Hydrogen energy ball valve on-line sealing detection equipment

Abstract

The application provides a hydrogen energy ball valve online sealing detection equipment and relates to the technical field of valve sealing detection. The application overcomes the disadvantages of traditional detection data processing redundancy and the inability to capture multi-factor coupling implicit failure through a progressive diagnosis framework of overall evaluation first and accurate tracing later. The quantitative tracing mechanism based on vector space projection greatly enhances the pertinence of maintenance decisions. Finally, through system-level risk visualization and intelligent path planning, the application realizes the leap from single-point state monitoring to full-station safety operation and maintenance decision support, and comprehensively improves the safety, reliability and maintenance efficiency of hydrogen energy facility operation.

Description

Technical Field

[0001] This invention belongs to the field of sealing detection technology, specifically relating to an online sealing detection device for hydrogen energy ball valves. Background Technology

[0002] Hydrogen energy, as a clean and efficient energy carrier, plays an important role in the energy transition. As a key fluid control component in hydrogen energy systems, the sealing performance of ball valves directly affects the safety, reliability, and operating efficiency of the system. The main sealing failure modes of hydrogen ball valves include internal leakage, external leakage, and hydrogen embrittlement. Once a hydrogen ball valve experiences internal leakage, external leakage, or hydrogen-induced damage (hydrogen embrittlement), it will not only lead to hydrogen leakage and energy loss, but may also cause serious safety accidents such as combustion and explosion. The risk of sealing failure is even more prominent in high-pressure, high-purity hydrogen environments. Therefore, real-time and accurate online sealing detection of hydrogen ball valves is extremely important.

[0003] Currently, traditional hydrogen energy ball valve sealing testing mainly relies on periodic offline disassembly and inspection, manual inspection, or single sensor monitoring, which has the following obvious shortcomings:

[0004] (1) Existing sealing testing equipment focuses on monitoring single parameters such as pressure or flow rate, which is difficult to fully reflect multiple sealing failure modes such as internal leakage, external leakage and hydrogen embrittlement. In particular, micro-damage such as hydrogen embrittlement is difficult to be captured by conventional means in the early stage. At the same time, it lacks the fusion analysis and state quantification assessment of multi-source data. It makes decisions based on single collected data, and the diagnostic results rely on human experience, which is prone to missed reports and false reports.

[0005] (2) Existing sealing detection equipment usually has a judgment unit that detects external leakage and internal leakage in parallel. Each unit needs to process the raw data independently and compare it with the threshold, which will lead to problems such as data processing redundancy, waste of computing resources, and system response delay. At the same time, the output results of each judgment unit are isolated from each other and lack coordination, which cannot capture such hidden failures coupled by multiple factors, resulting in delayed early warning.

[0006] (3) Even if the existing sealing detection equipment detects an abnormality, it is difficult to distinguish whether it is caused by internal leakage, external leakage or hydrogen embrittlement, resulting in a lack of targeted maintenance decisions;

[0007] (4) Existing sealing detection equipment is usually limited to single-point monitoring and cannot integrate the health status of multiple valves into a system-level risk distribution map, which is not conducive to maintenance personnel quickly locating high-risk areas and planning safe paths. Summary of the Invention

[0008] This invention provides an online sealing testing device for hydrogen-powered ball valves to solve at least one of the technical problems mentioned above.

[0009] To solve the above-mentioned technical problems, this invention discloses an online sealing detection device for hydrogen-powered ball valves, comprising:

[0010] The multi-source sealing data acquisition module is used to acquire the original multi-physics field data of each hydrogen ball valve in real time, and generate the internal leakage state vector, external leakage state vector and hydrogen embrittlement state vector of each hydrogen ball valve in the current monitoring cycle based on the original multi-physics field data.

[0011] The sealing status diagnosis and analysis module is used to determine the sealing health status of each hydrogen ball valve in the current monitoring cycle based on the internal leakage status vector, external leakage status vector, hydrogen embrittlement status vector, and the preset three-dimensional feature space of sealing health of each ball valve in the current monitoring cycle.

[0012] The sealing failure cause analysis module is used to trace and analyze the causes of sealing failures of hydrogen ball valves with abnormal sealing health status, and output a complete sealing failure contribution spectrum based on the traceability analysis results.

[0013] The comprehensive hazard assessment module is used to render the distribution map of hydrogen ball valves based on the sealing health status diagnosis results of each hydrogen ball valve in the current monitoring cycle, generate a dynamic hazard heat map, and generate the optimal maintenance and hazard avoidance path based on the hazard heat map.

[0014] Preferably, the multi-source sealing data acquisition module includes:

[0015] The internal leakage sensing and detection submodule includes two sets of high-precision micro differential pressure sensors installed in the upstream and downstream pipelines of the hydrogen ball valve, and a temperature sensor installed in the downstream pipeline of the hydrogen ball valve. The two sets of high-precision micro differential pressure sensors are used to collect pressure data of the upstream and downstream pipelines of the hydrogen ball valve when the valve core is closed, respectively, and the temperature sensor is used to collect temperature data of the downstream pipeline of the hydrogen ball valve.

[0016] The external leakage sensing and detection submodule includes an ultrasonic sensor and a hydrogen concentration sensor installed on the outer wall of the hydrogen ball valve stem packing assembly. The ultrasonic sensor is used to collect the sound pressure data of the hydrogen turbulent jet leaking from the hydrogen ball valve stem packing assembly, and the hydrogen concentration sensor is used to collect the hydrogen accumulation concentration data of the current environment where the hydrogen ball valve is located.

[0017] The hydrogen embrittlement sensing and detection submodule includes a high-frequency acoustic emission sensor and a triaxial vibration acceleration sensor installed on the body of the hydrogen ball valve. The high-frequency acoustic emission sensor is used to collect the voltage data detected by the high-frequency acoustic emission sensor when the valve core and valve seat of the hydrogen ball valve are in a coordinated motion state. The triaxial vibration acceleration sensor is used to collect the vibration data of the valve core and valve seat of the hydrogen ball valve in a coordinated motion state.

[0018] The internal leakage state vector generation submodule is used to generate the internal leakage state vector of the hydrogen ball valve for the current monitoring period based on the pressure data of the upstream and downstream pipelines of the hydrogen ball valve and the temperature data of the downstream pipeline of the hydrogen ball valve when the valve core is closed in the current monitoring period.

[0019] The external leakage state vector generation submodule is used to generate the external leakage state vector of the hydrogen ball valve for the current monitoring period based on the acoustic pressure data of the hydrogen turbulent jet collected by the ultrasonic sensor and the hydrogen accumulation concentration data collected by the hydrogen concentration sensor.

[0020] The hydrogen embrittlement state vector generation submodule is used to generate the hydrogen embrittlement state vector of the hydrogen ball valve in the current monitoring period based on the voltage data detected by the high-frequency acoustic emission sensor and the vibration data collected by the triaxial vibration acceleration sensor.

[0021] Preferably, the internal leakage state vector generation submodule includes:

[0022] A type of internal leakage vector element acquisition unit is used to calculate the pressure difference between the upstream and downstream pipelines of the hydrogen ball valve based on the detection values ​​of two sets of high-precision micro differential pressure sensors at each sampling time within the monitoring period. The pressure differences at several sampling times are arranged in time sequence to obtain the pressure difference sequence for the corresponding monitoring period. Linear fitting and quadratic polynomial fitting are performed on the pressure difference sequence for each monitoring period. The slope of the straight line obtained by linear fitting is used as the pressure difference attenuation slope for the corresponding monitoring period, and twice the coefficient of the quadratic term of the quadratic polynomial fitting result is used as the pressure difference attenuation curvature for the corresponding monitoring period.

[0023] The second type of internal leakage vector element acquisition unit is used to analyze the temperature data of the downstream pipeline of the hydrogen energy ball valve collected by the temperature sensor through wavelet transform, extract the characteristic instantaneous temperature drop pulse within the monitoring period, take the maximum value of the characteristic instantaneous temperature drop pulse as the instantaneous temperature drop pulse amplitude within the corresponding monitoring period, and take the number of characteristic instantaneous temperature drop pulses as the instantaneous temperature drop pulse frequency.

[0024] The internal leakage state vector generation unit arranges the differential pressure attenuation slope, differential pressure attenuation curvature, instantaneous temperature drop pulse amplitude, and instantaneous temperature drop pulse frequency of each monitoring cycle in order to form the internal leakage state vector.

[0025] Preferably, the external state vector generation submodule includes:

[0026] A type of external leakage vector element acquisition unit is used to perform fast Fourier transform on the sound pressure signal collected by the ultrasonic sensor at each sampling time in each monitoring cycle to obtain the average power spectral density of the corresponding monitoring cycle. Based on the average power spectral density of the corresponding monitoring cycle, the cumulative energy of the preset hydrogen turbulent jet characteristic frequency band of the corresponding monitoring cycle is obtained and used as the cumulative energy of the turbulent jet of the corresponding monitoring cycle. The standard deviation of the preset energy distribution of the hydrogen turbulent jet characteristic frequency band is used as the turbulent jet bandwidth of the corresponding monitoring cycle.

[0027] The second type of leakage vector element acquisition unit is used to obtain the average rate of change of hydrogen concentration in the corresponding monitoring period based on the hydrogen accumulation concentration data collected by the hydrogen concentration sensor at each sampling time of each monitoring period.

[0028] The external leakage state vector generation unit is used to arrange the cumulative energy of the turbulent jet, the bandwidth of the turbulent jet, and the average rate of change of hydrogen concentration in each monitoring cycle in order to form an external leakage state vector.

