Hydrogen valve leakage intelligent monitoring method and system based on multiple sensing parameters

Abstract

The application discloses a hydrogen valve leakage intelligent monitoring method and system based on multiple sensing parameters and relates to the technical field of intelligent sensing. The method comprises the following steps: simulating and constructing a hydrogen valve twin model; collecting pressure, temperature and hydrogen concentration in real time through a sensor to obtain a pressure sequence, a temperature sequence and a hydrogen concentration sequence in a historical monitoring adjustment time window; calculating a pressure-temperature joint change rate sequence and a hydrogen concentration difference sequence, performing emergency evaluation, determining a leakage monitoring emergency index, wherein the fitting weight proportion is obtained according to the hydrogen concentration difference sequence to adjust the benchmark weight set by the analytic hierarchy process; querying an emergency index-monitoring parameter group correlation table to determine a fitting monitoring parameter group, combining the fitting weight proportion to generate a fitting monitoring scheme, and performing monitoring. The application solves the technical problems that it is difficult to distinguish normal penetration from abnormal leakage in the prior art, that misjudgment and missed judgment are caused by the failure to fuse pressure and temperature parameters, and that the monitoring strategy lacks self-adaptive adjustment capability.

Description

Technical Field

[0001] This invention relates to the field of intelligent sensing technology, specifically to an intelligent monitoring method and system for hydrogen valve leakage based on multiple sensing parameters. Background Technology

[0002] In industrial hydrogen pipeline systems, high-pressure shut-off valves are crucial devices for controlling the flow and regulating the hydrogen medium. The moving seal area of ​​the stuffing box in the shut-off valve is located in the relatively moving part between the valve stem and the packing. This area is subjected to the coupling effect of alternating stress and medium pressure during valve opening and closing, making it a high-risk location for hydrogen leakage. Hydrogen molecules have extremely small kinetic diameters and extremely strong penetration and diffusion capabilities, making them highly susceptible to leakage through the microscopic gaps in the moving seal area of ​​the stuffing box under high-pressure conditions. Once a leak occurs in the moving seal area of ​​the stuffing box, the accumulated hydrogen, after mixing with air in the confined space, can, if it reaches the explosive limit, ignite upon contact with an ignition source, triggering a deflagration or detonation accident, posing a serious threat to personnel safety and equipment property. Therefore, real-time and accurate leakage monitoring of the moving seal area of ​​the stuffing box is a critical link in ensuring the safe operation of hydrogen pipelines.

[0003] Current hydrogen leak monitoring technologies primarily employ single-sensor detection methods, such as deploying fixed hydrogen concentration detectors near valves or using electrochemical sensors for point-based measurements. These methods have several shortcomings in practical applications. First, single-concentration monitoring cannot distinguish between normal permeation and abnormal leaks caused by degradation of the packing seal performance. Even under normal operating conditions, trace amounts of hydrogen permeate the moving seal area of ​​the packing gland, making it difficult to determine whether the seal has deteriorated based solely on the absolute concentration value. Second, pressure and temperature fluctuations significantly affect the sealing performance of the packing and the diffusion behavior of hydrogen, but existing methods do not incorporate pressure and temperature parameters into a comprehensive assessment system, leading to potential misjudgments or missed detections under conditions of sudden pressure changes or temperature increases. Third, traditional monitoring methods use fixed sampling frequencies and uniform monitoring parameter sets, failing to adaptively adjust monitoring strategies based on dynamic changes in leak risk. Under normal sealing conditions, high-frequency, multi-parameter monitoring consumes significant computational resources and communication bandwidth; however, when the leak risk increases, the response speed and monitoring sensitivity become insufficient.

[0004] Therefore, there is an urgent need for an intelligent monitoring method for hydrogen valve leakage that can integrate multiple sensor parameters and dynamically adjust the monitoring strategy. Summary of the Invention

[0005] This invention addresses the technical problems of existing technologies by providing a method and system for intelligent monitoring of hydrogen valve leakage based on multiple sensing parameters.

[0006] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: In a first aspect, the present invention provides an intelligent monitoring method for hydrogen valve leakage based on multiple sensing parameters, including: Simulations were used to construct a hydrogen valve twin model to predict the current hydrogen concentration in the stuffing gland dynamic sealing zone based on pressure and temperature parameters. During the operation of the high-pressure shut-off valve, the pressure, temperature and hydrogen concentration of the packing gland dynamic sealing area are collected in real time by sensors to obtain the pressure sequence, temperature sequence and hydrogen concentration sequence within the historical monitoring and adjustment time window. The pressure-temperature joint change rate sequence is calculated based on the pressure sequence and temperature sequence, and the hydrogen concentration difference sequence is calculated based on the hydrogen concentration sequence and the predicted hydrogen concentration sequence obtained by simulation through the hydrogen valve twin model. Based on the pressure sequence, temperature sequence, hydrogen concentration sequence, pressure-temperature combined rate of change sequence, and hydrogen concentration difference sequence, the urgency assessment of hydrogen leak monitoring is conducted to determine the leak monitoring urgency index. Specifically, the baseline weights set in advance using the analytic hierarchy process are adjusted according to the hydrogen concentration difference sequence to obtain the adaptation weight ratio, and the urgency assessment is conducted based on the adaptation weight ratio. Based on the leak monitoring emergency index, query the pre-built emergency index-monitoring parameter group association table to determine the appropriate monitoring parameter group, generate an appropriate monitoring scheme based on the appropriate weight ratio, and execute hydrogen leak monitoring of the stuffing gland dynamic sealing area within the next monitoring adjustment time window.

[0007] Secondly, the present invention provides an intelligent monitoring system for hydrogen valve leakage based on multiple sensing parameters, comprising: The model simulation module is used to simulate and build a hydrogen valve twin model for predicting the current hydrogen concentration in the stuffing gland dynamic sealing area based on pressure and temperature parameters. The data acquisition module is used to collect the pressure, temperature and hydrogen concentration of the packing gland moving sealing area in real time through sensors during the operation of the high-pressure shut-off valve, and obtain the pressure sequence, temperature sequence and hydrogen concentration sequence within the historical monitoring and adjustment time window; The sequence calculation module is used to calculate the pressure-temperature joint change rate sequence based on the pressure sequence and temperature sequence, and to calculate the hydrogen concentration difference sequence based on the hydrogen concentration sequence and the predicted hydrogen concentration sequence obtained by simulation through the hydrogen valve twin model. The emergency assessment module is used to assess the urgency of hydrogen leak monitoring based on the pressure sequence, temperature sequence, hydrogen concentration sequence, pressure-temperature combined rate of change sequence, and hydrogen concentration difference sequence, and to determine the leak monitoring urgency index. Specifically, the module adjusts the baseline weights set in advance using the analytic hierarchy process based on the hydrogen concentration difference sequence to obtain the appropriate weight ratio, and performs the urgency assessment based on the appropriate weight ratio. The scheme generation module is used to query the pre-built emergency index-monitoring parameter group association table according to the leakage monitoring emergency index, determine the appropriate monitoring parameter group, generate an appropriate monitoring scheme in combination with the appropriate weight ratio, and execute hydrogen leakage monitoring of the stuffing gland dynamic sealing area in the next monitoring adjustment time window.

[0008] The beneficial effects of this invention are: Compared to existing technologies, this invention first utilizes a hydrogen valve twin model to establish a mapping relationship between pressure, temperature, and hydrogen concentration under normal sealing conditions, providing a dynamic baseline for leak detection and effectively distinguishing between normal seepage and abnormal leaks. Secondly, it introduces a pressure-temperature joint change rate to characterize the impact of sudden changes in operating conditions on sealing performance, and introduces a hydrogen concentration difference to characterize the deviation between the measured concentration and the predicted normal concentration, constructing a multi-dimensional monitoring index system containing ten indicators. Thirdly, it employs the analytic hierarchy process (AHP) to set the baseline weights for each indicator and calculates the leakage deviation coefficient based on the hydrogen concentration difference sequence, dynamically adjusting the weights of concentration-related indicators to adaptively focus the urgency assessment results towards leakage-sensitive indicators. Finally, it matches monitoring parameter groups of different granularities according to the urgency index; the lower the urgency, the more comprehensive the monitoring parameters to facilitate trend analysis, while the higher the urgency, the more concise the monitoring parameters to ensure response speed. This invention achieves adaptive adjustment of the monitoring strategy under operating conditions, optimizing the allocation of computational resources while ensuring timely alarm response. Attached Figure Description

[0009] Figure 1 A flowchart illustrating the intelligent monitoring method for hydrogen valve leakage based on multiple sensing parameters provided by this invention; Figure 2 A schematic diagram of the intelligent monitoring method for hydrogen valve leakage based on multiple sensing parameters provided by the present invention; Figure 3 This is a schematic diagram of the structure of the intelligent monitoring system for hydrogen valve leakage based on multiple sensing parameters provided by the present invention.

[0010] In the attached diagram, the components represented by each number are as follows: Model simulation module 11, data acquisition module 12, sequence calculation module 13, emergency assessment module 14, and scheme generation module 15. Detailed Implementation

[0011] Example 1, as Figure 1 , Figure 2 As shown, this embodiment of the invention provides an intelligent monitoring method for hydrogen valve leakage based on multiple sensing parameters, including: S10: Simulation builds a hydrogen valve twin model to predict the current hydrogen concentration in the stuffing box dynamic sealing zone based on pressure and temperature parameters; In industrial hydrogen pipeline systems, the dynamic sealing zone of the stuffing box in a high-pressure shut-off valve refers to the annular cavity where the valve stem exits the valve body, used to install the sealing packing. In this zone, the valve stem needs to move axially up and down during valve opening and closing, constituting a dynamic seal structure. Compared to static seals, dynamic seals are subjected to alternating stress and continuous friction during operation, making the packing more prone to wear, loosening, and aging, leading to a progressive failure of the sealing performance. In the initial stage of good sealing performance, the packing compression is sufficient to block the diffusion path of hydrogen molecules along the valve stem surface, resulting in only a trace amount of permeation as designed. As service time increases, the packing gradually loosens or wears, the effective contact pressure at the sealing interface decreases, and hydrogen leakage develops from trace permeation to a small leak. When the sealing performance deteriorates severely, the gap between the packing and the valve stem continues to widen, eventually leading to a large leak. Therefore, the key to hydrogen leak monitoring lies in accurately identifying early deterioration trends in sealing performance through measurable physical parameters, issuing warnings before the leakage reaches dangerous levels.

[0012] The sealing performance of the dynamic sealing zone of the stuffing box is mainly affected by two external factors: medium pressure and ambient temperature. When the medium pressure increases, the axial compressive force on the packing increases, which theoretically improves the sealing effect. However, excessive pressure may cause plastic deformation or extrusion of the packing, thus accelerating seal failure. When the temperature rises, the difference in thermal expansion coefficient between the packing material and the valve stem and body materials may cause changes in the contact pressure distribution at the sealing interface. Simultaneously, high temperatures accelerate the chemical aging process of the packing material, reducing its elastic recovery ability and causing permanent compression deformation, making it impossible to maintain the initial sealing pressure. Therefore, the absolute value of the hydrogen concentration at the current moment is insufficient to determine whether the leakage is caused by normal aging of the packing or abnormal deterioration of the sealing performance; a baseline mapping relationship between pressure, temperature, and hydrogen concentration must be established.

[0013] This step constructs a hydrogen valve twin model using simulation technology. This hydrogen valve twin model is a data-driven fusion model based on physical mechanisms, used to simulate the hydrogen permeation and diffusion behavior in the moving seal area of ​​the stuffing box under different pressure and temperature conditions. In subsequent steps, the real-time collected pressure and temperature parameters are input into this hydrogen valve twin model to obtain the predicted hydrogen concentration corresponding to the normal sealing state under the current operating conditions. This prediction is then compared with the actual measured hydrogen concentration to separate the normal permeation component from the measured concentration, extracting the leakage characteristic quantity caused by the deterioration of sealing performance.

[0014] Specifically, in this embodiment, the simulation constructs a hydrogen valve twin model for predicting the current hydrogen concentration in the moving seal area of ​​the stuffing box based on pressure and temperature parameters, including: Simulation modeling was performed based on the structural parameters, material properties, and physicochemical characteristics of hydrogen in the dynamic sealing zone of the stuffing box of the high-pressure shut-off valve, and an initial twin model of the dynamic sealing zone of the stuffing box was established. Historical pressure data, historical temperature data, and corresponding historical hydrogen concentration data of the dynamic sealing zone of the high-pressure shut-off valve packing gland under different operating conditions were collected to construct a model training sample set. The initial twin model is trained using the model training sample set, so that the initial twin model learns the mapping relationship between pressure, temperature and hydrogen concentration in the stuffing box dynamic sealing area under normal sealing conditions. When the prediction error meets the preset accuracy threshold, the construction of the hydrogen valve twin model is completed. During the operation of the high-pressure shut-off valve, the real-time collected pressure and temperature parameters are input into the hydrogen valve twin model, and the predicted value of the current hydrogen concentration in the stuffing gland dynamic sealing area is output.

