance value is calculated and normalized to obtain a shape integrity index. Finally, the shape integrity index and the scratch feature quantification index are fused by weighted summation to obtain the external feature coefficient. ,in, These are the shape integrity index and the scratch feature quantification index, respectively. The corresponding weighting coefficients are used; the external safety quality score is obtained through an exponential function. .

[0041] The MobileNetV2 network mentioned above Semantic segmentation is performed to obtain the pressure vessel region mask. The specific implementation involves inputting the acquired color image of the pressure vessel into the MobileNetV2 network. The network, using a depthwise separable convolutional structure, effectively extracts multi-scale spatial features while maintaining low computational complexity. The front layers primarily capture low-level texture and edge information, the middle layers extract structural contours and geometric features, and the high layers extract semantic differences between the container body and the background. After feature extraction, the network uses a deconvolutional upsampling module to restore the high-dimensional feature mapping to the same spatial resolution as the input image, thereby generating a semantic segmentation result map. Each pixel in this result map is labeled as either a "container region" or a "non-container region" according to its semantic category, forming a pressure vessel region mask. To further improve segmentation accuracy, a class-balanced loss function is introduced during the training phase, assigning higher weights to container edge regions to reduce recognition errors caused by boundary blurring. Subsequently, morphological post-processing, including erosion and dilation operations, is performed on the mask obtained from semantic segmentation to remove isolated noise points and smooth region boundaries, ensuring the integrity and continuity of the container region. Ultimately, the position of each pixel in the resulting mask image corresponds one-to-one with the coordinates (x, y) of the original image, which can be used for subsequent scratch detection, contour analysis, and external safety quality score calculation. This method significantly improves the accuracy and stability of pressure vessel region identification while maintaining real-time performance, laying the foundation for multimodal data fusion evaluation.

[0042] The process involves using YOLOv5 to identify scratch regions in the corrected image and performing semantic segmentation to calculate the scratch area. This area is then compared to the surface area of ​​the pressure vessel and normalized to obtain a scratch feature quantification index. Simultaneously, semantic segmentation selects n pressure vessel contour coordinate points in the corrected image, calculates the Euclidean distance between each corresponding point and the corresponding point coordinates of the standard template, calculates the average distance value, and normalizes it to obtain a shape integrity index. Specifically, the process begins by inputting the illumination-corrected image into the YOLOv5 model for target detection and semantic segmentation. During the pre-training phase, the model uses an industrial defect dataset containing multiple types of scratch samples for transfer learning, capturing texture differences, edge breakage features, and grayscale distribution changes in the scratch region through a convolutional feature extraction layer. In the detection phase, the model generates multiple candidate region boxes and filters out the most likely scratch target regions based on a confidence threshold. Subsequently, the MobileNetV2 network is used to perform pixel-level classification of the detected regions, labeling each pixel as either "scratch" or "non-scratch," resulting in a high-resolution scratch mask image. In the scratch quantization stage, the system counts the total number of pixels marked as "scratches" in the mask image and, combined with the spatial calibration parameters of the camera system, converts the pixel area into the actual physical area. Then, using the total surface area of ​​the pressure vessel as a benchmark, the scratch area ratio is calculated and normalized to obtain a dimensionless scratch feature quantification index. This index numerically reflects the degree of surface damage; a larger value indicates a wider scratch coverage area and a higher potential threat to safety. To eliminate local deviations caused by differences in lighting and shooting angle, the system performs a weighted average of multiple viewpoint images before calculation to ensure stable and reliable quantization results. Simultaneously, to assess the integrity and geometric deformation of the overall shape of the pressure vessel, this embodiment selects n contour feature points as shape description nodes based on the segmented container contour. Specifically, the coordinate sequence of the container's outer boundary is obtained using the Canny edge detection and contour extraction algorithm; then, feature points are uniformly selected across the entire contour according to an equal-angle sampling strategy to ensure consistent point correspondence between different samples. Subsequently, the extracted feature points are matched one-to-one with corresponding points in a pre-established standard template model, and the Euclidean distance between each pair of corresponding points is calculated. This distance reflects the degree of deviation between the actual contour and the ideal contour; a larger distance indicates more significant local deformation. The average Euclidean distance of all corresponding points is calculated and normalized to obtain the shape integrity index. To improve the robustness of the evaluation results, a multi-angle matching correction strategy is further introduced: when there are deviations in the shape integrity index under different camera views, the system performs weighted fusion based on the viewpoint confidence weights to obtain the final comprehensive shape integrity value. Through this multi-parameter, multi-angle calculation method, a unified quantification of external surface defects and overall structural deformation of the pressure vessel is achieved, providing a reliable data foundation for calculating the external safety quality score.