[0029] Preferably, the hydrogen embrittlement state vector generation submodule includes:

[0030] A hydrogen embrittlement vector element acquisition unit is used to count the number of voltage signals detected by the high-frequency acoustic emission sensor at each sampling time in each monitoring cycle that exceed the preset voltage threshold, and use this as the total number of acoustic emission events. Based on the total number of acoustic emission events and the total number of sampling times in each monitoring cycle, the occurrence rate of acoustic emission events in each monitoring cycle is calculated. The mean value of the voltage signal corresponding to each acoustic emission event in the corresponding monitoring cycle is used as the acoustic emission event intensity evaluation value in the corresponding monitoring cycle.

[0031] The second-class hydrogen embrittlement vector element acquisition unit is used to calculate the root mean square value and kurtosis of the time-series composite acceleration amplitude sequence within each monitoring period based on the composite acceleration amplitude detected by the triaxial vibration accelerometer at each sampling time within each monitoring period, and obtain the vibration root mean square value and vibration kurtosis of the corresponding monitoring period.

[0032] The hydrogen embrittlement state vector generation unit is used to arrange the acoustic emission event occurrence rate, acoustic emission event intensity assessment value, vibration root mean square value and vibration kurtosis in order for each monitoring period to form a hydrogen embrittlement state vector.

[0033] Preferably, the sealing condition diagnostic analysis module includes:

[0034] The feature space construction submodule is used to construct the ball valve preset sealing health three-dimensional feature space based on the internal leakage state vector, external leakage state vector and hydrogen embrittlement state vector corresponding to all historical normal sealing monitoring cycles.

[0035] The state coordinate point determination submodule is used to project the internal leakage state vector, external leakage state vector, and hydrogen embrittlement state vector of each hydrogen ball valve in the current monitoring cycle onto the internal leakage principal component vector, external leakage principal component vector, and hydrogen embrittlement principal component vector, respectively, to obtain the internal leakage scalar value x, external leakage scalar value y, and hydrogen embrittlement scalar value z, and use (x, y, z) as the state coordinate point P of the current monitoring cycle;

[0036] The health deviation calculation submodule is used to calculate the Mahalanobis distance between the state coordinate point P of the current monitoring period and the cluster Q of state coordinate points of all historical normal monitoring periods. And use it as the seal health deviation for the current monitoring cycle;

[0037] The status assessment result output submodule is used to compare the seal health deviation of the current monitoring period with the preset seal health deviation threshold range and output the corresponding seal health status level, which includes healthy, alert, warning and abnormal.

[0038] Preferably, the feature space construction submodule includes:

[0039] The analysis matrix determination unit is used to construct analysis matrix one, analysis matrix two and analysis matrix three based on the internal leakage state vector, external leakage state vector and hydrogen embrittlement state vector corresponding to all historical normal sealing monitoring cycles;

[0040] Principal component vector one determination unit is used to perform principal component analysis on analysis matrix one to obtain the principal component vector of analysis matrix one, and use it as principal component vector one;

[0041] The principal component vector two determination unit is used to perform principal component analysis on analysis matrix two to obtain the principal component vector of analysis matrix two, and use it as the principal component vector two.

[0042] The principal component vector three determination unit is used to perform principal component analysis on the analysis matrix three to obtain the principal component vector of the analysis matrix three, and use it as the principal component vector three.

[0043] The feature space construction unit is used to orthogonalize the principal component vectors 1, 2, and 3 to form the internal leakage principal component vector, external leakage principal component vector, and hydrogen embrittlement principal component vector. Based on the internal leakage principal component vector, external leakage principal component vector, and hydrogen embrittlement principal component vector, the ball valve preset sealing health three-dimensional feature space is constructed.

[0044] Preferably, the seal failure cause analysis module includes:

[0045] The elementary failure mode spectrum library construction submodule is used to construct three mutually orthogonal elementary failure mode vectors, including elementary vector one representing pure internal leakage of the sealing part, elementary vector two representing pure external leakage of the packing part, and elementary vector three representing hydrogen embrittlement of the pure sealing part. The space composed of elementary vector one, elementary vector two, and elementary vector three is used as the elementary space.

[0046] The deviation vector acquisition submodule takes the difference between the state coordinate point P corresponding to the monitoring cycle with an abnormal sealing health status level and the center point O of the cluster Q of state coordinate points from all historical normal monitoring cycles as the deviation vector for the current abnormal monitoring cycle. ;

[0047] The failure mode determination submodule will determine the deviation vector of the current anomaly monitoring cycle. Projecting onto the primitive space, calculate the deviation vector for the current anomaly monitoring period. Projection coefficients are plotted on primitive vector one, primitive vector two, and primitive vector three respectively, and each projection coefficient is normalized to obtain the contribution percentage of each primitive failure mode at the state coordinate point P corresponding to the current anomaly monitoring cycle. The dominant failure type is determined based on the primitive failure mode with the largest contribution percentage, and a complete sealing failure contribution spectrum is output.

[0048] Preferably, the comprehensive hazard assessment module includes:

[0049] The Hazardous Heat Map Construction Submodule is used to create a 3D layout map based on hydrogen energy facilities, mark the precise installation location of each hydrogen ball valve, construct an initial hydrogen ball valve distribution map, and divide the hydrogen ball valve distribution map into several consecutive assessment areas.

[0050] The regional risk value calculation submodule is used to calculate the comprehensive risk thermal value of each assessment area based on the sealing health status level of each hydrogen ball valve in the current monitoring cycle, and based on the real-time risk values ​​of all hydrogen ball valves corresponding to each assessment area.

[0051] The risk assessment visualization submodule is used to generate a dynamic hazard heat map by rendering it on the hydrogen energy ball valve distribution map with a color gradient from cool to warm colors based on the comprehensive risk heat value of each assessment area.

[0052] The risk avoidance path generation module is used to automatically generate and display the optimal maintenance risk avoidance path to an assessment area when the comprehensive risk heat value of an assessment area exceeds a preset high-risk threshold.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] This invention synchronously collects and integrates multi-physics field data on internal leakage, external leakage, and hydrogen embrittlement through a multi-source sealing data acquisition module, laying the foundation for comprehensive evaluation. Addressing the challenges of inefficient independent assessment of individual components and the inability to capture latent failures caused by multi-factor coupling, this invention uses a sealing status diagnosis and analysis module to process the corresponding state vectors of internal leakage, external leakage, and hydrogen embrittlement in one go, determining the sealing health status of each hydrogen-powered ball valve in the current monitoring cycle. This not only avoids data redundancy and delays from parallel processing by multiple modules, but its multi-variable joint evaluation characteristics are also more sensitive to capturing subtle collaborative anomalies between various features, enabling early warning. Addressing the pain points of untraceable anomalies and lack of targeted maintenance decisions, this invention sets up a sealing failure cause analysis module that only conducts in-depth analysis of states assessed as abnormal overall, and outputs a complete sealing failure contribution spectrum based on the traceability analysis results. Finally, addressing the lack of system-level risk visualization, this invention uses a hazard comprehensive assessment module to directly and efficiently generate dynamic risk heat maps and intelligent risk avoidance paths using standardized diagnostic results, achieving a leap from single-point diagnosis to full-site safety operation and maintenance decision-making. Attached Figure Description

[0055] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0056] Figure 1 This is a schematic diagram of the composition of the online sealing detection device for hydrogen-powered ball valves of the present invention;

[0057] Figure 2 This is a schematic diagram of the hydrogen energy ball valve structure of the present invention.

[0058] In the diagram: 1. Upstream pipe of the hydrogen ball valve; 2. Downstream pipe of the hydrogen ball valve; 3. Stem packing assembly of the hydrogen ball valve; 4. Valve core of the hydrogen ball valve; 5. Valve seat of the hydrogen ball valve. Detailed Implementation

[0059] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0060] Furthermore, in this invention, the use of terms such as "first" and "second" is for descriptive purposes only and does not specifically refer to any order or sequence, nor is it intended to limit the invention. They are merely used to distinguish components or operations described using the same technical terms and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions and features of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If a combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0061] The present invention provides the following embodiments:

[0062] Example 1

[0063] This invention provides an online sealing testing device for hydrogen-powered ball valves, such as... Figure 1-2 As shown, it includes:

[0064] The multi-source sealing data acquisition module is used to acquire the original multi-physics field data of each hydrogen ball valve in real time, and generate the internal leakage state vector, external leakage state vector and hydrogen embrittlement state vector of each hydrogen ball valve in the current monitoring cycle based on the original multi-physics field data.

[0065] The sealing status diagnosis and analysis module is used to determine the sealing health status of each hydrogen ball valve in the current monitoring cycle based on the internal leakage status vector, external leakage status vector, hydrogen embrittlement status vector, and the preset three-dimensional feature space of sealing health of each ball valve in the current monitoring cycle.

[0066] The sealing failure cause analysis module is used to trace and analyze the causes of sealing failures of hydrogen ball valves with abnormal sealing health status, and output a complete sealing failure contribution spectrum based on the traceability analysis results.

[0067] The comprehensive hazard assessment module is used to render the distribution map of hydrogen ball valves based on the sealing health status diagnosis results of each hydrogen ball valve in the current monitoring cycle, generate a dynamic hazard heat map, and generate the optimal maintenance and hazard avoidance path based on the hazard heat map.

[0068] In this embodiment, the raw multiphysics data includes internal leakage-related data, external leakage-related data, and hydrogen embrittlement-related data;

[0069] The internal leakage related data includes the pressure data of the upstream pipe 1 and the downstream pipe 2 of the hydrogen ball valve when the valve core 4 of the hydrogen ball valve is closed, and the temperature data of the downstream pipe 2 of the hydrogen ball valve.

[0070] The external leakage data includes the sound pressure data of the hydrogen turbulent jet leaking from the three points of the hydrogen ball valve stem packing assembly and the hydrogen accumulation concentration data of the current environment where the hydrogen ball valve is located.