[0015] First, a simulation model of the moving sealing zone of the stuffing box in the high-pressure shut-off valve is established based on the structural parameters, material properties, and physicochemical characteristics of hydrogen. Specifically, in an optional embodiment, the structural parameters include the geometric dimensions of the packing groove, the number and thickness of the packing layers, and the valve stem diameter. These structural parameters determine the contact area of ​​the sealing interface and the geometric characteristics of the sealing gap. Material properties include the elastic modulus, Poisson's ratio, coefficient of thermal expansion, permeability coefficient of the packing, and the surface roughness of the valve stem and valve body materials. These material properties determine the deformation behavior of the packing under pressure loads and temperature, as well as the diffusion path of hydrogen molecules along the sealing interface. The physicochemical characteristics of hydrogen include the molecular dynamics diameter, viscosity, and diffusion coefficient of hydrogen. These physicochemical characteristics determine the permeation and diffusion rates of hydrogen under different pressure and temperature gradients.

[0016] Optionally, the initial twin model is constructed based on finite element analysis or multiphysics coupling simulation methods to simulate the contact pressure distribution at the interface between the packing and the valve stem, as well as the diffusion flux of hydrogen molecules along the sealing interface, under given pressure load and temperature boundary conditions, and output the steady-state permeation concentration.

[0017] In an optional embodiment, taking finite element analysis as an example, an axisymmetric geometric model of the dynamic sealing zone of the stuffing box is first established, including components such as the packing groove, packing ring, valve stem, and packing gland. The packing material is defined as a hyperelastic constitutive model, such as the Mooney-Rivlin model or the Ogden model, to describe the stress-strain relationship of the packing under large deformation conditions. To describe the stress relaxation and permanent compressive deformation that may occur in the packing material during long-term service, viscoelastic or creep components can be introduced into the constitutive model. The relevant viscoelastic parameters are used as calibration parameters for subsequent data-driven calibration, enabling the model to fit the long-term sealing performance degradation trend. The valve stem and packing gland are defined as linear elastic materials. During the mesh generation stage, the area near the contact interface between the packing and the valve stem is locally meshed, with a mesh size one-third that of the area far from the contact region, to improve the calculation accuracy of the contact pressure distribution. The applied loads and boundary conditions include: applying an axial preload, calculated from the bolt tightening torque, to the upper surface of the packing gland; applying a medium pressure load to the outer surface of the valve stem, with the pressure value set according to actual operating conditions; applying convective heat transfer boundary conditions to the outer wall of the stuffing box, with the heat transfer coefficient set according to ambient temperature and wind speed; and applying temperature degrees of freedom to each node of the packing ring, with the initial temperature set according to ambient temperature. The solution type is steady-state analysis or transient analysis, and the governing equations include the mechanical equilibrium equation, the heat conduction equation, and the mass conservation equation.

[0018] After solving, the post-processing extracts the normal contact pressure distribution at the interface between the packing and the valve stem, and integrates it along the length of the interface to obtain the total contact force. The effective sealing specific pressure is calculated based on the difference between the total contact force and the medium pressure, used to evaluate the sealing capacity of the sealing interface. Simultaneously, the contact pressure distribution at the interface is mapped to a local permeability distribution function using an exponential permeability-pressure empirical model. That is, the local permeability equals the initial permeability multiplied by an exponent raised to the power of e, where the exponent is the negative contact pressure divided by the reference pressure. This model reflects the physical law that the greater the contact pressure, the smaller the sealing gap, and the exponentially decreasing the permeability. Furthermore, combining the diffusion coefficient of hydrogen in the packing material, the diffusion flux of hydrogen molecules along the sealing interface is calculated. Integrating the diffusion flux along the outlet boundary area yields the steady-state permeation concentration, which is used as the output of the initial twin model.

[0019] It is important to note that key parameters in the initial twin model can be preliminarily set based on theoretical formulas or empirical values ​​from literature. For example, the hyperelastic constitutive model parameters of the packing can be obtained by fitting standard data from uniaxial tensile tests; the friction coefficient of the packing-valve stem interface can be set by consulting mechanical design manuals based on lubrication conditions and surface roughness, for example, 0.05 for good lubrication and 0.15 for dry friction; the hydrogen permeability coefficient of the packing can be obtained by consulting material databases based on the polymer type of the packing. The initial permeability and reference pressure can be preliminarily estimated based on the initial geometric clearance and material hardness of the sealing structure. The initial permeability can be determined based on hydrogen permeation experimental data under no-pressure conditions, and the reference pressure can be set according to the yield strength order of the packing material. The above preliminarily set parameters will be subsequently calibrated and optimized using the model training sample set to transform the initial twin model into a high-precision hydrogen valve twin model. Using the multiphysics coupling simulation method described above, the initial twin model can simultaneously consider the mechanical effect of pressure load on the contact pressure of the sealing interface and the thermal effect of temperature field on the mechanical properties of the packing material and the hydrogen diffusion coefficient, providing an initial state with a clear physical mechanism for the establishment of the hydrogen valve twin model.

[0020] Furthermore, historical pressure data, historical temperature data, and corresponding historical hydrogen concentration data of the moving seal area of ​​the high-pressure shut-off valve's stuffing box under different operating conditions were collected to construct a model training sample set. During the sealing test phase before the valve leaves the factory or the break-in phase during the initial operation of a newly installed valve, the medium pressure and ambient temperature of the moving seal area of ​​the stuffing box were collected using pressure and temperature sensors, respectively, while the hydrogen concentration value under the corresponding operating conditions was collected using a hydrogen concentration sensor. The data collection process covered the normal operating pressure range that the valve might experience, such as pressure ranges from 0 MPa to the valve's rated operating pressure, and temperature ranges from ambient low temperature to the valve's maximum allowable operating temperature. The collected data together constitute the model training sample set. Each model training sample contains a pressure value, a temperature value, and the measured hydrogen concentration value under that pressure and temperature combination. This hydrogen concentration value reflects the baseline level of hydrogen concentration in the moving seal area of ​​the stuffing box under normal sealing conditions, due to the permeation channels allowed by the design.

[0021] Furthermore, the initial twin model established above is trained using the model training sample set. This allows the initial twin model to learn the mapping relationship between pressure, temperature, and hydrogen concentration in the moving sealing area of ​​the packing gland under normal sealing conditions. When the prediction error meets the preset accuracy threshold, the construction of the hydrogen valve twin model is complete. Specifically, the training process adopts a supervised learning approach, using the pressure and temperature values ​​in the model training samples as input features and the corresponding hydrogen concentration values ​​as the supervision target to calibrate and optimize the key parameters in the initial twin model. The initial twin model contains several empirical coefficients that need to be calibrated, such as the friction coefficient of the packing-stem contact interface and the stress relaxation coefficient of the packing material. These coefficients cannot be accurately obtained from theoretical calculations and need to be identified through a data-driven approach.

[0022] In one optional embodiment, the training process is illustrated using a reduced-order modeling and particle swarm optimization algorithm as an example: The first stage involves constructing a reduced-order model. Since the initial twin model is a multi-physics coupled finite element model, the computational cost of a single solution is too high, making direct embedding into particle swarm optimization iterations impossible. Therefore, the initial twin model is first used to perform sparse sampling in the parameter space, generating a set of simulation samples covering the boundaries of the ranges of the empirical coefficients to be calibrated. Each sample corresponds to a set of empirical coefficient combinations, and the corresponding predicted hydrogen concentration value is output through finite element simulation. Using intrinsic orthogonal decomposition or radial basis function response surface methods, with the input parameters of the above samples as independent variables and the corresponding simulation output as the dependent variable, a reduced-order model is constructed. This reduced-order model can output the predicted hydrogen concentration value under a given combination of empirical coefficients with minimal computational cost, and the error is controlled within an acceptable engineering range.

[0023] In the second stage, all empirical coefficients to be calibrated in the initial twin model are combined into a multi-dimensional vector. The upper and lower boundaries of each dimension of this vector are set according to physical constraints or empirical values ​​from literature; for example, the range of the friction coefficient is set to 0.05 to 0.25, and the range of the stress relaxation coefficient is set to 0.8 to 1.2. A particle swarm containing multiple particles is initialized, with each particle representing a candidate value for a set of empirical coefficients to be calibrated. For each particle, its represented empirical coefficient is substituted into the reduced-order model, and simulation calculations are performed on each sample in the training sample set. The pressure and temperature values ​​of the sample are input, and the corresponding predicted hydrogen concentration value is output. The root mean square error (RMSE) between the predicted hydrogen concentration and the measured hydrogen concentration of all training samples under that particle is calculated and used as the fitness value of that particle. The smaller the RMSE, the better the combination of empirical coefficients corresponding to that particle.

[0024] In each iteration, the individual optimal position is updated based on the current fitness value of each particle, and the global optimal position is updated based on the individual optimal positions of all particles. Then, the position of each particle is adjusted according to the particle swarm velocity update formula. The velocity update formula is: New velocity = Inertia weight × Original velocity + Individual learning factor × Random number 1 × (Individual optimal position - Current position) + Social learning factor × Random number 2 × (Global optimal position - Current position). The position update formula is: New position = Original position + New velocity. Here, the inertia weight controls the degree to which the particle maintains its original motion trend, the individual learning factor controls the strength of the particle's convergence towards its own historical optimal position, the social learning factor controls the strength of the particle's convergence towards the swarm's global optimal position, and random number 1 and random number 2 are uniformly distributed random numbers between 0 and 1, used to increase the randomness of particle motion to avoid getting trapped in local optima. For example, the inertia weight can be set to 0.7, the individual learning factor can be set to 1.5, the social learning factor can be set to 1.5, the maximum number of iterations can be set to 200, and the iteration is terminated early when the global optimal fitness value no longer decreases for 20 consecutive iterations.

[0025] After the iteration terminates, the combination of empirical coefficients corresponding to the globally optimal position is used as the optimal parameter configuration for the initial twin model. This initial twin model is the trained hydrogen valve twin model. Through the above training process, the empirical coefficients in the hydrogen valve twin model are calibrated based on actual data, enabling the model to accurately describe the nonlinear mapping relationship between pressure, temperature, and hydrogen concentration in the stuffing box moving seal area under normal sealing conditions. After training, the obtained hydrogen valve twin model can predict the hydrogen concentration corresponding to the normal sealing state under untrained pressure and temperature combinations.

[0026] The preset accuracy threshold is the upper limit of the allowable deviation between the predicted and measured values ​​of the hydrogen valve twin model. It is used to determine whether the model's prediction accuracy meets the requirements of engineering applications. It can be set according to actual engineering requirements; for example, it can be set to ensure that the average absolute percentage error between the predicted and measured hydrogen concentrations does not exceed 5%. When the model's prediction error on the validation set is lower than this preset accuracy threshold, the hydrogen valve twin model is deemed to meet the usage requirements, and the construction is completed.

[0027] Specifically, in practical use, during the operation of the high-pressure shut-off valve, the real-time collected pressure and temperature parameters are input into the hydrogen valve twin model, outputting a predicted value of the current hydrogen concentration in the moving seal area of ​​the stuffing box. This predicted value represents the hydrogen concentration level that the moving seal area of ​​the stuffing box should exhibit under the current pressure and temperature conditions to maintain a normal sealing state, and can serve as a dynamic baseline for subsequent leakage judgment. When the real-time monitored measured hydrogen concentration is significantly higher than this predicted value, it indicates that in addition to the design-allowed permeation component, the measured concentration also includes a leakage component caused by the deterioration of sealing performance, which can be used to identify abnormal leakage trends in the moving seal area of ​​the stuffing box.

[0028] In summary, through the construction and application of the hydrogen valve twin model, the function of separating the normal permeation component from the measured hydrogen concentration was realized, providing a dynamic benchmark that adapts to the operating conditions for leak detection, and avoiding the problem of frequent false alarms due to pressure or temperature fluctuations causing fixed concentration thresholds.

[0029] S20: During the operation of the high-pressure shut-off valve, the pressure, temperature and hydrogen concentration of the packing gland dynamic sealing area are collected in real time by sensors to obtain the pressure sequence, temperature sequence and hydrogen concentration sequence within the historical monitoring and adjustment time window. Furthermore, after the hydrogen valve twin model is completed and put into use, it is necessary to collect key physical parameters of the moving seal area of ​​the stuffing box in real time during the continuous operation of the high-pressure shut-off valve, providing raw data for subsequent leakage monitoring and emergency assessment. Specifically, the execution steps of this embodiment involve real-time collection of pressure, temperature, and hydrogen concentration in the moving seal area of ​​the stuffing box using sensors, including: A pressure sensor is installed at the medium inlet of the dynamic sealing zone of the stuffing box to collect the medium pressure in the dynamic sealing zone of the stuffing box in real time. Temperature sensors are installed on the outer wall of the moving sealing area of ​​the stuffing box to collect the ambient temperature of the moving sealing area of ​​the stuffing box in real time. A MEMS hydrogen concentration sensor is installed at the contact point between the valve stem and the packing in the moving sealing area of ​​the stuffing box to collect the local hydrogen concentration in the moving sealing area of ​​the stuffing box in real time.