[0043] To achieve a comprehensive assessment of both internal condition and external structure, the internal and external safety quality scores are weighted and summed to obtain the pressure vessel's safety quality score. The weighting coefficients are determined through training and optimization based on the contributions of internal anomalies and external damage to failure events in historical samples, ensuring the representativeness of the weighted result. The weighted safety quality score reflects the overall safety status of the pressure vessel within a specific time period and is the core quantitative indicator for dynamic assessment.

[0044] To overcome the assessment lag and distortion problems caused by fixed thresholds in traditional safety assessment methods, a dynamic safety assessment threshold is obtained by dynamically correcting the preset initial safety assessment threshold based on the pressure vessel thickness and crack parameter data in the equipment's basic data. Specifically, ultrasonic technology is first used to acquire the pressure vessel's thickness and crack parameter data. The thickness data reflects the degree of wall thickness attenuation under long-term operating conditions and is a key indicator for measuring corrosion and wear. The crack parameter data, obtained using ultrasonic technology, includes crack length, crack depth, and crack opening width, used to describe the cumulative characteristics of structural damage. The pressure vessel thickness coefficient is then calculated. ,in The actual thickness of the pressure vessel. These represent the maximum and minimum allowable thicknesses for the pressure vessel, respectively. Simultaneously, for crack length, crack depth, and crack opening width, normalized values ​​are calculated based on their maximum allowable values ​​according to safety standards. Weighted summations are then performed using empirically assigned weights to obtain the crack parameter coefficients. The preset initial safety assessment threshold is dynamically corrected using the pressure vessel thickness coefficient and crack parameter coefficient to obtain the corrected dynamic safety assessment threshold. ,in These are the weighting coefficients. , which is the preset dynamic security assessment threshold These are the pressure vessel thickness coefficient and the crack parameter coefficient, respectively. Through the aforementioned dynamic correction mechanism, the problem that fixed thresholds cannot reflect the evolution of local damage in equipment is solved. This allows the safety assessment model to be updated in real time as the equipment deteriorates, achieving an adaptive assessment logic where the more severe the damage, the stricter the judgment; and the healthier the condition, the more lenient the threshold. Compared with traditional static threshold methods, this significantly improves the real-time performance, sensitivity, and reliability of safety assessments, providing a scientific and quantifiable dynamic basis for the safety monitoring of pressure vessels throughout their entire lifecycle.

[0045] Finally, the safety quality score is compared with the dynamic safety assessment threshold. If the safety quality score is lower than the dynamic safety assessment threshold, the pressure vessel is determined to be in an unsafe state; if the safety quality score is greater than or equal to the dynamic safety assessment threshold, the pressure vessel is determined to be in a safe state. This method can provide personalized safety assessment results for pressure vessels at different stages of use and with different material properties, providing reliable technical support for safe production in industries such as chemical, energy, and metallurgy.

[0046] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments for application in other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A pressure vessel safety assessment method based on big data and multi-parameter coupling, characterized in that, Includes the following steps: Acquire multi-dimensional monitoring data of the pressure vessel, including internal monitoring data, external monitoring data, and basic equipment data; The internal monitoring data includes internal temperature and pressure data; the external monitoring data includes image data of the external pressure vessel; and the equipment foundation data includes pressure vessel thickness and crack parameter data. Based on the internal temperature and internal pressure data at all acquisition times, the internal safety quality score of the pressure vessel is determined through time-series data analysis and data standardization scores. The image data in the external monitoring data is preprocessed and features are extracted to obtain the shape integrity index and scratch feature quantification index of the pressure vessel. Based on the shape integrity index and scratch feature quantification index, the external safety quality score of the pressure vessel is determined. The safety quality score of the pressure vessel is obtained by weighted summation of the internal safety quality score and the external safety quality score. Based on the pressure vessel thickness and crack parameter data in the equipment's basic data, the preset initial safety assessment threshold is dynamically corrected to obtain the dynamic safety assessment threshold. The safety quality score is compared with the dynamic safety assessment threshold. If the safety quality score is lower than the dynamic safety assessment threshold, the pressure vessel is determined to be in an unsafe state; if the safety quality score is greater than or equal to the dynamic safety assessment threshold, the pressure vessel is determined to be in a safe state. The specific implementation method for dynamically correcting the preset initial safety assessment threshold based on the pressure vessel thickness and crack parameter data in the equipment's basic data to obtain the dynamic safety assessment threshold includes: First, obtain the pressure vessel thickness and crack parameter data of the equipment, and then use ultrasonic technology to identify the pressure vessel thickness, crack length, crack depth, and crack opening width. Calculate the thickness coefficient of a pressure vessel ,in The actual thickness of the pressure vessel. These represent the maximum and minimum allowable thicknesses for the pressure vessel, respectively. Simultaneously, for crack length, crack depth, and crack opening width, normalized values ​​are calculated based on their maximum allowable values ​​according to safety standards. Weighted summations are then performed using empirically assigned weights to obtain the crack parameter coefficients. ; The preset initial safety assessment threshold is dynamically corrected by using the pressure vessel thickness coefficient and crack parameter coefficient, resulting in the corrected dynamic safety assessment threshold. ,in These are the weighting coefficients. The preset dynamic security assessment threshold, These are the pressure vessel thickness coefficient and the crack parameter coefficient, respectively.