[0071] The hydrogen embrittlement-related data includes voltage data detected by a high-frequency acoustic emission sensor of the valve core 4 and valve seat 5 of the hydrogen ball valve in their coordinated motion state, as well as vibration data of the valve core 4 and valve seat 5 of the hydrogen ball valve in their coordinated motion state.

[0072] In this embodiment, the preset sealing health three-dimensional feature space is a mathematical space used to quantitatively evaluate the sealing health status of the ball valve. This space consists of three mutually orthogonal coordinate axes, namely the internal leakage principal component vector axis, the external leakage principal component vector axis, and the hydrogen embrittlement principal component vector axis.

[0073] In this embodiment, the internal leakage state vector is a mathematical vector composed of a set of internal leakage feature elements, used to quantify and characterize the probability, severity and dynamic characteristics of internal leakage in the hydrogen ball valve sealing pair (the valve core 4 and the valve seat 5 of the hydrogen ball valve) during the current monitoring period.

[0074] The external leakage state vector is a mathematical vector composed of a set of external leakage characteristic elements, used to quantify and characterize the probability, leakage intensity and leakage flow characteristics of external leakage at the three points of the hydrogen ball valve stem packing assembly during the current monitoring period.

[0075] The hydrogen embrittlement state vector is a mathematical vector composed of a set of hydrogen embrittlement characteristic elements. It is used to quantify and characterize the possibility, degree of damage, and activity of microcracks or surface degradation caused by hydrogen-induced damage (hydrogen embrittlement) in the sealing pair of hydrogen-powered ball valves during the current monitoring period.

[0076] In this embodiment, based on the internal leakage state vector, external leakage state vector, hydrogen embrittlement state vector, and the preset three-dimensional feature space of the sealing health of each hydrogen ball valve in the current monitoring cycle, the sealing health status of each hydrogen ball valve in the current monitoring cycle is determined. Specifically, this includes: calculating the Mahalanobis distance between the state coordinate points corresponding to the internal leakage state vector, external leakage state vector, and hydrogen embrittlement state vector of the hydrogen ball valve in the current monitoring cycle and the cluster of state coordinate points of all historical normal monitoring cycles in the preset three-dimensional feature space of sealing health, thereby obtaining the sealing health deviation of the hydrogen ball valve in the current monitoring cycle.

[0077] In this embodiment, the sealing health status of the hydrogen ball valve includes healthy (the operating status of the hydrogen ball valve is within the historical normal fluctuation range, with no significant abnormal signs), alert (the status of the hydrogen ball valve begins to show identifiable deviations, but is still in the early or slight stage, requiring increased attention and observation), warning (the status of the hydrogen ball valve has significantly deviated from the normal range, the risk of failure has significantly increased, and preparations need to be made for maintenance intervention or in-depth inspection), and abnormal (the valve status is seriously deviated from the normal range, and it is determined to be in an abnormal state. This level will trigger the subsequent sealing failure cause analysis module to perform quantitative analysis of the cause of sealing failure and serve as input for risk calculation in the risk comprehensive assessment module).

[0078] In this embodiment, the seal failure contribution spectrum is a quantitative report generated after the cause analysis of the abnormal state, which is used to clarify the various failure modes that lead to the abnormality and their degree of impact.

[0079] In this embodiment, the hydrogen ball valve distribution map is a digital map constructed based on the actual layout of hydrogen energy facilities, reflecting the actual location of each hydrogen ball valve.

[0080] In this embodiment, the dynamic hazard heat map is a visualization interface generated by color rendering based on the hydrogen ball valve distribution map and the real-time assessment results. It is a color map that uses a continuous color gradient from cool to warm colors to represent the overall risk level of different areas. Its color will change dynamically as the assessment results of each monitoring cycle are updated.

[0081] In this embodiment, the optimal maintenance and risk avoidance path is the route that staff need to take to reach the abnormal hydrogen ball valve when they are inspecting it. The optimal maintenance and risk avoidance path will try to avoid other high-risk areas and choose a wide and safe passage to ensure that personnel can reach the target area safely.

[0082] The beneficial effects of the above technical solution are as follows: This invention synchronously collects and integrates multi-physics field data of internal leakage, external leakage, and hydrogen embrittlement through a multi-source sealing data acquisition module, laying the foundation for comprehensive evaluation. Addressing the problem of low efficiency and inability to capture hidden failures caused by independent judgment of individual items, this invention processes the corresponding state vectors of internal leakage, external leakage, and hydrogen embrittlement in one go through a sealing state diagnosis and analysis module, determining the sealing health status of each hydrogen ball valve in the current monitoring cycle. This not only avoids data redundancy and delays caused by parallel processing of multiple modules, but its multi-variable joint evaluation characteristics are also more sensitive to capturing subtle collaborative anomalies between various features, achieving early warning. Addressing the pain points of anomalies being difficult to trace and maintenance decisions lacking specificity, the sealing failure cause analysis module set up in this invention only performs in-depth analysis on states assessed as abnormal overall, and outputs a complete sealing failure contribution spectrum based on the traceability analysis results. Finally, addressing the lack of system-level risk visualization, this invention, through a comprehensive hazard assessment module, directly and efficiently generates dynamic risk heat maps and intelligent risk avoidance paths using standardized diagnostic results, achieving a leap from single-point diagnosis to full-site safety operation and maintenance decisions.

[0083] Example 2

[0084] Based on Example 1, the multi-source sealing data acquisition module includes:

[0085] The internal leakage sensing and detection submodule includes two sets of high-precision micro differential pressure sensors installed in the upstream pipeline 1 and the downstream pipeline 2 of the hydrogen ball valve, and a temperature sensor installed in the downstream pipeline 2 of the hydrogen ball valve. The two sets of high-precision micro differential pressure sensors are used to collect pressure data of the upstream pipeline 1 and the downstream pipeline 2 of the hydrogen ball valve when the valve core 4 of the hydrogen ball valve is closed, respectively. The temperature sensor is used to collect temperature data of the downstream pipeline 2 of the hydrogen ball valve.

[0086] The external leakage sensing and detection submodule includes an ultrasonic sensor and a hydrogen concentration sensor installed on the outer wall of the hydrogen ball valve stem packing assembly 3. The ultrasonic sensor is used to collect the sound pressure data of the hydrogen turbulent jet leaking from the hydrogen ball valve stem packing assembly 3, and the hydrogen concentration sensor is used to collect the hydrogen accumulation concentration data of the current hydrogen ball valve environment.

[0087] The hydrogen embrittlement sensing and detection submodule includes a high-frequency acoustic emission sensor and a triaxial vibration acceleration sensor installed on the body of the hydrogen ball valve. The high-frequency acoustic emission sensor is used to collect the voltage data detected by the high-frequency acoustic emission sensor when the valve core 4 and the valve seat 5 of the hydrogen ball valve are in a coordinated motion state. The triaxial vibration acceleration sensor is used to collect the vibration data of the valve core 4 and the valve seat 5 of the hydrogen ball valve in a coordinated motion state.

[0088] The internal leakage state vector generation submodule is used to generate the internal leakage state vector of the hydrogen ball valve in the current monitoring period based on the pressure data of the upstream pipeline 1 and the downstream pipeline 2 of the hydrogen ball valve and the temperature data of the downstream pipeline 2 of the hydrogen ball valve when the valve core 4 of the hydrogen ball valve is closed.

[0089] The external leakage state vector generation submodule is used to generate the external leakage state vector of the hydrogen ball valve for the current monitoring period based on the acoustic pressure data of the hydrogen turbulent jet collected by the ultrasonic sensor and the hydrogen accumulation concentration data collected by the hydrogen concentration sensor.

[0090] The hydrogen embrittlement state vector generation submodule is used to generate the hydrogen embrittlement state vector of the hydrogen ball valve in the current monitoring period based on the voltage data detected by the high-frequency acoustic emission sensor and the vibration data collected by the triaxial vibration acceleration sensor.

[0091] In this embodiment, the significance of the high-precision micro differential pressure sensor is as follows: When the valve core 4 of the hydrogen ball valve is closed, under ideal sealing conditions, the upstream pipe 1 and the downstream pipe 2 of the hydrogen ball valve maintain a constant pressure difference. If the pressure difference between the upstream pipe 1 and the downstream pipe 2 of the hydrogen ball valve continuously and monotonically decreases, it indicates that hydrogen flows from the high-pressure side to the low-pressure side through the micro-channel of the valve seat sealing pair. The slope of the pressure decay curve can be directly converted into the internal leakage equivalent leakage rate. The horizontal axis of the pressure decay curve is time, and the vertical axis is the pressure difference between the upstream pipe 1 and the downstream pipe 2 of the hydrogen ball valve.

[0092] In this embodiment, the significance of the temperature sensor is that when hydrogen passes through a tiny leak gap, it expands and absorbs heat instantly, resulting in a characteristic instantaneous temperature drop pulse downstream of the leak point. That is, the low temperature signal of the downstream pipe 2 of the hydrogen ball valve can be used as one of the bases for judging internal leakage.

[0093] In this embodiment, the significance of the ultrasonic sensor is that when hydrogen leaks through the valve stem packing assembly 3 of the hydrogen ball valve, a hydrogen turbulent jet is formed, which excites a sound pressure signal. This signal is essentially a gas dynamic noise signal, the energy intensity of which is related to the leakage rate and pressure, and the dominant frequency characteristics are related to the shape and size of the leak. Therefore, this sound signal can be used as one of the bases for judging external leakage.

[0094] In this embodiment, the significance of the hydrogen concentration sensor is that when hydrogen leaks through the valve stem packing assembly 3 of the hydrogen ball valve, the hydrogen concentration in the environment where the hydrogen ball valve is located will continuously accumulate, and thus the hydrogen concentration signal can be used as one of the bases for judging external leakage.