[0030] The dynamic sealing area of ​​the stuffing box is subjected to high-frequency pressure fluctuations and temperature changes during operation. Instantaneous measurements at a single moment are easily affected by measurement noise or random disturbances, failing to accurately reflect the changing trend of the sealing state. Therefore, this step introduces the concept of a historical monitoring and adjustment time window. This involves taking the current moment as the endpoint and extracting a fixed-length time interval before it. Within this time interval, data from multiple sampling points are continuously collected to form a time sequence. The length of the historical monitoring and adjustment time window can be set comprehensively based on the frequency of the valve's working cycle and the response speed requirements of the leakage monitoring system; for example, it can be set to 5 minutes or 10 minutes. The sampling interval can be set according to the sensor's sampling rate; for example, collecting one data point every 10 seconds would result in a 5-minute time window containing 30 data points each for the pressure, temperature, and hydrogen concentration sequences. By accumulating multiple sets of data within the historical monitoring and adjustment time window, the statistical characteristics of each physical quantity, such as the mean and fluctuation coefficient, can be calculated. Simultaneously, the changing trends of parameters can be identified, avoiding false alarms triggered by abnormal single-point measurements.

[0031] The pressure sensor is used to acquire the medium pressure in the moving seal area of ​​the stuffing box in real time. This pressure sensor is installed at the medium inlet of the moving seal area of ​​the stuffing box and directly measures the medium pressure acting on the packing. The medium inlet refers to the channel where high-pressure hydrogen medium enters the stuffing box cavity from the main flow channel of the valve body. The medium pressure at this location is directly related to the axial load on the packing. Increased pressure leads to increased axial clamping force on the packing and increases the pressure difference across the sealing interface, affecting the driving force for hydrogen diffusion along the sealing interface.

[0032] A temperature sensor is used to collect the ambient temperature of the dynamic sealing area of ​​the stuffing box in real time. This temperature sensor is installed on the outer wall surface of the dynamic sealing area of ​​the stuffing box to measure the actual operating temperature of the packing. The outer wall surface of the stuffing box is the thermal interface between the stuffing box cavity and the external environment, and its temperature reflects the actual thermal load borne by the packing material. The temperature sensor is installed close to the outer wall surface of the stuffing box, and the measured temperature value is used to characterize the current operating temperature of the packing. Increased temperature accelerates the chemical aging of the packing material, reduces its elastic recovery ability, and also changes the hydrogen diffusion coefficient, affecting the normal permeation concentration level.

[0033] In addition, a hydrogen concentration sensor is used to acquire the local hydrogen concentration in the moving seal area of ​​the stuffing box in real time. This hydrogen concentration sensor is positioned near the contact point between the valve stem and the packing, employing a MEMS miniature hydrogen concentration sensing chip, capable of detecting minute changes in hydrogen concentration. The MEMS hydrogen concentration sensor is a miniaturized hydrogen detection element manufactured based on microelectromechanical systems (MEMS) technology, using palladium alloy thin-film resistors or catalytic combustion principles, capable of detecting trace amounts of hydrogen at concentrations of one part per million. The key advantages of this sensor are its small size, fast response, and low power consumption, allowing it to be embedded in the narrow space near the contact point between the valve stem and the packing in the moving seal area of ​​the stuffing box. The sensor probe directly faces the sealing gap between the packing and the valve stem; when hydrogen escapes outward along this gap, the sensor can immediately capture changes in the concentration of leaked hydrogen.

[0034] Specifically, by deploying the three sets of sensors in a coordinated manner, synchronous, in-situ, and real-time acquisition of pressure, temperature, and hydrogen concentration can be achieved. The three sensors acquire data synchronously at the same sampling frequency, ensuring that the pressure, temperature, and hydrogen concentration data at the same moment correspond to each other in time.

[0035] Finally, through the above real-time acquisition process, the pressure sequence P within the historical monitoring and adjustment time window can be obtained. (t) Temperature sequence T (t) and hydrogen concentration sequence C (t) , where t represents the sampling time sequence number within the time window. The above time series contains information on the evolution of the operating conditions of the dynamic sealing area of ​​the stuffing box over a recent period, providing basic data for subsequent calculations of the combined pressure-temperature change rate and hydrogen concentration difference.

[0036] S30: Calculate the pressure-temperature joint change rate sequence based on the pressure sequence and temperature sequence, and calculate the hydrogen concentration difference sequence based on the hydrogen concentration sequence and the predicted hydrogen concentration sequence obtained by simulation through the hydrogen valve twin model. Furthermore, after obtaining the pressure, temperature, and hydrogen concentration sequences within the historical monitoring and adjustment time window, feature extraction needs to be performed on these raw sequences to transform them into derived feature sequences that can characterize changes in the sealing state. The raw values ​​in the pressure and temperature sequences reflect the operating conditions, but the sealing performance of the stuffing box dynamic sealing zone depends not only on the absolute values ​​of pressure and temperature at the current moment, but also on the rate of change of pressure and temperature. When the pressure or temperature rises rapidly, the contact interface between the packing and the valve stem is in a non-equilibrium state. The contact pressure distribution at the sealing interface and the creep response of the packing lag behind the changes in operating conditions. Under these conditions, the risk of leakage is significantly higher than under steady-state conditions.

[0037] Therefore, this step first extracts the pressure change rate and temperature change rate from the pressure and temperature sequences, and then weights and fuses them into a pressure-temperature joint change rate sequence, which is used to quantify the impact intensity of the operating conditions. The pressure change rate reflects the severity of fluctuations in medium pressure, and the temperature change rate reflects the severity of fluctuations in thermal load. The weighted fusion of the two can comprehensively characterize the impact effect of sudden changes in operating conditions on the sealing interface.

[0038] Simultaneously, the pressure values ​​at each moment in the pressure sequence and the temperature values ​​at each moment in the temperature sequence are input into the hydrogen valve twin model. This hydrogen valve twin model, based on the pressure-temperature-hydrogen concentration mapping relationship under normal sealing conditions, outputs the predicted hydrogen concentration value for each moment, forming a predicted hydrogen concentration sequence. This predicted hydrogen concentration sequence represents the hydrogen concentration level that the moving seal area of ​​the stuffing box should exhibit while maintaining a normal sealing state under the current pressure and temperature conditions. The difference between the measured hydrogen concentration sequence and the predicted hydrogen concentration sequence is calculated moment by moment to obtain the hydrogen concentration difference sequence.

[0039] Specifically, the hydrogen concentration difference reflects the portion of the measured hydrogen concentration that exceeds the normal permeability baseline, i.e., the leakage component caused by deterioration of sealing performance. When the hydrogen concentration difference continues to increase, it indicates that the sealing gap between the packing and the valve stem is constantly widening, and the leakage is continuously rising; when the hydrogen concentration difference fluctuates greatly, it indicates that the sealing performance is unstable and there is intermittent leakage. This hydrogen concentration difference sequence is the core basis for dynamically adjusting the index weights and determining the leakage development trend in subsequent emergency assessments.

[0040] Specifically, in this embodiment, a pressure-temperature joint rate of change sequence is calculated based on the pressure sequence and temperature sequence, and a hydrogen concentration difference sequence is calculated based on the hydrogen concentration sequence and the predicted hydrogen concentration sequence obtained through simulation using the hydrogen valve twin model, including: Calculate the rate of change of pressure and the rate of change of temperature at each time step for the pressure and temperature sequences, respectively, to obtain the pressure rate of change sequence and the temperature rate of change sequence. Based on the pressure change rate sequence and temperature change rate sequence, the pressure change rate and temperature change rate at each time are weighted and fused to obtain the pressure-temperature joint change rate at each time, which constitutes the pressure-temperature joint change rate sequence. The weights of the pressure change rate and temperature change rate are pre-calibrated based on the degree of influence of the pressure change rate and temperature change rate on the sealing failure of the stuffing box dynamic sealing area in historical working condition data. The hydrogen concentration difference at each moment is obtained by subtracting the predicted hydrogen concentration sequence from the hydrogen concentration sequence at each moment, thus forming a hydrogen concentration difference sequence.

[0041] First, calculate the rate of change of pressure and the rate of change of temperature at each moment in the pressure and temperature sequences, respectively, to obtain the pressure rate of change sequence and the temperature rate of change sequence. The method for calculating the rate of change of pressure is as follows: for the i-th sampling point in the pressure sequence, take the pressure value at the previous moment, calculate the pressure difference, and divide the pressure difference by the sampling time interval to obtain the rate of change of pressure at that moment. The method for calculating the rate of change of temperature is similar. For example, assuming the sampling time interval is 10 seconds, the pressure value at a certain moment is 5.2 MPa, and the pressure value at the previous moment is 5.0 MPa, then the rate of change of pressure = 0.2 / 10 = 0.02 MPa per second.

[0042] Specifically, the pressure change rate reflects the rate at which the medium pressure rises or falls. An excessively rapid rate of pressure rise can lead to a sharp increase in the axial load on the packing, potentially causing packing extrusion. Conversely, an excessively rapid rate of pressure fall can cause a rapid decrease in the pressure difference across the sealing interface, potentially resulting in insufficient packing rebound and loss of sealing pressure. The temperature change rate reflects the rate of change in the thermal load on the packing. Sudden temperature rises or falls can generate additional stress at the sealing interface due to differences in thermal expansion coefficients.

[0043] Secondly, based on the pressure change rate sequence and temperature change rate sequence, the pressure change rate and temperature change rate at each time point are weighted and fused to obtain the pressure-temperature joint change rate at each time point, forming a pressure-temperature joint change rate sequence. Specifically, the weighted fusion calculation formula is: Pressure-Temperature Joint Change Rate = α × Pressure Change Rate + β × Temperature Change Rate, where α is the pressure change rate weight, β is the temperature change rate weight, and the sum of α and β is 1. The weights of the pressure change rate and temperature change rate are pre-calibrated based on the degree of influence of the pressure change rate and temperature change rate on the sealing failure of the packing gland moving seal area in historical operating data. For example, if historical data statistical analysis shows that pressure fluctuation is the main cause of packing leakage in this valve, then the weight of α can be set to 0.7 and the weight of β can be set to 0.3; if the influence of temperature change is more significant, then the weight of α can be set to 0.3 and the weight of β can be set to 0.7; under normal operating conditions, both the weight of α and the weight of β can be set to 0.5. The combined pressure-temperature change rate comprehensively reflects the intensity of the working condition impact. The larger the value, the more drastic the pressure or temperature change is at the current moment, the more the sealing interface is in a non-equilibrium state, and the risk of leakage is significantly increased.

[0044] Furthermore, the predicted hydrogen concentration sequence is subtracted from the actual hydrogen concentration sequence at each time step to obtain the hydrogen concentration difference at each time step, thus forming a hydrogen concentration difference sequence. For example, let C be the measured hydrogen concentration at the t-th sampling time. actual(t) The predicted hydrogen concentration output by the hydrogen valve twin model is C. predicted(t) Then the hydrogen concentration difference ΔC at that moment (t) =C actual(t) -C predicted(t) When ΔC (t) When ΔC is positive, it indicates that the measured hydrogen concentration is higher than the predicted concentration under normal sealing conditions, and the difference reflects the leakage component caused by the deterioration of sealing performance; when ΔC (t) When ΔC is close to zero, it indicates that the measured hydrogen concentration is basically consistent with the predicted concentration, and the sealing condition is normal; when ΔC (t) Negative values ​​may be caused by sensor measurement errors or prediction biases in the hydrogen valve twin model, requiring comprehensive judgment in conjunction with other indicators. The hydrogen concentration difference sequence reflects the actual size of the leak and its trend over time, serving as the core basis for determining the severity of the leak and dynamically adjusting indicator weights in subsequent emergency assessments.

[0045] In summary, through the above calculations, the original pressure, temperature, and hydrogen concentration sequences were transformed into a pressure-temperature combined rate of change sequence and a hydrogen concentration difference sequence. These two sequences characterize the sealing status of the stuffing box dynamic sealing zone from two dimensions: the intensity of the operating condition impact and the degree of leakage deviation, respectively. The pressure-temperature combined rate of change sequence and the hydrogen concentration difference sequence, together with the original sequences, constitute a multi-dimensional monitoring index system, providing comprehensive feature input for subsequent emergency assessments.

[0046] S40: Based on the pressure sequence, temperature sequence, hydrogen concentration sequence, pressure-temperature combined rate of change sequence and hydrogen concentration difference sequence, perform an emergency assessment of hydrogen leak monitoring and determine the emergency index of leak monitoring. In this process, the baseline weights set in advance using the analytic hierarchy process are adjusted according to the hydrogen concentration difference sequence to obtain the appropriate weight ratio, and the emergency assessment is performed based on the appropriate weight ratio. Furthermore, after obtaining the pressure, temperature, hydrogen concentration, pressure-temperature combined rate of change, and hydrogen concentration difference sequences, statistical features need to be extracted from these five sequences to quantify the current leakage urgency of the dynamic sealing zone of the stuffing box. These five sequences reflect the state characteristics of the dynamic sealing zone of the stuffing box in different physical dimensions: the pressure sequence reflects the load level borne by the sealing interface; the temperature sequence reflects the risk of packing aging; the hydrogen concentration sequence reflects the absolute value of the current leakage; the pressure-temperature combined rate of change sequence reflects the impact intensity of sudden changes in operating conditions on the sealing interface; and the hydrogen concentration difference sequence reflects the trend of deterioration in sealing performance.