2. The pressure vessel safety assessment method based on big data and multi-parameter coupling according to claim 1, characterized in that, The specific implementation of determining the internal safety quality score of a pressure vessel based on internal temperature and pressure data at all acquisition times, through time-series data analysis and combined with data standardization scores, includes: The collected time-series temperature and pressure data sequences were reconstructed using nonlinear fitting algorithms to obtain temperature fitting curves in the continuous time dimension. and pressure fitting curve , where t is the time index; Extract a peak and a trough from the temperature fitting curve, and record the time period from the acquisition time of the corresponding trough to the acquisition time of the adjacent peak as a complete operating condition fluctuation period of the pressure vessel; at the same time, collect the pressure fitting curve of the same complete operating condition fluctuation period. The temperature change rate curve is obtained by taking the first derivative of the temperature fitting curve for a complete operating condition fluctuation period. The curve of the degree of temperature fluctuation is obtained by taking the second derivative. The pressure change rate curve was obtained in the same way. And the curve of the intensity of pressure fluctuation ; Calculate temperature rate deviation ,in, These represent the start and end times of a complete operating condition fluctuation segment. The maximum and minimum values ​​of the pre-set safety threshold for the rate of temperature change are determined; the temperature fluctuation deviation is also calculated. ,in The maximum and minimum values ​​of the pre-defined temperature fluctuation deviation threshold are used; and the pressure rate deviation is obtained using the same logic. Deviation from pressure fluctuation ; Define rate co-factor ,in It is an extremely small constant; the volatility co-factor is defined in the same way. Then, the rate synergy factor and the fluctuation synergy factor are fused by weighted summation to obtain the comprehensive synergy deviation coefficient C; The internal safety quality score is obtained by using an exponential function. .

3. The pressure vessel safety assessment method based on big data and multi-parameter coupling according to claim 1, characterized in that, The specific implementation of preprocessing and feature extraction of image data from external monitoring data to obtain the shape integrity index and scratch feature quantification index of the pressure vessel is as follows: Synchronously acquire RGB images of the pressure vessel and polarization image The spatial transformation matrix is ​​calculated based on the SIFT feature point matching algorithm, and the polarization degree image is registered to the RGB image coordinate system to obtain the registered polarization degree map. ; right The image is decomposed into illumination and reflection components. The light component is ,in, Gaussian kernels of 3 scales, It is a very small constant. This is the polarization suppression coefficient; Using MobileNetV2 network Semantic segmentation is performed to obtain the pressure vessel region mask. ,in This is indicated as the pressure vessel area. Indicated as background area; Regional adaptive correction of the illumination component is obtained ,in For the corrected illumination components, This represents the average value of the illumination component within the pressure vessel area. For dynamic adjustment coefficients; The reconstructed image after illumination correction yields the corrected image. ; YOLOv5 is used to identify scratch areas in the corrected image and semantic segmentation is performed to calculate the scratch area. The scratch area is compared with the surface area of ​​the pressure vessel and normalized to obtain a scratch feature quantification index. At the same time, semantic segmentation is performed to select n pressure vessel contour coordinate points in the corrected image. The Euclidean distance of each corresponding point is calculated with the corresponding coordinates of the standard template, the average distance value is calculated, and then normalized to obtain the shape integrity index.

4. The pressure vessel safety assessment method based on big data and multi-parameter coupling according to claim 1, characterized in that, The specific implementation of determining the external safety quality score of a pressure vessel based on the aforementioned shape integrity index and scratch feature quantification index includes: The external feature coefficient is obtained by weighted summation and fusion of the shape integrity index and the scratch feature quantification index. ,in, These are the shape integrity index and the scratch feature quantification index, respectively. The corresponding weighting coefficients are used; the external safety quality score is obtained through an exponential function. .

Images