[0095] In this embodiment, the significance of the high-frequency acoustic emission sensor lies in the fact that the valve core 4 or the valve seat 5 of the hydrogen ball valve will generate microcracks during hydrogen embrittlement. During the engagement of the valve core 4 and the valve seat 5, the stress wave generated at the crack location will propagate to the high-frequency acoustic emission sensor. Since metal is an excellent acoustic conductor and the propagation path of high-frequency signals in the structure is clear, by analyzing the time difference of the signal reaching the high-frequency acoustic emission sensor, the acoustic emission source can be located, thereby determining whether the damage occurs at the valve core 4 or the valve seat 5 of the hydrogen ball valve. Thus, by capturing the sound pressure signal emitted during the engagement of the valve core 4 or the valve seat 5 of the hydrogen ball valve, the characteristics of the sound pressure signal can be analyzed to determine whether hydrogen embrittlement exists inside the material.

[0096] In this embodiment, the significance of the triaxial vibration acceleration sensor is that the valve core 4 or the valve seat 5 of the hydrogen ball valve will have uneven surfaces in the hydrogen embrittlement state, and will generate strong vibrations when they cooperate with each other. By capturing the vibration data through the triaxial vibration acceleration sensor, it can be used as one of the bases for judging hydrogen embrittlement.

[0097] The beneficial effects of the above technical solution are as follows: Unlike the isolated data approach in the prior art, the present invention is based on the physical mechanisms of internal leakage, external leakage, and hydrogen embrittlement. It precisely deploys high-precision micro-differential pressure sensors, temperature sensors, ultrasonic sensors, hydrogen concentration sensors, high-frequency acoustic emission sensors, and triaxial vibration acceleration sensors, forming a cross-validation and complementary sensing network of multiple physical quantities. Through the internal leakage, external leakage, and hydrogen embrittlement state vector generation submodule, the original data is extracted into highly generalized structured feature vectors in real time, which greatly reduces the data transmission and processing load and overcomes the drawbacks of the prior art of excessive original data and large processing delays.

[0098] Example 3

[0099] Based on Example 2, the internal leakage state vector generation submodule includes:

[0100] A type of internal leakage vector element acquisition unit is used to calculate the pressure difference between the upstream pipeline 1 and the downstream pipeline 2 of the hydrogen ball valve based on the detection values ​​of two sets of high-precision micro differential pressure sensors at each sampling time within the monitoring period. The pressure difference at several sampling times is arranged in time sequence to obtain the pressure difference sequence of the corresponding monitoring period. Linear fitting and quadratic polynomial fitting are performed on the pressure difference sequence of each monitoring period. The slope of the straight line obtained by linear fitting is used as the pressure difference attenuation slope of the corresponding monitoring period, and twice the coefficient of the quadratic term of the quadratic polynomial fitting result is used as the pressure difference attenuation curvature of the corresponding monitoring period.

[0101] The second type of internal leakage vector element acquisition unit is used to analyze the temperature data of the downstream pipeline 2 of the hydrogen energy ball valve collected by the temperature sensor through wavelet transform, extract the characteristic instantaneous temperature drop pulse within the monitoring period, take the maximum value of the characteristic instantaneous temperature drop pulse as the instantaneous temperature drop pulse amplitude within the corresponding monitoring period, and take the number of characteristic instantaneous temperature drop pulses as the instantaneous temperature drop pulse frequency.

[0102] The internal leakage state vector generation unit arranges the differential pressure attenuation slope, differential pressure attenuation curvature, instantaneous temperature drop pulse amplitude, and instantaneous temperature drop pulse frequency of each monitoring cycle in order to form the internal leakage state vector.

[0103] In this embodiment, the pressure difference between the upstream pipeline 1 and the downstream pipeline 2 of the hydrogen ball valve at the j-th sampling time of the i-th monitoring cycle is: ,in The value detected by the high-precision micro differential pressure sensor corresponding to the upstream pipeline 1 of the hydrogen ball valve at the j-th sampling time of the i-th monitoring cycle is given. The value is the detection value of the high-precision micro differential pressure sensor corresponding to the downstream pipeline 2 of the hydrogen ball valve at the j-th sampling time of the i-th monitoring cycle.

[0104] In this embodiment, the differential pressure sequence for the i-th monitoring period is: { , , , , ... } where n is the total number of samples taken by the high-precision micro differential pressure sensor for each monitoring cycle.

[0105] The beneficial effects of the above technical solution are as follows: This invention constructs a feature vector that can more finely and robustly characterize the internal leakage state, significantly improving the accuracy of internal leakage detection and early detection capability. This invention not only calculates the pressure difference attenuation slope to assess the average leakage rate, but also innovatively introduces a quadratic fitting curvature to characterize the dynamic evolution trend of whether the leakage process is uniform, accelerating, or decelerating. At the same time, by extracting the amplitude and frequency features of the instantaneous temperature drop pulse from the downstream temperature signal through wavelet transform, it combines the pressure difference attenuation reflecting the flow characteristics with the temperature pulse reflecting the thermodynamic effect. The multi-evidence fusion mechanism can effectively distinguish between real internal leakage and interference such as sensor drift, providing a more robust and information-rich internal leakage state input for subsequent overall health assessment.

[0106] Example 4

[0107] Based on Example 2, the external leakage state vector generation submodule includes:

[0108] A type of external leakage vector element acquisition unit is used to perform fast Fourier transform on the sound pressure signal collected by the ultrasonic sensor at each sampling time in each monitoring cycle to obtain the average power spectral density of the corresponding monitoring cycle. Based on the average power spectral density of the corresponding monitoring cycle, the cumulative energy of the preset hydrogen turbulent jet characteristic frequency band of the corresponding monitoring cycle is obtained and used as the cumulative energy of the turbulent jet of the corresponding monitoring cycle. The standard deviation of the preset energy distribution of the hydrogen turbulent jet characteristic frequency band is used as the turbulent jet bandwidth of the corresponding monitoring cycle.

[0109] The second type of leakage vector element acquisition unit is used to obtain the average rate of change of hydrogen concentration in the corresponding monitoring period based on the hydrogen accumulation concentration data collected by the hydrogen concentration sensor at each sampling time of each monitoring period.

[0110] The external leakage state vector generation unit is used to arrange the cumulative energy of the turbulent jet, the bandwidth of the turbulent jet, and the average rate of change of hydrogen concentration in each monitoring cycle in order to form an external leakage state vector.

[0111] In this embodiment, the preset characteristic frequency band of hydrogen turbulent jet is: when hydrogen leaks through the tiny gaps in the valve stem packing to form a turbulent jet, it will excite a wideband gas dynamic noise. Among them, the frequency components most related to the leakage state and with the highest signal-to-noise ratio are concentrated in a specific frequency band, which is the characteristic frequency band of hydrogen turbulent jet.

[0112] In this embodiment, performing a fast Fourier transform on the sound pressure signal collected by the ultrasonic sensor at each sampling moment within each monitoring cycle to obtain the average power spectral density of the corresponding monitoring cycle is a prior art technique, specifically including:

[0113] To obtain stable frequency domain characteristics within a monitoring period, the continuous sound pressure signal acquired by the ultrasonic sensor is processed according to the following steps: First, the sound pressure signal of the entire monitoring period is divided into multiple consecutive short time intervals; then, a fast Fourier transform is performed on each frame of the signal to convert it from a time-varying waveform into a spectrum reflecting the intensity of each frequency component; next, the power spectral density of each frame is calculated based on the spectrum, i.e., the energy distribution carried by each frequency component; finally, the power spectral density of all frames is averaged at the same frequency point to obtain the average power spectral density representing the entire monitoring period. This average power spectral density can stably characterize the energy distribution of the signal in the frequency domain, laying the foundation for subsequent extraction of turbulent jet characteristics related to hydrogen leakage.

[0114] In this embodiment, the preset characteristic frequency band of the hydrogen turbulent jet is set as [ Given that the average power spectral density is S(f), the cumulative energy of the characteristic frequency band of the hydrogen turbulent jet is... .

[0115] In this embodiment, the standard deviation of the energy distribution of the hydrogen turbulent jet characteristic frequency band is used as the turbulent jet bandwidth feature. The energy corresponding to each hydrogen turbulent jet characteristic frequency is the corresponding characteristic frequency multiplied by the average power spectral density. The turbulent jet bandwidth feature is the standard deviation of the energy corresponding to each characteristic frequency in the hydrogen turbulent jet characteristic frequency band.

[0116] In this embodiment, the average rate of change of hydrogen concentration within the corresponding monitoring period is the average of the rates of change of hydrogen concentration at several adjacent sampling times.

[0117] The beneficial effects of the above technical solution are as follows: This invention extracts two frequency domain features that are strongly correlated with the physical mechanism of leakage and have strong anti-noise capabilities: the cumulative energy of turbulent jet and the bandwidth, by performing fast Fourier transform and characteristic frequency band analysis on ultrasonic signals. At the same time, it integrates the time domain trend index of the average change rate of hydrogen concentration, so that the external leakage state vector can not only quickly capture the intensity of sudden leakage through acoustic energy, but also sensitively detect the cumulative effect of slow micro-leakage through concentration changes. This forms a more comprehensive and in-depth description of external leakage events, providing a reliable external leakage dimension criterion for overall assessment.

[0118] Example 5

[0119] Based on Example 2, the hydrogen embrittlement state vector generation submodule includes:

[0120] A hydrogen embrittlement vector element acquisition unit is used to count the number of voltage signals detected by the high-frequency acoustic emission sensor at each sampling time in each monitoring cycle that exceed the preset voltage threshold, and use this as the total number of acoustic emission events. Based on the total number of acoustic emission events and the total number of sampling times in each monitoring cycle, the occurrence rate of acoustic emission events in each monitoring cycle is calculated. The mean value of the voltage signal corresponding to each acoustic emission event in the corresponding monitoring cycle is used as the acoustic emission event intensity evaluation value in the corresponding monitoring cycle.