[0047] The five monitoring indicators mentioned above vary in their sensitivity to indicating leakage risk, and their importance also differs at different stages of the leak. In the early stages of a leak, the absolute value of hydrogen concentration is low, but the upward trend of the hydrogen concentration difference is more sensitive; when operating conditions fluctuate drastically, the immediate impact of the combined pressure-temperature change rate is greater than the steady-state concentration value. Therefore, a simple weighted average of the indicators using fixed weights is not feasible; the weights of each indicator need to be dynamically adjusted according to the current leak status.

[0048] This step first uses the Analytic Hierarchy Process (AHP) to set the baseline weights for each monitoring indicator. The AHP is a multi-criteria decision-making method that combines qualitative and quantitative approaches. By constructing a hierarchical analysis judgment matrix for the indicator parameter set, pairwise importance comparisons are performed on each indicator parameter. The relative importance is quantified using a 1-9 scale, and a normalized weight vector is calculated after a consistency test.

[0049] Specifically, in this embodiment, the step of pre-setting the baseline weights using the analytic hierarchy process includes: Configure a set of indicator parameters, wherein the set of indicator parameters includes average pressure, pressure fluctuation coefficient, average temperature, temperature fluctuation coefficient, average hydrogen concentration, hydrogen concentration fluctuation coefficient, average pressure-temperature combined change rate, pressure-temperature combined change rate fluctuation coefficient, average hydrogen concentration difference, and hydrogen concentration difference fluctuation coefficient. A hierarchical analysis judgment matrix is ​​constructed for the set of index parameters. By comparing the importance of each index parameter in the set of index parameters pairwise, and based on the degree of influence of each index parameter on the monitoring of hydrogen leakage in the dynamic sealing area of ​​the stuffing box, the relative importance of each index parameter is quantified and assigned using the 1-9 scale method. A consistency check is performed on the judgment matrix, and the consistency ratio is calculated. When the consistency ratio is less than 0.1, the judgment matrix passes the consistency check. The weight vector of each indicator parameter is calculated for the judgment matrix that passes the consistency check, and the weight vector is normalized to obtain the benchmark weight of each indicator parameter in the indicator parameter set.

[0050] First, configure the parameter set. This parameter set includes ten parameters: average pressure, pressure fluctuation coefficient, average temperature, temperature fluctuation coefficient, average hydrogen concentration, hydrogen concentration fluctuation coefficient, average pressure-temperature combined change rate, average pressure-temperature combined change rate fluctuation coefficient, average hydrogen concentration difference, and hydrogen concentration difference fluctuation coefficient. These ten parameters characterize the sealing status of the stuffing box dynamic sealing zone from different dimensions. The physical meanings of the average value and fluctuation coefficient are as follows: the average value reflects the average level of the parameter within the historical monitoring and adjustment time window; for example, the average pressure reflects the average medium pressure during this period, and the average hydrogen concentration reflects the average leakage concentration during this period. The fluctuation coefficient equals the standard deviation divided by the average value, used to measure the relative dispersion of the parameter within this period. A larger fluctuation coefficient indicates more drastic changes in the parameter and a more unstable sealing status.

[0051] Secondly, a hierarchical analysis (AHP) judgment matrix for the indicator parameter set is constructed. The AHP is a multi-criteria decision-making method that combines qualitative and quantitative approaches. Its core is to construct a square matrix, where the rows and columns correspond to the various indicators in the indicator parameter set. The element in the i-th row and j-th column of the matrix represents the importance ratio of the i-th indicator to the j-th indicator. By comparing the pairwise importance of each indicator parameter in the indicator parameter set, and based on the degree of influence of each indicator parameter on the monitoring of hydrogen leakage in the dynamic sealing area of ​​the stuffing box, a 1-9 scale is used to quantify the relative importance between each indicator parameter. The assignment rules for the 1-9 scale are as follows: scale 1 indicates that the two indicators are equally important; scale 3 indicates that one indicator is slightly more important than the other; scale 5 indicates significant importance; scale 7 indicates strong importance; scale 9 indicates extreme importance; and scales 2, 4, 6, and 8 represent the median values ​​of the above adjacent judgments.

[0052] The scaling criteria are based on a comprehensive analysis of the physical mechanisms and engineering experience of hydrogen leakage by domain experts. For example, in one optional embodiment, the specific determination process is as follows: The average pressure reflects the average medium pressure experienced by the dynamic sealing zone of the stuffing box. Increased pressure increases the pressure difference across the sealing interface, enhancing the driving force for hydrogen diffusion along the sealing gap. However, increased pressure also increases the axial clamping force on the packing, which to some extent helps maintain the sealing contact pressure. Therefore, the average pressure has a dual effect on leakage, and its importance is relatively moderate. The pressure fluctuation coefficient reflects the severity of medium pressure fluctuations. Frequent or rapid pressure fluctuations can cause fatigue deformation of the packing under alternating loads, accelerating the degradation of sealing performance. Compared to steady-state pressure, pressure fluctuations have a more significant impact on seal life; therefore, the pressure fluctuation coefficient is usually more important than the average pressure.

[0053] The average temperature reflects the thermal load level of the environment in which the packing is located. Increased temperature accelerates the thermal oxidation and aging of the packing material, reduces its elastic recovery ability, and causes permanent compressive deformation, making it unable to maintain the initial sealing pressure. Simultaneously, increased temperature also increases the thermal motion rate of hydrogen molecules, increasing permeability. For valves operating in high-temperature environments, the importance of the average temperature is significantly enhanced. The temperature fluctuation coefficient reflects the severity of thermal load fluctuations. Frequent temperature fluctuations cause additional stress between the packing and the valve stem / body due to differences in thermal expansion coefficients, accelerating interface wear and seal failure. Its importance is usually comparable to or slightly lower than the average temperature.

[0054] The average hydrogen concentration reflects the average leakage rate during the current monitoring period. When the seal is normal, the average hydrogen concentration is at a low background permeability level; when the seal performance deteriorates, the average hydrogen concentration increases significantly. This indicator is the most direct evidence of whether a leak has occurred and has the highest importance among all indicators. The hydrogen concentration fluctuation coefficient reflects the degree of fluctuation in leakage. When the leak is intermittent or fluctuating, this indicator can capture abnormal fluctuations that a stable average cannot reflect, and its importance is second only to the average.

[0055] In addition, the mean pressure-temperature combined change rate reflects the average intensity of the operating condition impact during the monitoring period. Rapid changes in pressure and temperature can cause the sealing interface to be in a non-equilibrium state, significantly increasing the risk of instantaneous leakage. This indicator has early warning value for sudden leaks, and its importance is second only to the direct concentration indicator. The pressure-temperature combined change rate fluctuation coefficient reflects the degree of fluctuation in the intensity of the operating condition impact. This indicator has relatively low importance. The mean hydrogen concentration difference reflects the average extent to which the measured hydrogen concentration exceeds the normal permeability benchmark. It is the pure leakage component after stripping away the operating condition influences, directly characterizing the severity of sealing performance degradation, and its importance is second only to the mean hydrogen concentration. The hydrogen concentration difference fluctuation coefficient reflects the degree of fluctuation in the leakage component. When the seal is in an intermittent leakage state, this indicator can reflect the instability of the sealing performance, and its importance is moderate.

[0056] Based on the above physical criteria, pairwise comparisons were performed on the ten indicators to determine the relative importance scale of each pair. For example, comparing the mean hydrogen concentration with the mean pressure: the mean hydrogen concentration is direct evidence of leakage, while the mean pressure is only an influencing factor. Therefore, the mean hydrogen concentration is significantly more important than the mean pressure, and the scale is set to 5. The corresponding element in the judgment matrix is ​​assigned a value of 5, while the element for the mean pressure is assigned a value of 1 / 5 compared to the mean hydrogen concentration. Comparing the mean hydrogen concentration with the mean hydrogen concentration difference: both are direct leakage indicators, but the mean hydrogen concentration reflects the total leakage including normal seepage, while the mean hydrogen concentration difference reflects the pure leakage after deducting the influence of operating conditions. When the seal is normal, the former is dominant, while the latter is more sensitive when a leak occurs. Both are equally important, and the scale is set to 1. Through the above pairwise comparisons, the judgment matrix is ​​assigned values. It should be noted that the judgment matrix must be jointly determined by a team of experts in valve design and sealing based on the physical mechanism of hydrogen leakage and on-site operation and maintenance experience to ensure the rationality of the scale assignment and the engineering applicability.

[0057] Furthermore, a consistency test is performed on the judgment matrix, and the consistency ratio is calculated. If the consistency ratio is less than 0.1, the judgment matrix passes the consistency test. Since the judgment matrix is ​​constructed by manually assigning values ​​pairwise, logical inconsistencies may exist, such as contradictory judgments like A being superior to B, or B being superior to C but C being superior to A. The consistency test is used to assess whether the degree of inconsistency is within an acceptable range.

[0058] Specifically, first, the largest eigenvalue λ of the judgment matrix is ​​calculated. max Then calculate the consistency index CI, the formula is CI = (λ) / (λ) max - (Judgment matrix order) / (Judgment matrix order - 1). Then calculate the consistency ratio CR, calculated as CR = CI / RI, where RI is the average random consistency index, its value obtained from a table based on the order of the judgment matrix. For example, for a 10th order matrix, RI is 1.49. When CR is less than 0.1, the judgment matrix passes the consistency test, indicating that the degree of inconsistency in the judgment matrix is ​​within the allowable range, and the weight calculation results are reliable. If CR is greater than or equal to 0.1, the judgment matrix needs to be reconstructed, and the scale assignments for pairwise comparisons adjusted until the consistency test passes.

[0059] Next, the weight vectors of each indicator parameter are calculated for the judgment matrix that has passed the consistency test, and the weight vectors are normalized to obtain the baseline weights of each indicator parameter in the indicator parameter set. The weight vectors can be calculated using the eigenvector method or the geometric mean method. In an optional embodiment, the eigenvector method is used as an example: the maximum eigenvalue λ of the judgment matrix is ​​calculated. maxThe corresponding eigenvector is used to divide each element of the eigenvector by the sum of all elements, resulting in a normalized weight vector. Each component in this weight vector is the baseline weight for the corresponding indicator parameter. The baseline weight represents the theoretical contribution of each indicator to the leakage risk under normal sealing conditions, and the sum of the baseline weights for the ten indicators is 1.

[0060] Furthermore, the baseline weights are adjusted based on the hydrogen concentration difference sequence to obtain the appropriate weight ratio. The hydrogen concentration difference is the core indicator for measuring the deviation of the measured concentration from the normal permeability baseline. When the hydrogen concentration difference is small, it indicates that the sealing condition is normal or there is only a trace amount of permeation. In this case, the indicators can be integrated according to the baseline weights. When the hydrogen concentration difference increases, it indicates that the sealing performance has deteriorated and leakage is occurring or aggravating. In this case, the weights of indicators directly related to leakage, such as the hydrogen concentration difference and its fluctuation coefficient, should be increased, while the weights of other indicators should be reduced accordingly, so that the urgency assessment results adaptively focus on leakage-sensitive indicators.

[0061] Specifically, in this embodiment, adjusting the baseline weights based on the hydrogen concentration difference sequence to obtain the adaptation weight ratio includes: Calculate the mean value of the actual hydrogen concentration difference and the fluctuation coefficient of the actual hydrogen concentration difference within the historical monitoring and adjustment time window of the hydrogen concentration difference sequence; Based on the pre-calibrated baseline thresholds for the average hydrogen concentration difference and the fluctuation coefficient of the hydrogen concentration difference, the average actual hydrogen concentration difference and the fluctuation coefficient of the actual hydrogen concentration difference are normalized respectively to obtain the normalized value of the average hydrogen concentration difference and the normalized value of the fluctuation coefficient of the hydrogen concentration difference. The leakage deviation coefficient for the current period is obtained by weighted summing of the normalized value of the mean hydrogen concentration difference and the normalized value of the hydrogen concentration difference fluctuation coefficient; wherein, the weights of the normalized value of the mean hydrogen concentration difference and the normalized value of the hydrogen concentration difference fluctuation coefficient are pre-calibrated based on the degree of influence of the two indicators on the degree of leakage deviation. Based on the leakage deviation coefficient, the product of the leakage deviation coefficient and the preset adjustment coefficient is used as the weight adjustment range to adjust the benchmark weights corresponding to the average hydrogen concentration difference and the hydrogen concentration difference fluctuation coefficient in the set of index parameters. The adjusted weights do not exceed the preset weight upper limit. Based on the increased weight increment, the benchmark weights corresponding to the other indicator parameters are proportionally reduced according to their respective benchmark weights to obtain the appropriate weight ratio. The sum of the weights of each indicator parameter after adjustment remains 1.

[0062] First, the mean actual hydrogen concentration difference and the fluctuation coefficient of the actual hydrogen concentration difference sequence are calculated within the historical monitoring and adjustment time window. The hydrogen concentration difference sequence is obtained by subtracting the measured hydrogen concentration sequence from the predicted hydrogen concentration sequence at each time step. The arithmetic mean of this sequence is calculated to obtain the mean actual hydrogen concentration difference, which represents the average extent to which the measured concentration exceeds the normal permeability baseline within this time window. The standard deviation of this sequence is calculated and divided by the mean to obtain the fluctuation coefficient of the actual hydrogen concentration difference, which represents the relative fluctuation of the leakage component within this time window.

[0063] Secondly, based on the pre-calibrated baseline thresholds for the average hydrogen concentration difference and the fluctuation coefficient of the hydrogen concentration difference using normal operating condition data, the actual average hydrogen concentration difference and the actual fluctuation coefficient of the hydrogen concentration difference are normalized to obtain normalized values ​​for the average hydrogen concentration difference and the fluctuation coefficient of the hydrogen concentration difference. Specifically, the baseline threshold for the average hydrogen concentration difference is the upper limit determined by statistically analyzing the distribution range of the average hydrogen concentration difference in normal operating condition data, under conditions where the valve is in a normal sealing state and there is only design-allowed permeation without abnormal leakage. For example, it can be set to the value corresponding to three times the standard deviation of the average hydrogen concentration difference under normal operating conditions. The baseline threshold for the fluctuation coefficient of the hydrogen concentration difference is calculated similarly. The normalization calculation formula is: Normalized value = Actual value / Baseline threshold, and the calculation result is limited to the range of 0 to 1. That is, if the actual value exceeds the baseline threshold, the normalized value is 1. The closer the normalized value is to 1, the closer the current leakage deviation is to or exceeds the upper limit of fluctuation under normal operating conditions.

[0064] Next, the normalized values ​​of the mean hydrogen concentration difference and the normalized values ​​of the hydrogen concentration difference fluctuation coefficient are weighted and summed to obtain the leakage deviation coefficient for the current period. Specifically, the weighted summation formula is: Leakage deviation coefficient = w1 × Normalized value of the mean hydrogen concentration difference + w2 × Normalized value of the hydrogen concentration difference fluctuation coefficient, where w1 and w2 are pre-calibrated weight coefficients, and the sum of w1 and w2 is 1. The weights of the normalized values ​​of the mean hydrogen concentration difference and the normalized value of the hydrogen concentration difference fluctuation coefficient are pre-calibrated based on the ability of the two indicators to characterize the degree of leakage deviation. Since the mean hydrogen concentration difference directly reflects the average exceedance of the leakage amount, while the hydrogen concentration difference fluctuation coefficient only reflects the fluctuation characteristics of the leakage, the former is usually more important than the latter. Therefore, w1 can be set to 0.7 and w2 to 0.3. The leakage deviation coefficient ranges from 0 to 1, with a larger value indicating a more severe deviation of the current sealing state from the normal baseline.

[0065] Furthermore, based on the leakage deviation coefficient, the product of the leakage deviation coefficient and the preset adjustment coefficient is used as the weight adjustment range to increase the baseline weights corresponding to the average hydrogen concentration difference and the hydrogen concentration difference fluctuation coefficient in the indicator parameter set. The preset adjustment coefficient is a pre-set hyperparameter used to control the sensitivity of the weight adjustment. It can be set comprehensively according to the system's sensitivity requirements to leakage changes and the tolerance of the field application for weight stability; for example, it can be set to 0.5. Weight adjustment range = Leakage deviation coefficient × Preset adjustment coefficient. The adjusted weight of the average hydrogen concentration difference = Baseline weight × (1 + Weight adjustment range), and the adjusted weight of the hydrogen concentration difference fluctuation coefficient = Baseline weight × (1 + Weight adjustment range).

[0066] Simultaneously, the adjusted weights are set to not exceed a preset weight upper limit. This preset weight upper limit is a constraint value comprehensively set based on the physical upper limit of the indicator parameters' ability to characterize leakage risk and the consensus from engineering experience that a single indicator should not completely dominate the assessment results. It is used to prevent the weights of the average hydrogen concentration difference or the hydrogen concentration difference fluctuation coefficient from being increased to an unreasonable level when the leakage deviation coefficient is extremely large, causing the assessment results to be dominated by a single indicator and losing the comprehensive judgment advantage of multi-parameter fusion. For example, a weight upper limit of 0.40 can be set. When the weight calculated according to the adjustment formula exceeds this upper limit, the upper limit value is directly taken as the adjusted weight, and the excess portion is not adjusted downwards, ensuring that the weights of each indicator always remain within a reasonable range.

[0067] Finally, based on the increased weight increments, the benchmark weights corresponding to the remaining indicator parameters are proportionally reduced according to their respective benchmark weights to obtain the appropriate weight proportions. Specifically, let the increased weight increment of the average hydrogen concentration difference = increased weight - benchmark weight, and the increased weight increment of the hydrogen concentration difference fluctuation coefficient is similarly calculated. The two weight increments are added together to obtain the total weight increase. For the eight indicators in the indicator parameter set other than the average hydrogen concentration difference and the hydrogen concentration difference fluctuation coefficient, the sum of their benchmark weights is 1 minus the sum of the benchmark weights of the two indicators mentioned above. The weights are reduced proportionally according to the proportion of each indicator's benchmark weight to the sum of the eight indicator benchmark weights, so that the reduction of each indicator = the indicator's benchmark weight × (total weight increase / sum of the eight indicator benchmark weights).

[0068] In summary, after the above adjustments, the sum of the weights of all ten indicators remains at 1. The adjusted weights are the adaptive weight proportions. The characteristic of this adaptive weight proportion is that when the leakage deviation coefficient is large, the weights of indicators related to hydrogen concentration difference are significantly increased, causing the urgency assessment results to adaptively focus on leakage-sensitive indicators; when the leakage deviation coefficient is close to 0, the adjustment amount approaches 0, and the adaptive weight proportion degenerates into the baseline weight. Through this dynamic weight adjustment mechanism, the urgency assessment can automatically adjust the contribution of each indicator according to the current leakage status, improving the accuracy of the assessment results.

[0069] Further, after weight adjustment, the actual monitoring indicator set is weighted according to the appropriate weight ratio to obtain the leak monitoring urgency index. Specifically, in this embodiment, based on the pressure sequence, temperature sequence, hydrogen concentration sequence, pressure-temperature combined rate of change sequence, and hydrogen concentration difference sequence, the urgency of hydrogen leak monitoring is assessed to determine the leak monitoring urgency index, including: Based on the pressure sequence, temperature sequence, hydrogen concentration sequence, pressure-temperature combined rate of change sequence, and hydrogen concentration difference sequence, the actual average pressure, actual pressure fluctuation coefficient, actual average temperature, actual temperature fluctuation coefficient, actual average hydrogen concentration, actual hydrogen concentration fluctuation coefficient, actual average pressure-temperature combined rate of change, actual pressure-temperature combined rate of change fluctuation coefficient, actual average hydrogen concentration difference, and actual hydrogen concentration difference fluctuation coefficient are calculated and used as the actual monitoring index set. Based on the normalized reference parameters determined in advance according to normal operating condition data, the actual monitoring indicators in the actual monitoring indicator set are subjected to segmented normalization processing to obtain the normalized value of each actual monitoring indicator. The normalized values ​​of each actual monitoring indicator in the actual monitoring indicator set are weighted according to the adaptation weight ratio to obtain the leakage monitoring emergency index. The leakage monitoring emergency index is positively correlated with each actual monitoring indicator in the actual monitoring indicator set. The larger the leakage monitoring emergency index, the higher the risk of hydrogen leakage in the dynamic sealing area of ​​the stuffing box.

[0070] First, based on the pressure series, temperature series, hydrogen concentration series, pressure-temperature combined rate of change series, and hydrogen concentration difference series, the actual average pressure, actual pressure fluctuation coefficient, actual average temperature, actual temperature fluctuation coefficient, actual average hydrogen concentration, actual hydrogen concentration fluctuation coefficient, actual average pressure-temperature combined rate of change series, actual pressure-temperature combined rate of change series, actual average hydrogen concentration difference, and actual hydrogen concentration difference fluctuation coefficient are calculated as the actual monitoring index set. The calculation method for the above ten indicators is as follows: For each series, firstly, the arithmetic mean of the series within the historical monitoring adjustment time window is calculated as the corresponding mean index; then, the standard deviation of the series is calculated, and the fluctuation coefficient is obtained by dividing the standard deviation by the mean. Taking the pressure series as an example, the actual average pressure is equal to the sum of all pressure values ​​in the pressure series divided by the number of sampling points, and the actual pressure fluctuation coefficient is equal to the standard deviation of the pressure series divided by the actual average pressure. For hydrogen concentration sequences and hydrogen concentration difference sequences, when calculating the fluctuation coefficient, if the mean approaches zero, to avoid division by zero errors, the fluctuation coefficient can be set to directly take the value after standard deviation normalization or a preset minimum value can be used as the lower limit of the denominator.

[0071] Secondly, before calculating the leakage monitoring emergency index by weighted fusion of the actual monitoring index set, it is necessary to normalize each actual monitoring index to eliminate the dimensional differences between different indicators and ensure the comparability of each indicator during weighted fusion. To this end, normalized baseline parameters corresponding to each actual monitoring index are obtained in advance based on historical data under normal operating conditions. These normalized baseline parameters include a normal threshold and an alarm threshold. The normal threshold is the upper limit determined by statistically analyzing the distribution range of each actual monitoring index in the normal operating condition data, under conditions where the valve is in a normal sealing state and there is only design-allowed permeation without abnormal leakage. For example, it can be set as the value corresponding to three times the standard deviation of the index's mean under normal operating conditions. The alarm threshold is a critical value set according to the critical standard for valve sealing performance deterioration or relevant safety regulations. When the actual monitoring index exceeds the alarm threshold, it indicates that the sealing performance has severely deteriorated and the risk of leakage is extremely high. Specifically, these normalized baseline parameters are used to delineate the boundaries between the normal and abnormal ranges of each actual monitoring index, providing a benchmark for subsequent segmented normalization processing.

[0072] Specifically, for each actual monitoring indicator, based on the relationship between its actual value and the normal and alarm thresholds, a piecewise normalization method is used to map the actual value of the indicator to a range of 0 to 1. The necessity of piecewise normalization lies in the fact that the actual monitoring indicators include indicators such as the average hydrogen concentration difference and the hydrogen concentration difference fluctuation coefficient. Under normal sealing conditions, the actual values ​​of these indicators approach zero, and the dynamic range of historical normal operating data is extremely small. However, when a real leak occurs, the actual values ​​of these indicators will significantly increase, far exceeding the historical normal operating range. If the traditional extreme value normalization method is used, on the one hand, the difference between the maximum and minimum values ​​under normal operating conditions approaches zero, leading to instability in the denominator; on the other hand, the actual value during a leak far exceeds the historical extreme value, causing the normalization result to be distorted. Therefore, this step uses a piecewise normalization method to process each actual monitoring indicator.

[0073] The specific implementation of segmented normalization processing is based on the relationship between the actual value of each monitored indicator and the normal threshold and alarm threshold. The normalization process is divided into three intervals. For example, suppose the actual value of a monitored indicator is x, the normal threshold is N, and the alarm threshold is A. When x does not exceed N, the indicator is in the normal interval, and the normalized value = x / N, ranging from 0 to 1, reflecting the relative level of the indicator within the normal range. When x exceeds N but does not exceed A, the indicator is in the warning interval, and the normalized value = 1 + (xN) / (AN), ranging from 1 to 2, reflecting the degree to which the indicator deviates from the normal range; the larger the value, the more severe the deviation. When x exceeds A, the indicator is in the danger interval, and the normalized value = 2 + (xA) / A, ranging from 2 to 3. The upper limit of the range can be set according to actual needs, reflecting the severity of the indicator's abnormality. Through the above segmented normalization process, when the actual value is within the normal range, the normalized value is compressed into the interval between 0 and 1; when the actual value exceeds the normal threshold but does not reach the alarm threshold, the normalized value linearly expands to the interval between 1 and 2; when the actual value exceeds the alarm threshold, the normalized value further expands to the interval between 2 and 3. This process ensures the stability of the denominator, while enabling the normalized value to continuously and monotonically reflect the change process of each actual monitoring indicator from normal to abnormal.

[0074] It should be noted that the value range division of each interval in the above segmented normalization process can be adjusted according to the actual application scenario. The segmentation method of 0 to 1, 1 to 2, and 2 to 3 in this embodiment is only for illustrative purposes and does not constitute a limitation on the normalization process.