[0121] The second-class hydrogen embrittlement vector element acquisition unit is used to calculate the root mean square value and kurtosis of the time-series composite acceleration amplitude sequence within each monitoring period based on the composite acceleration amplitude detected by the triaxial vibration accelerometer at each sampling time within each monitoring period, and obtain the vibration root mean square value and vibration kurtosis of the corresponding monitoring period.

[0122] The hydrogen embrittlement state vector generation unit is used to arrange the acoustic emission event occurrence rate, acoustic emission event intensity assessment value, vibration root mean square value and vibration kurtosis in order for each monitoring period to form a hydrogen embrittlement state vector.

[0123] In this embodiment, the acoustic emission event occurrence rate for each monitoring cycle is the quotient of the total number of acoustic emission events in each monitoring cycle and the total number of sampling times in each monitoring cycle.

[0124] In this embodiment, the formula for calculating the composite acceleration amplitude of the three directions at the j-th sampling time of the i-th monitoring period is:

[0125] ;in, , and Let x, y, and z be the accelerations in the x, y, and z directions, respectively, at the j-th sampling time of the i-th monitoring period. It represents the combined acceleration amplitude of the acceleration in the three directions at the j-th sampling time of the i-th monitoring period.

[0126] In this embodiment, the time-series-based synthetic acceleration amplitude sequence within the i-th monitoring period is: { };in, These are the composite acceleration amplitudes at the 1st, 2nd, 3rd, 4th...nth sampling times in the i-th monitoring period.

[0127] In this embodiment, the root mean square value of the time-series-based synthetic acceleration amplitude sequence within the monitoring period is calculated using the following formula: ;in, is the root mean square value of the time-series synthesized acceleration amplitude sequence for the i-th monitoring period, representing the average energy level of the vibration signal within the corresponding monitoring period, and n is the total number of sampling times for the i-th monitoring period.

[0128] In this embodiment, the kurtosis of the time-series-based synthetic acceleration amplitude sequence within the monitoring period is:

[0129] ;in, This represents the kurtosis of the time-series-based synthetic acceleration amplitude sequence within the i-th monitoring period. This value is highly sensitive to impact components (such as spikes) in the signal. Surface irregularities caused by hydrogen embrittlement can generate such impacts when the valve core 4 and the valve seat 5 of the hydrogen ball valve are engaged.

[0130] The working principle and beneficial effects of the above technical solution: This invention creatively utilizes the acoustic emission and vibration signals excited during the coordinated movement of the valve core 4 and the valve seat 5 of the hydrogen-powered ball valve to construct a comprehensive feature vector that can dynamically assess the activity of hydrogen embrittlement microcracks and the degree of material surface deterioration online. This achieves early online warning of hydrogen embrittlement, a hidden failure, filling the technical gap of traditional reliance on periodic offline detection. By statistically analyzing the occurrence rate and average intensity of acoustic emission events, this invention quantifies the activity and energy release level of hydrogen-induced microcrack generation and propagation. By calculating the root mean square value and kurtosis of the vibration signal, it reflects the vibration energy and impact characteristics caused by surface roughening due to hydrogen embrittlement during coordinated movement. This provides a multi-scale perspective from microscopic initiation to macroscopic manifestation for online monitoring of hydrogen embrittlement damage, achieving the effect of detecting hydrogen embrittlement in the early stage of hydrogen embrittlement failure.

[0131] Example 6

[0132] Based on Example 1, the sealing condition diagnostic analysis module includes:

[0133] The feature space construction submodule is used to construct the ball valve preset sealing health three-dimensional feature space based on the internal leakage state vector, external leakage state vector and hydrogen embrittlement state vector corresponding to all historical normal sealing monitoring cycles.

[0134] The state coordinate point determination submodule is used to project the internal leakage state vector, external leakage state vector, and hydrogen embrittlement state vector of each hydrogen ball valve in the current monitoring cycle onto the internal leakage principal component vector, external leakage principal component vector, and hydrogen embrittlement principal component vector, respectively, to obtain the internal leakage scalar value x, external leakage scalar value y, and hydrogen embrittlement scalar value z, and use (x, y, z) as the state coordinate point P of the current monitoring cycle;

[0135] The health deviation calculation submodule is used to calculate the Mahalanobis distance between the state coordinate point P of the current monitoring period and the cluster Q of state coordinate points of all historical normal monitoring periods. And use it as the seal health deviation for the current monitoring cycle;

[0136] The status assessment result output submodule is used to compare the seal health deviation of the current monitoring period with the preset seal health deviation threshold range and output the corresponding seal health status level, which includes healthy, alert, warning and abnormal.

[0137] In this embodiment, the cluster Q of all historical normal monitoring cycle status coordinate points is a data distribution composed of status coordinate points corresponding to historical normal monitoring cycles.

[0138] In this embodiment, the Mahalanobis distance between the state coordinate point P of the current monitoring cycle and the cluster Q of state coordinate points of all historical normal monitoring cycles is calculated. Specifically, it includes:

[0139] ;in, This represents the Mahalanobis distance between the current monitoring cycle's state coordinate point P and the cluster Q of all historical normal monitoring cycle state coordinate points. This represents the status coordinate point for the current monitoring period. This is the mean vector of the coordinates of all historical normal monitoring cycle status points. The covariance matrix is ​​calculated based on the cluster Q of state coordinate points from all historical normal monitoring cycles. The inverse of the covariance matrix of cluster Q, representing all historical normal monitoring cycle status coordinate points. For vectors The transpose of .

[0140] In this embodiment, the sealing health deviation of the current monitoring period is compared with a preset sealing health deviation threshold range, and the corresponding sealing health status level is output, specifically including:

[0141] Pre-set a set of ordered thresholds for seal health deviation; for example, set three thresholds. , , ,and ;

[0142] like If so, the sealing condition is determined to be healthy;

[0143] like If so, the seal is deemed to be in good condition and requires attention.

[0144] like If so, the sealing health status is determined to be a warning;

[0145] like If so, the sealing health status is determined to be abnormal.

[0146] The beneficial effects of the above technical solution are as follows: Unlike traditional technologies that only provide isolated parameter alarms or rely on manual comprehensive judgment, this invention projects the three types of state vectors of internal leakage, external leakage, and hydrogen embrittlement onto a unified three-dimensional feature space, and calculates the Mahalanobis distance between the current state point and the historical normal state cluster, to obtain a single deviation value that comprehensively considers all features and their correlations. This not only achieves accurate classification of health status from qualitative to quantitative (health, attention, warning, abnormal), but also, because Mahalanobis distance can automatically eliminate the influence of dimensions and consider the correlation between features, it is extremely sensitive to weak anomalies that occur in multiple dimensions in a coordinated manner. Thus, it scientifically and efficiently completes the comprehensive screening and level determination of the overall sealing health status of the valve.

[0147] Example 7

[0148] Based on Example 6, the feature space construction submodule includes:

[0149] The analysis matrix determination unit is used to construct analysis matrix one, analysis matrix two and analysis matrix three based on the internal leakage state vector, external leakage state vector and hydrogen embrittlement state vector corresponding to all historical normal sealing monitoring cycles;

[0150] Principal component vector one determination unit is used to perform principal component analysis on analysis matrix one to obtain the principal component vector of analysis matrix one, and use it as principal component vector one;

[0151] The principal component vector two determination unit is used to perform principal component analysis on analysis matrix two to obtain the principal component vector of analysis matrix two, and use it as the principal component vector two.

[0152] The principal component vector three determination unit is used to perform principal component analysis on the analysis matrix three to obtain the principal component vector of the analysis matrix three, and use it as the principal component vector three.

[0153] The feature space construction unit is used to orthogonalize the principal component vectors 1, 2, and 3 to form the internal leakage principal component vector, external leakage principal component vector, and hydrogen embrittlement principal component vector. Based on the internal leakage principal component vector, external leakage principal component vector, and hydrogen embrittlement principal component vector, the ball valve preset sealing health three-dimensional feature space is constructed.

[0154] In this embodiment, constructing the analysis matrix one based on the internal leakage state vectors corresponding to all historical normal sealing monitoring cycles includes: arranging and combining the internal leakage state vectors of several normal sealing monitoring cycles in a time sequence, either horizontally or vertically, to form the analysis matrix one.

[0155] In this embodiment, constructing analysis matrix two based on the external leakage state vectors corresponding to several historical normal sealing monitoring cycles includes: arranging and combining the external leakage state vectors of several normal sealing monitoring cycles in a time sequence, either horizontally or vertically, to form analysis matrix two.

[0156] In this embodiment, constructing analysis matrix three based on hydrogen embrittlement state vectors corresponding to several historical normal sealing monitoring cycles includes: arranging and combining the hydrogen embrittlement state vectors of several normal sealing monitoring cycles in a time sequence, either horizontally or vertically, to form analysis matrix three.

[0157] In this embodiment, "performing principal component analysis on analysis matrix one to obtain the principal component vector of analysis matrix one, and using it as principal component vector one," "performing principal component analysis on analysis matrix two to obtain the principal component vector of analysis matrix two, and using it as principal component vector two," and "performing principal component analysis on analysis matrix three to obtain the principal component vector of analysis matrix three, and using it as principal component vector three" are all existing technologies. Specifically, taking "performing principal component analysis on analysis matrix one to obtain the principal component vector of analysis matrix one, and using it as principal component vector one" as an example:

[0158] First, calculate the mean of each column in analysis matrix one. Then, subtract the mean of the column from each value in that column to obtain the centered analysis matrix one. Based on the centered analysis matrix one, calculate its covariance matrix. This covariance matrix reflects the correlation between different leakage feature dimensions (such as pressure drop attenuation slope, curvature, etc.). Perform eigenvalue decomposition on the calculated covariance matrix to obtain a set of eigenvalues ​​and corresponding eigenvectors. The magnitude of each eigenvalue represents the degree of data change in the direction of its corresponding eigenvector. Sort all eigenvalues ​​from largest to smallest and select the eigenvector corresponding to the largest eigenvalue as principal component vector one. Principal component vector one represents the most important change pattern in the historical leakage data, that is, principal component vector one is the direction that best distinguishes different leakage states.

[0159] In this embodiment, principal component vector one, principal component vector two, and principal component vector three are orthogonalized, specifically using the Gram-Schmidt orthogonalization method.

[0160] Taking the principal component vector as the first coordinate axis direction, denoted as vector... Calculate the principal component vectors in vector... The projection component in the direction is then subtracted from the principal component vector 2 to obtain the vector. The new perpendicular vector, after being normalized, is denoted as the direction vector of the second coordinate axis. Calculate the principal component vectors three respectively in the vector sum vector The projection components in the direction are then subtracted from the principal component vectors simultaneously to obtain the vectors simultaneously with the vectors. sum vector A new vector perpendicular to all axes is normalized and denoted as the direction of the third coordinate axis. , where vector ,vector sum vector These are the internal leakage principal component vector, the external leakage principal component vector, and the hydrogen embrittlement principal component vector, respectively.

[0161] The beneficial effects of the above technical solution are as follows: Based on a large amount of historical normal data, the present invention extracts the most important variation directions of internal leakage, external leakage, and hydrogen embrittlement state as principal component vectors through principal component analysis, ensuring that the evaluation benchmark truly reflects the natural fluctuation distribution of the equipment in a healthy state. Gram-Schmidt orthogonalization is used to make the three coordinate axes independent of each other, which not only enhances the interpretability of the space, but also makes each axis mainly represent the health change of a failure mode, providing a rigorous mathematical basis and geometric framework for the subsequent scientific decomposition of the overall health deviation and tracing the cause to specific failure modes.

[0162] Example 8

[0163] Based on Example 6, the seal failure cause analysis module includes:

[0164] The elementary failure mode spectrum library construction submodule is used to construct three mutually orthogonal elementary failure mode vectors, including elementary vector one representing pure internal leakage of the sealing part, elementary vector two representing pure external leakage of the packing part, and elementary vector three representing hydrogen embrittlement of the pure sealing part. The space composed of elementary vector one, elementary vector two, and elementary vector three is used as the elementary space.

[0165] The deviation vector acquisition submodule takes the difference between the state coordinate point P corresponding to the monitoring cycle with an abnormal sealing health status level and the center point O of the cluster Q of state coordinate points from all historical normal monitoring cycles as the deviation vector for the current abnormal monitoring cycle. ;

[0166] The failure mode determination submodule will determine the deviation vector of the current anomaly monitoring cycle. Projecting onto the primitive space, calculate the deviation vector for the current anomaly monitoring period. Projection coefficients are plotted on primitive vector one, primitive vector two, and primitive vector three respectively, and each projection coefficient is normalized to obtain the contribution percentage of each primitive failure mode at the state coordinate point P corresponding to the current anomaly monitoring cycle. The dominant failure type is determined based on the primitive failure mode with the largest contribution percentage, and a complete sealing failure contribution spectrum is output.

[0167] In this embodiment, three mutually orthogonal primitive failure mode vectors are constructed, including primitive vector one representing pure internal leakage of the sealing pair, primitive vector two representing pure external leakage of the packing, and primitive vector three representing hydrogen embrittlement of the pure sealing pair. Specifically, they include:

[0168] Step 1: Filter monitoring cycle data from the historical fault database that only exhibits a single fault type:

[0169] Pure internal leakage fault data: Only internal leakage characteristics are present, with no external leakage or hydrogen embrittlement characteristics;

[0170] Pure external leakage fault data: Only external leakage characteristics are present, with no internal leakage or hydrogen embrittlement characteristics;

[0171] Pure hydrogen embrittlement fault data: Only hydrogen embrittlement characteristics are present, with no internal or external leakage characteristics;

[0172] Step 2: For the three types of single fault data mentioned above, calculate the mean vector of their corresponding state vectors, which will serve as the prototype vectors for the corresponding fault modes:

[0173] The mean of the internal leakage state vector corresponding to a pure internal leakage fault is used as the internal leakage prototype vector.

[0174] The mean of the external leakage state vector corresponding to a pure external leakage fault is used as the external leakage prototype vector.

[0175] The mean of the hydrogen embrittlement state vector corresponding to a pure hydrogen embrittlement fault is used as the hydrogen embrittlement prototype vector.

[0176] Step 3: Using the Gram-Schmidt orthogonal method, orthogonalize the internal leakage prototype vector, external leakage prototype vector, and hydrogen embrittlement prototype vector to obtain mutually orthogonal primitive failure mode vectors:

[0177] Pick ;

[0178] calculate : Subtract its value from the exposed prototype vector Project the image onto the direction and normalize the result.

[0179] calculate : Subtract its value from the hydrogen embrittlement prototype vector and Project the image along the direction and normalize the result.

[0180] in, , and These are, respectively, the first primitive vector characterizing internal leakage of a pure sealing pair, the second primitive vector characterizing external leakage of a pure packing pair, and the third primitive vector characterizing hydrogen embrittlement of a pure sealing pair.

[0181] In this embodiment, the center point O of the cluster Q of all historical normal monitoring cycle state coordinate points is the arithmetic mean of all state coordinate points in the cluster Q of all historical normal monitoring cycle state coordinate points, and the deviation vector of the current abnormal monitoring cycle is... .

[0182] In this embodiment, the projection coefficients are normalized to obtain the contribution percentage of each primitive failure mode at the state coordinate point P corresponding to the current anomaly monitoring period. The specific operation steps include:

[0183] The deviation vector of the current anomaly monitoring period Projecting onto primitive vector one, primitive vector two, and primitive vector three yields three projection coefficients:

[0184] Deviation vector of the current anomaly monitoring period Projection coefficients on the first primitive vector:

[0185] ;

[0186] Deviation vector of the current anomaly monitoring period Projection coefficients on primitive vector two:

[0187] ;

[0188] Deviation vector of the current anomaly monitoring period The projection coefficients on the three primitive vectors:

[0189] ;in, Represents the dot product;

[0190] The percentage contribution of the internal leakage primitive failure mode to the state coordinate point P corresponding to the current anomaly monitoring cycle:

[0191] ;

[0192] The percentage contribution of the external leakage primitive failure mode to the state coordinate point P corresponding to the current anomaly monitoring cycle:

[0193] ;

[0194] The percentage contribution of the hydrogen embrittlement elementary failure mode to the state coordinate point P corresponding to the current anomaly monitoring cycle:

[0195] .

[0196] In this embodiment, the dominant failure type is determined based on the elementary failure mode with the largest contribution percentage, and a complete sealing failure contribution spectrum is output, as follows:

[0197] Dominant failure mode: A failure mode of an elementary element that contributes ≥50% of the total failure rate;

[0198] Associated anomaly modes: primitive failure modes with a contribution percentage of <50%, further subdivided into:

[0199] Minor anomalies: Contribution percentage between 20% and 50%;

[0200] Abnormal tendencies: Contribution percentage between 5% and 20%;

[0201] Ignoreable anomalies: Contribution percentage < 5% (can be excluded from the report);

[0202] The dominant failure modes include: internal leakage of the sealing pair, external leakage of the packing, and hydrogen embrittlement of the sealing pair;

[0203] Associated abnormal modes include: slight internal leakage of the sealing pair, slight external leakage of the packing, slight hydrogen embrittlement of the sealing pair, tendency for internal leakage of the sealing pair, tendency for external leakage of the packing, and tendency for hydrogen embrittlement of the sealing pair.

[0204] The beneficial effects of the above technical solution are as follows: This invention provides a complete quantitative fault attribution method based on vector space projection, which can accurately decompose the overall anomaly into the contribution of different failure modes. When the sealing condition diagnosis module determines that there is an anomaly, the analysis is initiated: First, a mutually orthogonal primitive failure mode vector space derived from historical fault data of pure internal leakage, pure external leakage, and pure hydrogen embrittlement is constructed. Then, the deviation vector of the abnormal state point relative to the normal cluster center is calculated and projected onto the primitive space. Finally, by calculating the projection coefficient and normalizing, the contribution percentage of internal leakage, external leakage, and hydrogen embrittlement to the current abnormal state is directly output. Finally, the sealing failure contribution spectrum is output, providing maintenance personnel with clear information on the dominant failure type and associated anomalies. This transforms maintenance decisions from experience-based guesswork to data-based precise positioning, greatly improving the pertinence and efficiency of maintenance work.

[0205] Example 9

[0206] Based on Example 1, the comprehensive hazard assessment module includes:

[0207] The Hazardous Heat Map Construction Submodule is used to create a 3D layout map based on hydrogen energy facilities, mark the precise installation location of each hydrogen ball valve, construct an initial hydrogen ball valve distribution map, and divide the hydrogen ball valve distribution map into several consecutive assessment areas.

[0208] The regional risk value calculation submodule is used to calculate the comprehensive risk thermal value of each assessment area based on the sealing health status level of each hydrogen ball valve in the current monitoring cycle, and based on the real-time risk values ​​of all hydrogen ball valves corresponding to each assessment area.

[0209] The risk assessment visualization submodule is used to generate a dynamic hazard heat map by rendering it on the hydrogen energy ball valve distribution map with a color gradient from cool to warm colors based on the comprehensive risk heat value of each assessment area.