[0075] Subsequently, after segmented normalization, the actual monitoring index set is weighted according to the appropriate weight ratio to obtain the leakage monitoring emergency index. The appropriate weight ratio is the weight coefficient corresponding to the ten indicators obtained by dynamically adjusting the hydrogen concentration difference sequence in the aforementioned steps. Each weight coefficient is between 0 and 1, and the sum of the ten weight coefficients is 1. The weighted calculation formula is: Leakage Monitoring Emergency Index = Σ(Normalized actual value of the i-th indicator × Applied weight of the i-th indicator), where i ranges from 1 to 10. Specifically, the leakage monitoring emergency index is positively correlated with each actual monitoring indicator in the actual monitoring index set. That is, an increase in the dimensionless value of any indicator will lead to a corresponding increase in the leakage monitoring emergency index, but the magnitude of the increase for different indicators is determined by the corresponding appropriate weight. The leakage monitoring emergency index ranges from 0 to 3. A larger value indicates a higher risk of hydrogen leakage in the dynamic sealing area of ​​the stuffing box, requiring more urgent monitoring measures or triggering an early warning.

[0076] In summary, through the above urgency assessment, this step compresses multi-dimensional, multi-time-series information into a dimensionless leak monitoring urgency index, which provides a quantitative decision-making basis for the adaptive adjustment of subsequent monitoring strategies.

[0077] S50: Based on the leak monitoring emergency index, query the pre-built emergency index-monitoring parameter group association table, determine the appropriate monitoring parameter group, generate an appropriate monitoring scheme in combination with the appropriate weight ratio, and execute hydrogen leak monitoring of the stuffing gland dynamic sealing area within the next monitoring adjustment time window.

[0078] Finally, after calculating the leakage monitoring urgency index for the current period, the monitoring strategy for the next monitoring adjustment time window needs to be dynamically adjusted based on this index. Hydrogen leakage in the dynamic sealing area of ​​the stuffing box exhibits a progressive failure characteristic, gradually developing from normal sealing to minor leakage and then to major leakage. When the sealing condition is normal or there is only a trace amount of seepage, the leakage risk is low. At this time, a more comprehensive set of monitoring parameters should be used, collecting all five monitoring parameters, including pressure, temperature, hydrogen concentration, pressure-temperature combined change rate, and hydrogen concentration difference, to accumulate complete operating condition and leakage data for trend analysis and model calibration. When the leakage monitoring urgency index increases, it indicates that the sealing performance has deteriorated and the leakage risk has increased. At this time, the monitoring parameter set should be appropriately simplified, reducing the frequency of collecting parameters not directly related to leakage or suspending collection, focusing monitoring resources on hydrogen concentration to shorten the monitoring cycle, improve response speed, and ensure that an alarm can be triggered as soon as the leakage intensifies.

[0079] Specifically, this step uses an emergency index-monitoring parameter group association table, with the leak monitoring emergency index as the query key, to determine the appropriate monitoring parameter group corresponding to the current risk level through a table lookup. Specifically, in this embodiment, determining the appropriate monitoring parameter group by querying the pre-built emergency index-monitoring parameter group association table based on the leak monitoring emergency index includes: Based on historical operating data and leakage test data, the value range of the leakage monitoring emergency index is divided into several emergency index intervals in advance. For each emergency index interval, a corresponding monitoring parameter group is configured, and an emergency index-monitoring parameter group association table is constructed. The number of monitoring parameters in the monitoring parameter group is inversely proportional to the urgency of the emergency index interval. Specifically, when the leak monitoring emergency index is in the lowest emergency index range, the corresponding monitoring parameter group is the maximum monitoring parameter group, which includes five monitoring parameters: pressure, temperature, hydrogen concentration, pressure-temperature combined change rate, and hydrogen concentration difference; when the leak monitoring emergency index is in the highest emergency index range, the corresponding monitoring parameter group is the minimum monitoring parameter group, which includes only the single monitoring parameter of hydrogen concentration. The selection of monitoring parameters in each monitoring parameter group is based on the sorting result of the benchmark weights, and the monitoring parameters are selected in descending order of the benchmark weights of the corresponding indicator parameters. During the operation of the high-pressure shut-off valve, the leakage monitoring emergency index for the current time period is matched with the emergency index-monitoring parameter group association table to determine the appropriate monitoring parameter group for the current time period. The appropriate monitoring parameter group is used to determine the parameter range participating in the weighted fusion calculation of leakage monitoring. The pressure sensor, temperature sensor and MEMS hydrogen concentration sensor continuously collect data during the operation of the high-pressure shut-off valve and are not affected by the switching of the appropriate monitoring parameter group.

[0080] First, based on historical operating data and leakage test data, the range of the leakage monitoring emergency index is divided into several consecutive emergency index intervals. For each emergency index interval, a corresponding monitoring parameter group is configured, and an emergency index-monitoring parameter group association table is constructed. The number of monitoring parameters in a monitoring parameter group is inversely proportional to the urgency of the emergency index interval; that is, the smaller the emergency index, the more monitoring parameters the monitoring parameter group contains, and vice versa.

[0081] For example, in an optional embodiment, the value range of 0 to 1 can be divided into four intervals: interval one is 0 to 0.2, corresponding to low risk; interval two is 0.2 to 0.5, corresponding to medium risk; interval three is 0.5 to 0.8, corresponding to high risk; and interval four is 0.8 to 1.0, corresponding to extremely high risk. It should be noted that the boundary values ​​for dividing the emergency index intervals can be adjusted according to the tolerance for leakage risk and alarm delay requirements in actual engineering practice. For situations with low risk tolerance, the boundary values ​​for each interval can be appropriately reduced.

[0082] Secondly, the selection of monitoring parameters in each monitoring parameter group is based on the ranking results of the benchmark weights. The parameters are selected in descending order of their benchmark weights to ensure that parameters with higher importance are prioritized for inclusion in the monitoring parameter group. Specifically, the benchmark weights of ten indicators—mean pressure, pressure fluctuation coefficient, mean temperature, temperature fluctuation coefficient, mean hydrogen concentration, hydrogen concentration fluctuation coefficient, mean pressure-temperature combined rate of change, pressure-temperature combined rate of change fluctuation coefficient, mean hydrogen concentration difference, and hydrogen concentration difference fluctuation coefficient—are calculated first using the analytic hierarchy process (AHP). These ten indicators are then ranked in descending order of their benchmark weights. A correspondence exists between the monitoring parameters in each monitoring parameter group and the aforementioned indicators: mean pressure and pressure fluctuation coefficient correspond to pressure monitoring parameters; mean temperature and temperature fluctuation coefficient correspond to temperature monitoring parameters; mean hydrogen concentration and hydrogen concentration fluctuation coefficient correspond to hydrogen concentration monitoring parameters; mean pressure-temperature combined rate of change and pressure-temperature combined rate of change fluctuation coefficient correspond to pressure-temperature combined rate of change monitoring parameters; and mean hydrogen concentration difference and hydrogen concentration difference fluctuation coefficient correspond to hydrogen concentration difference monitoring parameters. Therefore, the priority order of the five monitoring parameters can be determined based on the ranking results of the benchmark weights of the indicator parameters.

[0083] For example, if the average hydrogen concentration and the average hydrogen concentration difference rank first and second in the benchmark weighting, then the hydrogen concentration monitoring parameter and the hydrogen concentration difference monitoring parameter have the highest priority; if the average pressure rank second, then the pressure monitoring parameter has the next highest priority, and so on.

[0084] For example, when the leak monitoring emergency index is in the lowest emergency index range, the corresponding monitoring parameter set is the maximum monitoring parameter set, which includes all five monitoring parameters: pressure, temperature, hydrogen concentration, pressure-temperature combined change rate, and hydrogen concentration difference. At this time, the sealing condition is normal or there is only a very small leak, and the risk is low. The most comprehensive monitoring parameter set should be used to accumulate complete operating condition and leak data for long-term trend analysis and online calibration of the hydrogen valve twin model. When the leak monitoring emergency index is in the highest emergency index range, the corresponding monitoring parameter set is the minimum monitoring parameter set, which only includes the single monitoring parameter of hydrogen concentration. At this time, the sealing performance is severely deteriorated, and the risk of leakage is extremely high. The collection of other monitoring parameters should be suspended, and all monitoring resources should be focused on hydrogen concentration, the most direct evidence of leakage, monitoring concentration changes at the highest sampling frequency to ensure that an alarm is triggered immediately if the leakage further increases. For the intermediate emergency index range, the number of monitoring parameters in the corresponding monitoring parameter group is between the largest and smallest groups. For example, the monitoring parameter group corresponding to the medium-risk range may include three monitoring parameters: hydrogen concentration, pressure, and temperature. The monitoring parameter group corresponding to the high-risk range may include two monitoring parameters: hydrogen concentration and hydrogen concentration difference.

[0085] It is important to note that the sensor acquisition layer is responsible for continuously and in real-time acquiring pressure, temperature, and hydrogen concentration data in the moving seal area of ​​the stuffing box, and is unaffected by switching of the adaptive monitoring parameter group. The fusion calculation layer is responsible for performing weighted fusion calculations on the acquired data based on the adaptive monitoring parameter group and its weighting proportions to obtain the leak monitoring urgency index. Switching of the adaptive monitoring parameter group only affects the fusion calculation layer and does not affect the normal operation of the sensor acquisition layer.

[0086] During the operation of the high-pressure shut-off valve, the pressure sensor, temperature sensor, and MEMS hydrogen concentration sensor continuously acquire data to ensure that the hydrogen valve twin model receives complete pressure and temperature parameters within any monitoring time window. This allows for the continuous output of the predicted hydrogen concentration in the stuffing box's moving seal area, ensuring uninterrupted calculation of the hydrogen concentration difference sequence. At the fusion calculation layer, the switching of the adaptive monitoring parameter group is manifested as follows: from the complete set of ten actual monitoring indicators, the corresponding indicator parameters are selected for weighted fusion calculation, while other indicator parameters are temporarily excluded from the fusion calculation within the current time window. When the emergency index is low, all ten actual monitoring indicators corresponding to the five monitoring parameters participate in the fusion calculation, achieving deep multi-parameter fusion to ensure monitoring accuracy. As the emergency index increases, the number of indicator parameters participating in the fusion calculation decreases sequentially, reducing computational complexity and shortening response time. This achieves an adaptive balance between monitoring accuracy and response time while ensuring continuous system operation.

[0087] Finally, the current time period's leakage monitoring urgency index is matched with the urgency index-monitoring parameter group association table to determine the appropriate monitoring parameter group for the current time period. Specifically, the matching method involves searching the urgency index interval to which the current urgency index belongs in the urgency index-monitoring parameter group association table, and then retrieving the corresponding monitoring parameter group as the appropriate monitoring parameter group. This table lookup method avoids online calculations and meets real-time requirements. The obtained appropriate monitoring parameter group will be used for data fusion calculations within the next monitoring adjustment time window, achieving a dynamic balance between monitoring resource consumption and risk response speed.

[0088] Further, after determining the adaptation monitoring parameter group, an adaptation monitoring scheme is generated by combining the adaptation weight ratios obtained in the aforementioned steps. Specifically, in this embodiment, generating the adaptation monitoring scheme by combining the adaptation weight ratios includes: Extract the indicator parameter weights corresponding to each adaptation monitoring parameter in the adaptation monitoring parameter group from the adaptation weight ratio; The weights of the mean index and the fluctuation coefficient index corresponding to each adaptive monitoring parameter are summed to obtain the combined weight of each adaptive monitoring parameter. The combined weights of the various adaptive monitoring parameters are normalized so that the sum of the normalized weights of the various adaptive monitoring parameters is 1, thus obtaining the parameter weight distribution. The parameter weight distribution and the adaptation weight ratio are combined to generate an adaptation monitoring scheme.

[0089] First, the weights of the corresponding indicator parameters in the adapted monitoring parameter group are extracted from the adaptation weight ratio. The adaptation weight ratio is the weight coefficient of each of the ten indicator parameters obtained after weight adjustment driven by hydrogen concentration difference by superimposing the baseline weights of the analytic hierarchy process in the aforementioned urgency assessment step. There is a correspondence between the ten indicator parameters and the five monitoring parameters: the average pressure and pressure fluctuation coefficient correspond to the pressure monitoring parameter, the average temperature and temperature fluctuation coefficient correspond to the temperature monitoring parameter, the average hydrogen concentration and hydrogen concentration fluctuation coefficient correspond to the hydrogen concentration monitoring parameter, the average pressure-temperature combined change rate and pressure-temperature combined change rate fluctuation coefficient correspond to the pressure-temperature combined change rate monitoring parameter, and the average hydrogen concentration difference and hydrogen concentration difference fluctuation coefficient correspond to the hydrogen concentration difference monitoring parameter. For each monitoring parameter in the adapted monitoring parameter group, its corresponding mean indicator weight and fluctuation coefficient indicator weight are extracted from the adaptation weight ratio.