[0210] The risk avoidance path generation module is used to automatically generate and display the optimal maintenance risk avoidance path to an assessment area when the comprehensive risk heat value of an assessment area exceeds a preset high-risk threshold.

[0211] In this embodiment, based on the three-dimensional layout map of the hydrogen energy facility, the precise installation location of each hydrogen ball valve is marked, and an initial hydrogen ball valve distribution map is constructed. This can be achieved in the following three ways:

[0212] The first method involves importing the design drawings (such as CAD 2D drawings or BIM 3D models) of hydrogen energy facilities into the system, and then associating and mapping the geometric coordinates in the drawings with the actual hydrogen energy ball valve equipment through predefined or automatic recognition of graphic elements (such as valve symbols and pipes).

[0213] The second method involves loading a floor plan of the hydrogen energy facility (e.g., PNG or JPG format). Through a human-computer interaction interface, the administrator can manually drag and drop the icon of each hydrogen ball valve and precisely position it to the actual installation location on the drawing. The system records the position of each icon in the pixel coordinate system of the drawing or the converted actual spatial coordinate system.

[0214] The third method involves using laser scanning or photogrammetry to acquire three-dimensional point cloud data of hydrogen energy facilities. Then, through point cloud segmentation and recognition algorithms, valve equipment is automatically or semi-automatically identified, and its spatial coordinates are determined, thereby generating a three-dimensional visualization map containing the location of the equipment.

[0215] In this embodiment, the hydrogen energy ball valve distribution map can be divided into several continuous evaluation areas according to the following rules:

[0216] Based on the process flow of hydrogen energy facilities, such as hydrogen production area, purification area, compression area, hydrogen storage area, and refueling area, ball valves in the same process section have similar operating pressure and media conditions.

[0217] The factory is divided according to its physical structure, such as different floors, different fire compartments, different rooms or pipe corridors, to facilitate on-site inspection and emergency management.

[0218] Combining the hydrogen diffusion model, and considering ventilation conditions, space volume, and the location of potential leak sources, a continuous space that may be affected by the same leak event is divided into an assessment area.

[0219] On digital maps, continuous evaluation areas can be generated automatically or semi-automatically using predefined geofence polygons, region growing algorithms utilizing image processing, or spatial clustering algorithms based on valve location density.

[0220] In this embodiment, the sealing health status level of each hydrogen ball valve in the current monitoring cycle is specifically as follows:

[0221] When the diagnosis result is healthy, the seal health status level is 0; when the diagnosis result is alert, the seal health status level is 1; when the diagnosis result is warning, the seal health status level is 2; and when the diagnosis result is abnormal, the seal health status level is 3.

[0222] The dominant failure types include: internal leakage of the sealing pair, external leakage of the packing, and hydrogen embrittlement of the sealing pair.

[0223] In this embodiment, based on the real-time risk values ​​of all hydrogen ball valves corresponding to each assessment area, the comprehensive risk thermal value of each assessment area is calculated using the following formula:

[0224] ;in, Let the comprehensive risk thermal value be the k-th assessment area. Let e ​​be the logarithm to the base e. This represents the average real-time risk value of all hydrogen ball valves within the k-th assessment region. This represents the maximum real-time risk value of all hydrogen ball valves within the k-th assessment region. and These are the influence coefficients of the average real-time risk value of all hydrogen-powered ball valves on the regional hazard level and the influence coefficients of the maximum real-time risk value of all hydrogen-powered ball valves on the regional hazard level, respectively.

[0225] In this embodiment, based on the comprehensive risk heat value of each assessment area, a dynamic hazard heat map is generated on the hydrogen ball valve distribution map using a color gradient from cool to warm colors. This is achieved using Web-based geographic information system rendering technology or a custom Canvas / SVG graphics rendering technology. Specifically, a continuous color gradient from cool colors (e.g., blue / green, representing low risk) to warm colors (e.g., yellow / orange / red, representing high risk) is predefined. The comprehensive risk heat values ​​calculated for all assessment areas are normalized to the [0,1] interval. Based on the normalized comprehensive risk heat value of each area, the corresponding color (RGB value) is interpolated on the color gradient. On the display device, each assessment area in the hydrogen ball valve distribution map is filled or rendered semi-transparently according to the calculated color, thereby generating an intuitive hazard heat map. This map can be dynamically updated, changing with the refresh of assessment results in each monitoring cycle.

[0226] In this embodiment, when the comprehensive risk thermal value of a certain assessment area exceeds the preset high-risk threshold, it indicates that there is a high risk of hydrogen leakage or equipment failure in that area, and manual maintenance needs to bypass this area.

[0227] In this embodiment, when the comprehensive risk thermal value of a certain assessment area exceeds a preset high-risk threshold, the optimal evacuation path to that area is automatically generated and displayed, specifically including:

[0228] Starting from the real-time location of the staff receiving the alarm, and ending at the safe entrance or designated safe observation point of the assessment area that exceeds the preset high-risk threshold (not directly pointing to the faulty valve itself, ensuring that personnel can observe and conduct preliminary inspections from a safe distance), the optimal path is automatically calculated based on a weighted path planning algorithm. The path weight can consider geometric distance (the length of each evacuation path from the start to the end) and path risk coefficient (the comprehensive risk heat value level of other assessment areas traversed by each evacuation path from the start to the end, prioritizing evacuation paths that avoid other medium- and high-risk areas). Finally, the optimal evacuation path is dynamically overlaid on the hazard heat map as a highlighted line (such as a green dashed line with an arrow), and the path length and estimated travel time are marked. At the same time, voice or text prompts are provided to guide staff to travel safely according to the optimal evacuation path.

[0229] The beneficial effects of the above technical solution are as follows: Based on the health status level of each hydrogen energy ball valve, the present invention generates a dynamic hazard heat map covering the entire plant through aggregation calculation. The risk level of different areas is intuitively rendered with warm and cool color gradients, making the overall safety situation clear at a glance. When the risk of a certain area exceeds the high-risk threshold, it can automatically plan an optimal maintenance and hazard avoidance path to avoid other high-risk areas and guide the staff in real time. This extends safety protection from passive alarm to active path planning and behavior guidance, significantly improving the safety level and operational efficiency of emergency response and daily maintenance of hydrogen energy facilities.

[0230] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An online sealing testing device for hydrogen-powered ball valves, characterized in that: include: The multi-source sealing data acquisition module is used to acquire the original multi-physics field data of each hydrogen ball valve in real time, and generate the internal leakage state vector, external leakage state vector and hydrogen embrittlement state vector of each hydrogen ball valve in the current monitoring cycle based on the original multi-physics field data. The sealing status diagnosis and analysis module is used to determine the sealing health status of each hydrogen ball valve in the current monitoring cycle based on the internal leakage status vector, external leakage status vector, hydrogen embrittlement status vector, and the preset three-dimensional feature space of sealing health of each ball valve in the current monitoring cycle. The sealing failure cause analysis module is used to trace and analyze the causes of sealing failures of hydrogen ball valves with abnormal sealing health status, and output a complete sealing failure contribution spectrum based on the traceability analysis results. The comprehensive hazard assessment module is used to render the distribution map of hydrogen ball valves based on the sealing health status diagnosis results of each hydrogen ball valve in the current monitoring cycle, generate a dynamic hazard heat map, and generate the optimal maintenance and risk avoidance path based on the hazard heat map. The multi-source sealing data acquisition module includes: The internal leakage sensing and detection submodule includes two sets of high-precision micro differential pressure sensors installed on the upstream pipeline (1) and downstream pipeline (2) of the hydrogen ball valve, and a temperature sensor installed on the downstream pipeline (2) of the hydrogen ball valve. The two sets of high-precision micro differential pressure sensors are used to collect pressure data of the upstream pipeline (1) and downstream pipeline (2) of the hydrogen ball valve when the valve core (4) of the hydrogen ball valve is closed, respectively. The temperature sensor is used to collect temperature data of the downstream pipeline (2) of the hydrogen ball valve. The external leakage sensing and detection submodule includes an ultrasonic sensor and a hydrogen concentration sensor installed on the outer wall of the hydrogen ball valve stem packing assembly (3). The ultrasonic sensor is used to collect the sound pressure data of the hydrogen turbulent jet leaking from the hydrogen ball valve stem packing assembly (3), and the hydrogen concentration sensor is used to collect the hydrogen accumulation concentration data of the current hydrogen ball valve environment. The hydrogen embrittlement sensing and detection submodule includes a high-frequency acoustic emission sensor and a triaxial vibration acceleration sensor installed on the body of the hydrogen ball valve. The high-frequency acoustic emission sensor is used to collect the voltage data detected by the high-frequency acoustic emission sensor when the valve core (4) and the valve seat (5) of the hydrogen ball valve are in a coordinated motion state. The triaxial vibration acceleration sensor is used to collect the vibration data of the valve core (4) and the valve seat (5) of the hydrogen ball valve in a coordinated motion state. The internal leakage state vector generation submodule is used to generate the internal leakage state vector of the hydrogen ball valve in the current monitoring period based on the pressure data of the upstream pipeline (1) and downstream pipeline (2) of the hydrogen ball valve and the temperature data of the downstream pipeline (2) of the hydrogen ball valve in the current monitoring period when the valve core (4) of the hydrogen ball valve is closed. The external leakage state vector generation submodule is used to generate the external leakage state vector of the hydrogen ball valve for the current monitoring period based on the acoustic pressure data of the hydrogen turbulent jet collected by the ultrasonic sensor and the hydrogen accumulation concentration data collected by the hydrogen concentration sensor. The hydrogen embrittlement state vector generation submodule is used to generate the hydrogen embrittlement state vector of the hydrogen ball valve in the current monitoring period based on the voltage data detected by the high-frequency acoustic emission sensor and the vibration data collected by the triaxial vibration acceleration sensor. The internal leakage state vector is a mathematical vector composed of a set of internal leakage characteristic elements, used to quantify and characterize the probability, severity and dynamic characteristics of internal leakage in the hydrogen ball valve sealing pair during the current monitoring period. The external leakage state vector is a mathematical vector composed of a set of external leakage characteristic elements, used to quantify and characterize the possibility, leakage intensity and leakage flow characteristics of external leakage at the valve stem packing assembly (3) of the hydrogen ball valve during the current monitoring period; The hydrogen embrittlement state vector is a mathematical vector composed of a set of hydrogen embrittlement characteristic elements. It is used to quantify and characterize the possibility, degree of damage, and activity of microcracks or surface degradation in the hydrogen-powered ball valve sealing pair due to hydrogen-induced damage during the current monitoring period.