[0090] Secondly, the weights of the mean index and the fluctuation coefficient index corresponding to each adapted monitoring parameter are summed to obtain the combined weight of each adapted monitoring parameter. Since each monitoring parameter corresponds to two index parameters, such as the pressure monitoring parameter corresponding to the pressure mean and the pressure fluctuation coefficient, the weights of these two indicators are added together to obtain the combined weight of the pressure monitoring parameter. The combined weight reflects the overall importance of the monitoring parameter in the urgency assessment. For extreme cases that include only a single monitoring parameter, such as the adapted monitoring parameter group corresponding to the highest urgency index interval only including the hydrogen concentration monitoring parameter, the combined weight of the hydrogen concentration monitoring parameter is equal to the sum of the weight of the hydrogen concentration mean index and the weight of the hydrogen concentration fluctuation coefficient index.

[0091] Next, the combined weights of each adapted monitoring parameter are normalized so that the sum of the normalized weights of all adapted monitoring parameters is 1, resulting in the parameter weight distribution. The reason for normalization is that the number of monitoring parameters in the adapted monitoring parameter group may be less than five, while the original adaptation weight ratio was designed for all ten indicator parameters, with the sum of each indicator weight being 1. When only some monitoring parameters are selected, the sum of the combined weights of the selected monitoring parameters is less than 1. In order to reasonably allocate the importance proportion of each monitoring parameter in subsequent monitoring schemes, the combined weights of the selected monitoring parameters need to be rescaled.

[0092] Specifically, suppose there are K selected monitoring parameters, where K is between 1 and 5, and the combined weight of the i-th monitoring parameter is W. i Then the normalized weight of the i-th monitoring parameter = W i / (W from 1 to K) j (The sum of the weights). After normalization, the sum of the weights of the selected K monitoring parameters is 1, and the weights maintain the same proportional relationship as the original combined weights. That is, the monitoring parameters with larger original combined weights still have larger weights after normalization.

[0093] Finally, the parameter weight distribution and adaptation weight ratio are integrated to generate an adaptive monitoring scheme. The adaptive monitoring scheme is a structured monitoring strategy configuration that includes at least the following three elements: First, a list of monitoring parameters to be collected, i.e., the monitoring parameters specified in the adaptive monitoring parameter group; second, the sampling frequency of each monitoring parameter, which can be set according to the normalized weight of the monitoring parameter. Monitoring parameters with higher weights have higher sampling frequencies; for example, the monitoring parameter with the highest normalized weight can be sampled every 5 seconds, and the monitoring parameter with the lowest weight can be sampled every 30 seconds; third, the weight configuration of each monitoring parameter in the next round of urgency assessment, i.e., the normalized parameter weight distribution. In addition, the adaptive monitoring scheme may also include auxiliary configuration information such as alarm thresholds for each monitoring parameter.

[0094] By generating an adaptive monitoring plan, data collection and urgency assessment can be performed according to the adaptive monitoring plan within the next monitoring adjustment time window, ultimately achieving adaptive adjustment of the monitoring strategy.

[0095] In summary, the embodiments of this application have at least the following technical effects: This invention first constructs a hydrogen valve twin model through simulation, using pressure and temperature parameters to predict the hydrogen concentration in the dynamic sealing zone of the stuffing box under normal sealing conditions, providing a dynamic baseline for leak detection. Second, it collects pressure, temperature, and hydrogen concentration data in real time using sensors, constructing a time series within a historical monitoring and adjustment window, and calculating the pressure-temperature joint change rate series and the hydrogen concentration difference series, achieving quantitative extraction of the coupling characteristics of multiple physical quantities. Third, it uses the analytic hierarchy process (AHP) to set the baseline weights of the indicator parameters, and dynamically adjusts the weights based on the hydrogen concentration difference series to obtain an appropriate weight ratio, enabling urgency assessment to adaptively adjust the importance weights of each indicator according to the actual degree of leak deviation.

[0096] Finally, by querying the correlation table based on the emergency index of leak monitoring, a suitable monitoring parameter group is determined. In low-risk situations, comprehensive multi-parameter monitoring is employed to accumulate data, while in high-risk situations, it is simplified to single hydrogen concentration monitoring to improve response speed, achieving a dynamic balance between monitoring resource consumption and risk response speed. This invention solves the technical problems in existing technologies, such as the difficulty in distinguishing between normal seepage and abnormal leaks, misjudgments and missed judgments due to unblended pressure and temperature parameters, and the lack of adaptive adjustment capabilities in monitoring strategies.

[0097] Example 2, as Figure 3 As shown, based on the same inventive concept as the intelligent monitoring method for hydrogen valve leakage based on multiple sensing parameters provided in Embodiment 1, this embodiment of the invention also provides an intelligent monitoring system for hydrogen valve leakage based on multiple sensing parameters, including: Model simulation module 11 is used to simulate and build a hydrogen valve twin model for predicting the current hydrogen concentration in the stuffing box dynamic sealing area based on pressure and temperature parameters. Data acquisition module 12 is used to collect the pressure, temperature and hydrogen concentration of the packing gland moving sealing area in real time through sensors during the operation of the high pressure shut-off valve, and obtain the pressure sequence, temperature sequence and hydrogen concentration sequence within the historical monitoring and adjustment time window; The sequence calculation module 13 is used to calculate the pressure-temperature joint change rate sequence based on the pressure sequence and temperature sequence, and to calculate the hydrogen concentration difference sequence based on the hydrogen concentration sequence and the predicted hydrogen concentration sequence obtained by simulation through the hydrogen valve twin model. Emergency assessment module 14 is used to assess the urgency of hydrogen leak monitoring based on the pressure sequence, temperature sequence, hydrogen concentration sequence, pressure-temperature combined rate of change sequence and hydrogen concentration difference sequence, and to determine the leak monitoring urgency index. The module adjusts the baseline weights set in advance using the analytic hierarchy process based on the hydrogen concentration difference sequence to obtain the adaptation weight ratio, and performs the urgency assessment based on the adaptation weight ratio. The scheme generation module 15 is used to query the pre-built emergency index-monitoring parameter group association table according to the leakage monitoring emergency index, determine the appropriate monitoring parameter group, generate an appropriate monitoring scheme in combination with the appropriate weight ratio, and perform hydrogen leakage monitoring of the stuffing gland dynamic sealing area in the next monitoring adjustment time window.

[0098] Specifically, the model simulation module 11 is used for: Simulations were used to construct a hydrogen valve twin model to predict the current hydrogen concentration in the moving seal zone of the stuffing box based on pressure and temperature parameters. This included: Simulation modeling was performed based on the structural parameters, material properties, and physicochemical characteristics of hydrogen in the dynamic sealing zone of the stuffing box of the high-pressure shut-off valve, and an initial twin model of the dynamic sealing zone of the stuffing box was established. Historical pressure data, historical temperature data, and corresponding historical hydrogen concentration data of the dynamic sealing zone of the high-pressure shut-off valve packing gland under different operating conditions were collected to construct a model training sample set. The initial twin model is trained using the model training sample set, so that the initial twin model learns the mapping relationship between pressure, temperature and hydrogen concentration in the stuffing box dynamic sealing area under normal sealing conditions. When the prediction error meets the preset accuracy threshold, the construction of the hydrogen valve twin model is completed. During the operation of the high-pressure shut-off valve, the real-time collected pressure and temperature parameters are input into the hydrogen valve twin model, and the predicted value of the current hydrogen concentration in the dynamic sealing zone of the stuffing box is output.

[0099] Specifically, the data acquisition module 12 is used for: The pressure, temperature, and hydrogen concentration in the moving seal area of ​​the stuffing box are collected in real time by sensors, including: A pressure sensor is installed at the medium inlet of the dynamic sealing zone of the stuffing box to collect the medium pressure in the dynamic sealing zone of the stuffing box in real time. Temperature sensors are installed on the outer wall of the moving sealing area of ​​the stuffing box to collect the ambient temperature of the moving sealing area of ​​the stuffing box in real time. A MEMS hydrogen concentration sensor is installed at the contact point between the valve stem and the packing in the moving sealing area of ​​the stuffing box to collect the local hydrogen concentration in the moving sealing area of ​​the stuffing box in real time.

[0100] Specifically, the sequence calculation module 13 is used for: A pressure-temperature joint rate of change sequence is calculated based on the pressure and temperature sequences. A hydrogen concentration difference sequence is calculated based on the hydrogen concentration sequence and the predicted hydrogen concentration sequence obtained through simulation using the hydrogen valve twin model, including: Calculate the rate of change of pressure and the rate of change of temperature at each time step for the pressure and temperature sequences, respectively, to obtain the pressure rate of change sequence and the temperature rate of change sequence. Based on the pressure change rate sequence and temperature change rate sequence, the pressure change rate and temperature change rate at each time are weighted and fused to obtain the pressure-temperature joint change rate at each time, which constitutes the pressure-temperature joint change rate sequence. The weights of the pressure change rate and temperature change rate are pre-calibrated based on the degree of influence of the pressure change rate and temperature change rate on the sealing failure of the stuffing box dynamic sealing area in historical working condition data. The hydrogen concentration difference at each moment is obtained by subtracting the predicted hydrogen concentration sequence from the hydrogen concentration sequence at each moment, thus forming a hydrogen concentration difference sequence.

[0101] Specifically, the emergency assessment module 14 is used for: The steps for setting baseline weights in advance using the analytic hierarchy process include: Configure a set of indicator parameters, wherein the set of indicator parameters includes average pressure, pressure fluctuation coefficient, average temperature, temperature fluctuation coefficient, average hydrogen concentration, hydrogen concentration fluctuation coefficient, average pressure-temperature combined change rate, pressure-temperature combined change rate fluctuation coefficient, average hydrogen concentration difference, and hydrogen concentration difference fluctuation coefficient. A hierarchical analysis judgment matrix is ​​constructed for the set of index parameters. By comparing the importance of each index parameter in the set of index parameters pairwise, and based on the degree of influence of each index parameter on the monitoring of hydrogen leakage in the dynamic sealing area of ​​the stuffing box, the relative importance of each index parameter is quantified and assigned using the 1-9 scale method. A consistency check is performed on the judgment matrix, and the consistency ratio is calculated. When the consistency ratio is less than 0.1, the judgment matrix passes the consistency check. The weight vector of each indicator parameter is calculated for the judgment matrix that passes the consistency check, and the weight vector is normalized to obtain the benchmark weight of each indicator parameter in the indicator parameter set.

[0102] Further, the adaptation weight ratio is obtained by adjusting the baseline weights based on the hydrogen concentration difference sequence, including: Calculate the mean value of the actual hydrogen concentration difference and the fluctuation coefficient of the actual hydrogen concentration difference within the historical monitoring and adjustment time window of the hydrogen concentration difference sequence; Based on the pre-calibrated baseline thresholds for the average hydrogen concentration difference and the fluctuation coefficient of the hydrogen concentration difference, the average actual hydrogen concentration difference and the fluctuation coefficient of the actual hydrogen concentration difference are normalized respectively to obtain the normalized value of the average hydrogen concentration difference and the normalized value of the fluctuation coefficient of the hydrogen concentration difference. The leakage deviation coefficient for the current period is obtained by weighted summing of the normalized value of the mean hydrogen concentration difference and the normalized value of the hydrogen concentration difference fluctuation coefficient; wherein, the weights of the normalized value of the mean hydrogen concentration difference and the normalized value of the hydrogen concentration difference fluctuation coefficient are pre-calibrated based on the degree of influence of the two indicators on the degree of leakage deviation. Based on the leakage deviation coefficient, the product of the leakage deviation coefficient and the preset adjustment coefficient is used as the weight adjustment range to adjust the benchmark weights corresponding to the average hydrogen concentration difference and the hydrogen concentration difference fluctuation coefficient in the set of index parameters. The adjusted weights do not exceed the preset weight upper limit. Based on the increased weight increment, the benchmark weights corresponding to the other indicator parameters are proportionally reduced according to their respective benchmark weights to obtain the appropriate weight ratio. The sum of the weights of each indicator parameter after adjustment remains 1.

[0103] Furthermore, based on the aforementioned pressure sequence, temperature sequence, hydrogen concentration sequence, pressure-temperature combined rate of change sequence, and hydrogen concentration difference sequence, an urgency assessment of hydrogen leak monitoring is conducted to determine the leak monitoring urgency index, including: Based on the pressure sequence, temperature sequence, hydrogen concentration sequence, pressure-temperature combined rate of change sequence, and hydrogen concentration difference sequence, the actual average pressure, actual pressure fluctuation coefficient, actual average temperature, actual temperature fluctuation coefficient, actual average hydrogen concentration, actual hydrogen concentration fluctuation coefficient, actual average pressure-temperature combined rate of change, actual pressure-temperature combined rate of change fluctuation coefficient, actual average hydrogen concentration difference, and actual hydrogen concentration difference fluctuation coefficient are calculated and used as the actual monitoring index set. Based on the normalized reference parameters determined in advance according to normal operating condition data, the actual monitoring indicators in the actual monitoring indicator set are subjected to segmented normalization processing to obtain the normalized value of each actual monitoring indicator. The normalized values ​​of each actual monitoring indicator in the actual monitoring indicator set are weighted according to the adaptation weight ratio to obtain the leakage monitoring emergency index. The leakage monitoring emergency index is positively correlated with each actual monitoring indicator in the actual monitoring indicator set. The larger the leakage monitoring emergency index, the higher the risk of hydrogen leakage in the dynamic sealing area of ​​the stuffing box.