2. The online sealing testing device for hydrogen-powered ball valves according to claim 1, characterized in that: The internal leakage state vector generation submodule includes: A type of internal leakage vector element acquisition unit is used to calculate the pressure difference between the upstream pipeline (1) and the downstream pipeline (2) of the hydrogen ball valve based on the detection values ​​of two sets of high-precision micro differential pressure sensors at each sampling time in the monitoring cycle. The pressure difference at several sampling times is arranged in time sequence to obtain the pressure difference sequence of the corresponding monitoring cycle. Linear fitting and quadratic polynomial fitting are performed on the pressure difference sequence of each monitoring cycle. The slope of the straight line obtained by linear fitting is used as the pressure difference attenuation slope of the corresponding monitoring cycle. Twice the coefficient of the quadratic term of the quadratic polynomial fitting result is used as the pressure difference attenuation curvature of the corresponding monitoring cycle. The second type of internal leakage vector element acquisition unit is used to analyze the temperature data of the downstream pipeline (2) of the hydrogen energy ball valve collected by the temperature sensor through wavelet transform, extract the characteristic instantaneous temperature drop pulse within the monitoring period, take the maximum value of the characteristic instantaneous temperature drop pulse as the instantaneous temperature drop pulse amplitude within the corresponding monitoring period, and take the number of characteristic instantaneous temperature drop pulses as the instantaneous temperature drop pulse frequency. The internal leakage state vector generation unit arranges the differential pressure attenuation slope, differential pressure attenuation curvature, instantaneous temperature drop pulse amplitude, and instantaneous temperature drop pulse frequency of each monitoring cycle in order to form the internal leakage state vector.

3. The online sealing testing device for hydrogen-powered ball valves according to claim 1, characterized in that: The external state vector generation submodule includes: A type of external leakage vector element acquisition unit is used to perform fast Fourier transform on the sound pressure signal collected by the ultrasonic sensor at each sampling time in each monitoring cycle to obtain the average power spectral density of the corresponding monitoring cycle. Based on the average power spectral density of the corresponding monitoring cycle, the cumulative energy of the preset hydrogen turbulent jet characteristic frequency band of the corresponding monitoring cycle is obtained and used as the cumulative energy of the turbulent jet of the corresponding monitoring cycle. The standard deviation of the preset energy distribution of the hydrogen turbulent jet characteristic frequency band is used as the turbulent jet bandwidth of the corresponding monitoring cycle. The second type of leakage vector element acquisition unit is used to obtain the average rate of change of hydrogen concentration in the corresponding monitoring period based on the hydrogen accumulation concentration data collected by the hydrogen concentration sensor at each sampling time of each monitoring period. The external leakage state vector generation unit is used to arrange the cumulative energy of the turbulent jet, the bandwidth of the turbulent jet, and the average rate of change of hydrogen concentration in each monitoring cycle in order to form an external leakage state vector.

4. The online sealing testing device for hydrogen-powered ball valves according to claim 1, characterized in that: The hydrogen embrittlement state vector generation submodule includes: A hydrogen embrittlement vector element acquisition unit is used to count the number of voltage signals detected by the high-frequency acoustic emission sensor at each sampling time in each monitoring cycle that exceed the preset voltage threshold, and use this as the total number of acoustic emission events. Based on the total number of acoustic emission events and the total number of sampling times in each monitoring cycle, the occurrence rate of acoustic emission events in each monitoring cycle is calculated. The mean value of the voltage signal corresponding to each acoustic emission event in the corresponding monitoring cycle is used as the acoustic emission event intensity evaluation value in the corresponding monitoring cycle. The second-class hydrogen embrittlement vector element acquisition unit is used to calculate the root mean square value and kurtosis of the time-series composite acceleration amplitude sequence within each monitoring period based on the composite acceleration amplitude detected by the triaxial vibration accelerometer at each sampling time within each monitoring period, and obtain the vibration root mean square value and vibration kurtosis of the corresponding monitoring period. The hydrogen embrittlement state vector generation unit is used to arrange the acoustic emission event occurrence rate, acoustic emission event intensity assessment value, vibration root mean square value and vibration kurtosis in order for each monitoring period to form a hydrogen embrittlement state vector.

5. The online sealing testing device for hydrogen-powered ball valves according to claim 1, characterized in that: The sealing condition diagnostic analysis module includes: The feature space construction submodule is used to construct the ball valve preset sealing health three-dimensional feature space based on the internal leakage state vector, external leakage state vector and hydrogen embrittlement state vector corresponding to all historical normal sealing monitoring cycles. The state coordinate point determination submodule is used to project the internal leakage state vector, external leakage state vector, and hydrogen embrittlement state vector of each hydrogen ball valve in the current monitoring cycle onto the internal leakage principal component vector, external leakage principal component vector, and hydrogen embrittlement principal component vector, respectively, to obtain the internal leakage scalar value x, external leakage scalar value y, and hydrogen embrittlement scalar value z, and use (x, y, z) as the state coordinate point P of the current monitoring cycle; The health deviation calculation submodule is used to calculate the Mahalanobis distance between the state coordinate point P of the current monitoring period and the cluster Q of state coordinate points of all historical normal monitoring periods. And use it as the seal health deviation for the current monitoring cycle; The status assessment result output submodule is used to compare the seal health deviation of the current monitoring period with the preset seal health deviation threshold range and output the corresponding seal health status level, which includes healthy, alert, warning and abnormal.

6. The online sealing testing device for hydrogen energy ball valves according to claim 5, characterized in that: Feature space The submodules include: The analysis matrix determination unit is used to construct analysis matrix one, analysis matrix two and analysis matrix three based on the internal leakage state vector, external leakage state vector and hydrogen embrittlement state vector corresponding to all historical normal sealing monitoring cycles; Principal component vector one determination unit is used to perform principal component analysis on analysis matrix one to obtain the principal component vector of analysis matrix one, and use it as principal component vector one; The principal component vector two determination unit is used to perform principal component analysis on analysis matrix two to obtain the principal component vector of analysis matrix two, and use it as the principal component vector two. The principal component vector three determination unit is used to perform principal component analysis on the analysis matrix three to obtain the principal component vector of the analysis matrix three, and use it as the principal component vector three. The feature space construction unit is used to orthogonalize the principal component vectors 1, 2, and 3 to form the internal leakage principal component vector, external leakage principal component vector, and hydrogen embrittlement principal component vector. Based on the internal leakage principal component vector, external leakage principal component vector, and hydrogen embrittlement principal component vector, the ball valve preset sealing health three-dimensional feature space is constructed.

7. The online sealing testing device for a hydrogen-powered ball valve according to claim 5, characterized in that: The seal failure cause analysis module includes: The elementary failure mode spectrum library construction submodule is used to construct three mutually orthogonal elementary failure mode vectors, including elementary vector one representing pure internal leakage of the sealing part, elementary vector two representing pure external leakage of the packing part, and elementary vector three representing hydrogen embrittlement of the pure sealing part. The space composed of elementary vector one, elementary vector two, and elementary vector three is used as the elementary space. The deviation vector acquisition submodule takes the difference between the state coordinate point P corresponding to the monitoring cycle with an abnormal sealing health status level and the center point O of the cluster Q of state coordinate points from all historical normal monitoring cycles as the deviation vector for the current abnormal monitoring cycle. ; The failure mode determination submodule will determine the deviation vector of the current anomaly monitoring cycle. Projecting onto the primitive space, calculate the deviation vector for the current anomaly monitoring period. Projection coefficients are plotted on primitive vector one, primitive vector two, and primitive vector three respectively, and each projection coefficient is normalized to obtain the contribution percentage of each primitive failure mode at the state coordinate point P corresponding to the current anomaly monitoring cycle. The dominant failure type is determined based on the primitive failure mode with the largest contribution percentage, and a complete sealing failure contribution spectrum is output.

8. The online sealing testing device for hydrogen energy ball valves according to claim 1, characterized in that: The comprehensive hazard assessment module includes: The Hazardous Heat Map Construction Submodule is used to create a 3D layout map based on hydrogen energy facilities, mark the precise installation location of each hydrogen ball valve, construct an initial hydrogen ball valve distribution map, and divide the hydrogen ball valve distribution map into several consecutive assessment areas. The regional risk value calculation submodule is used to calculate the comprehensive risk thermal value of each assessment area based on the sealing health status level of each hydrogen ball valve in the current monitoring cycle, and based on the real-time risk values ​​of all hydrogen ball valves corresponding to each assessment area. The risk assessment visualization submodule is used to generate a dynamic hazard heat map by rendering it on the hydrogen energy ball valve distribution map with a color gradient from cool to warm colors based on the comprehensive risk heat value of each assessment area. The risk avoidance path generation module is used to automatically generate and display the optimal maintenance risk avoidance path to an assessment area when the comprehensive risk heat value of an assessment area exceeds a preset high-risk threshold.

Images