[0104] The scheme generation module 15 is specifically used for: Based on the leak monitoring emergency index, a pre-built emergency index-monitoring parameter group association table is queried to determine the appropriate monitoring parameter group, including: Based on historical operating data and leakage test data, the value range of the leakage monitoring emergency index is divided into several emergency index intervals in advance. For each emergency index interval, a corresponding monitoring parameter group is configured, and an emergency index-monitoring parameter group association table is constructed. The number of monitoring parameters in the monitoring parameter group is inversely proportional to the urgency of the emergency index interval. Specifically, when the leak monitoring emergency index is in the lowest emergency index range, the corresponding monitoring parameter group is the maximum monitoring parameter group, which includes five monitoring parameters: pressure, temperature, hydrogen concentration, pressure-temperature combined change rate, and hydrogen concentration difference; when the leak monitoring emergency index is in the highest emergency index range, the corresponding monitoring parameter group is the minimum monitoring parameter group, which includes only the single monitoring parameter of hydrogen concentration. The selection of monitoring parameters in each monitoring parameter group is based on the sorting result of the benchmark weights, and the monitoring parameters are selected in descending order of the benchmark weights of the corresponding indicator parameters. During the operation of the high-pressure shut-off valve, the leakage monitoring emergency index for the current time period is matched with the emergency index-monitoring parameter group association table to determine the appropriate monitoring parameter group for the current time period. The appropriate monitoring parameter group is used to determine the parameter range participating in the weighted fusion calculation of leakage monitoring. The pressure sensor, temperature sensor and MEMS hydrogen concentration sensor continuously collect data during the operation of the high-pressure shut-off valve and are not affected by the switching of the appropriate monitoring parameter group.

[0105] Furthermore, an adaptation monitoring scheme is generated based on the aforementioned adaptation weight ratio, including: Extract the indicator parameter weights corresponding to each adaptation monitoring parameter in the adaptation monitoring parameter group from the adaptation weight ratio; The weights of the mean index and the fluctuation coefficient index corresponding to each adaptive monitoring parameter are summed to obtain the combined weight of each adaptive monitoring parameter. The combined weights of the various adaptive monitoring parameters are normalized so that the sum of the normalized weights of the various adaptive monitoring parameters is 1, thus obtaining the parameter weight distribution. The parameter weight distribution and the adaptation weight ratio are combined to generate an adaptation monitoring scheme.

Claims

1. A method for intelligent monitoring of hydrogen valve leakage based on multiple sensor parameters, characterized in that, The method is applicable to hydrogen leakage monitoring in the dynamic sealing zone of the stuffing box of a high-pressure shut-off valve in an industrial hydrogen pipeline, including: Simulations were used to construct a hydrogen valve twin model to predict the current hydrogen concentration in the stuffing gland dynamic sealing zone based on pressure and temperature parameters. During the operation of the high-pressure shut-off valve, the pressure, temperature and hydrogen concentration of the packing gland dynamic sealing area are collected in real time by sensors to obtain the pressure sequence, temperature sequence and hydrogen concentration sequence within the historical monitoring and adjustment time window. The pressure-temperature joint change rate sequence is calculated based on the pressure sequence and temperature sequence, and the hydrogen concentration difference sequence is calculated based on the hydrogen concentration sequence and the predicted hydrogen concentration sequence obtained by simulation through the hydrogen valve twin model. Based on the pressure sequence, temperature sequence, hydrogen concentration sequence, pressure-temperature combined rate of change sequence, and hydrogen concentration difference sequence, the urgency assessment of hydrogen leak monitoring is conducted to determine the leak monitoring urgency index. Specifically, the baseline weights set in advance using the analytic hierarchy process are adjusted according to the hydrogen concentration difference sequence to obtain the adaptation weight ratio, and the urgency assessment is conducted based on the adaptation weight ratio. Based on the leak monitoring emergency index, query the pre-built emergency index-monitoring parameter group association table to determine the appropriate monitoring parameter group, generate an appropriate monitoring scheme based on the appropriate weight ratio, and execute hydrogen leak monitoring of the stuffing gland dynamic sealing area in the next monitoring adjustment time window. The steps involved in setting the baseline weights using the analytic hierarchy process (AHP) beforehand include: Configure a set of indicator parameters, wherein the set of indicator parameters includes average pressure, pressure fluctuation coefficient, average temperature, temperature fluctuation coefficient, average hydrogen concentration, hydrogen concentration fluctuation coefficient, average pressure-temperature combined change rate, pressure-temperature combined change rate fluctuation coefficient, average hydrogen concentration difference, and hydrogen concentration difference fluctuation coefficient. A hierarchical analysis judgment matrix is ​​constructed for the set of index parameters. By comparing the importance of each index parameter in the set of index parameters pairwise, and based on the degree of influence of each index parameter on the monitoring of hydrogen leakage in the dynamic sealing area of ​​the stuffing box, the relative importance of each index parameter is quantified and assigned using the 1-9 scale method. A consistency check is performed on the judgment matrix, and the consistency ratio is calculated. When the consistency ratio is less than 0.1, the judgment matrix passes the consistency check. The weight vector of each indicator parameter is calculated for the judgment matrix that passes the consistency check, and the weight vector is normalized to obtain the benchmark weight of each indicator parameter in the indicator parameter set. The process of adjusting the baseline weights based on the hydrogen concentration difference sequence to obtain the adaptation weight ratio includes: Calculate the mean value of the actual hydrogen concentration difference and the fluctuation coefficient of the actual hydrogen concentration difference within the historical monitoring and adjustment time window of the hydrogen concentration difference sequence; Based on the pre-calibrated baseline thresholds for the average hydrogen concentration difference and the fluctuation coefficient of the hydrogen concentration difference, the average actual hydrogen concentration difference and the fluctuation coefficient of the actual hydrogen concentration difference are normalized respectively to obtain the normalized value of the average hydrogen concentration difference and the normalized value of the fluctuation coefficient of the hydrogen concentration difference. The leakage deviation coefficient for the current period is obtained by weighted summing of the normalized value of the mean hydrogen concentration difference and the normalized value of the hydrogen concentration difference fluctuation coefficient; wherein, the weights of the normalized value of the mean hydrogen concentration difference and the normalized value of the hydrogen concentration difference fluctuation coefficient are pre-calibrated based on the degree of influence of the two indicators on the degree of leakage deviation. Based on the leakage deviation coefficient, the product of the leakage deviation coefficient and the preset adjustment coefficient is used as the weight adjustment range to adjust the benchmark weights corresponding to the average hydrogen concentration difference and the hydrogen concentration difference fluctuation coefficient in the set of index parameters. The adjusted weights do not exceed the preset weight upper limit. Based on the increased weight increment, the benchmark weights corresponding to the other indicator parameters are proportionally reduced according to their respective benchmark weights to obtain the appropriate weight ratio. The sum of the weights of each indicator parameter after adjustment remains 1. Specifically, based on the pressure sequence, temperature sequence, hydrogen concentration sequence, pressure-temperature combined rate of change sequence, and hydrogen concentration difference sequence, an urgency assessment of hydrogen leak monitoring is conducted to determine the leak monitoring urgency index, including: Based on the pressure sequence, temperature sequence, hydrogen concentration sequence, pressure-temperature combined rate of change sequence, and hydrogen concentration difference sequence, the actual average pressure, actual pressure fluctuation coefficient, actual average temperature, actual temperature fluctuation coefficient, actual average hydrogen concentration, actual hydrogen concentration fluctuation coefficient, actual average pressure-temperature combined rate of change, actual pressure-temperature combined rate of change fluctuation coefficient, actual average hydrogen concentration difference, and actual hydrogen concentration difference fluctuation coefficient are calculated and used as the actual monitoring index set. Based on the normalized reference parameters determined in advance according to normal operating condition data, the actual monitoring indicators in the actual monitoring indicator set are subjected to segmented normalization processing to obtain the normalized value of each actual monitoring indicator. The normalized values ​​of each actual monitoring indicator in the actual monitoring indicator set are weighted according to the adaptation weight ratio to obtain the leakage monitoring emergency index. The leakage monitoring emergency index is positively correlated with each actual monitoring indicator in the actual monitoring indicator set. The larger the leakage monitoring emergency index, the higher the risk of hydrogen leakage in the dynamic sealing area of ​​the stuffing box.

2. The intelligent monitoring method for hydrogen valve leakage based on multiple sensor parameters according to claim 1, characterized in that, Simulations were used to construct a hydrogen valve twin model to predict the current hydrogen concentration in the moving seal zone of the stuffing box based on pressure and temperature parameters. This included: Simulation modeling was performed based on the structural parameters, material properties, and physicochemical characteristics of hydrogen in the dynamic sealing zone of the stuffing box of the high-pressure shut-off valve, and an initial twin model of the dynamic sealing zone of the stuffing box was established. Historical pressure data, historical temperature data, and corresponding historical hydrogen concentration data of the dynamic sealing zone of the high-pressure shut-off valve packing gland under different operating conditions were collected to construct a model training sample set. The initial twin model is trained using the model training sample set, so that the initial twin model learns the mapping relationship between pressure, temperature and hydrogen concentration in the stuffing box dynamic sealing area under normal sealing conditions. When the prediction error meets the preset accuracy threshold, the construction of the hydrogen valve twin model is completed. During the operation of the high-pressure shut-off valve, the real-time collected pressure and temperature parameters are input into the hydrogen valve twin model, and the predicted value of the current hydrogen concentration in the dynamic sealing zone of the stuffing box is output.

3. The intelligent monitoring method for hydrogen valve leakage based on multiple sensing parameters according to claim 1, characterized in that, The pressure, temperature, and hydrogen concentration in the moving seal area of ​​the stuffing box are collected in real time by sensors, including: A pressure sensor is installed at the medium inlet of the dynamic sealing zone of the stuffing box to collect the medium pressure in the dynamic sealing zone of the stuffing box in real time. Temperature sensors are installed on the outer wall of the moving sealing area of ​​the stuffing box to collect the ambient temperature of the moving sealing area of ​​the stuffing box in real time. A MEMS hydrogen concentration sensor is installed at the contact point between the valve stem and the packing in the moving sealing area of ​​the stuffing box to collect the local hydrogen concentration in the moving sealing area of ​​the stuffing box in real time.

4. The intelligent monitoring method for hydrogen valve leakage based on multiple sensor parameters according to claim 1, characterized in that, A pressure-temperature joint rate of change sequence is calculated based on the pressure and temperature sequences. A hydrogen concentration difference sequence is calculated based on the hydrogen concentration sequence and the predicted hydrogen concentration sequence obtained through simulation using the hydrogen valve twin model, including: Calculate the rate of change of pressure and the rate of change of temperature at each time step for the pressure and temperature sequences, respectively, to obtain the pressure rate of change sequence and the temperature rate of change sequence. Based on the pressure change rate sequence and temperature change rate sequence, the pressure change rate and temperature change rate at each time are weighted and fused to obtain the pressure-temperature joint change rate at each time, which constitutes the pressure-temperature joint change rate sequence. The weights of the pressure change rate and temperature change rate are pre-calibrated based on the degree of influence of the pressure change rate and temperature change rate on the sealing failure of the stuffing box dynamic sealing area in historical working condition data. The hydrogen concentration difference at each moment is obtained by subtracting the predicted hydrogen concentration sequence from the hydrogen concentration sequence at each moment, thus forming a hydrogen concentration difference sequence.

5. The intelligent monitoring method for hydrogen valve leakage based on multiple sensor parameters according to claim 1, characterized in that, Based on the leak monitoring emergency index, a pre-built emergency index-monitoring parameter group association table is queried to determine the appropriate monitoring parameter group, including: Based on historical operating data and leakage test data, the value range of the leakage monitoring emergency index is divided into several emergency index intervals in advance. For each emergency index interval, a corresponding monitoring parameter group is configured, and an emergency index-monitoring parameter group association table is constructed. The number of monitoring parameters in the monitoring parameter group is inversely proportional to the urgency of the emergency index interval. Specifically, when the leak monitoring emergency index is in the lowest emergency index range, the corresponding monitoring parameter group is the maximum monitoring parameter group, which includes five monitoring parameters: pressure, temperature, hydrogen concentration, pressure-temperature combined change rate, and hydrogen concentration difference; when the leak monitoring emergency index is in the highest emergency index range, the corresponding monitoring parameter group is the minimum monitoring parameter group, which includes only the single monitoring parameter of hydrogen concentration. The selection of monitoring parameters in each monitoring parameter group is based on the sorting result of the benchmark weights, and the monitoring parameters are selected in descending order of the benchmark weights of the corresponding indicator parameters. During the operation of the high-pressure shut-off valve, the leakage monitoring emergency index for the current time period is matched with the emergency index-monitoring parameter group association table to determine the appropriate monitoring parameter group for the current time period. The appropriate monitoring parameter group is used to determine the parameter range participating in the weighted fusion calculation of leakage monitoring. The pressure sensor, temperature sensor and MEMS hydrogen concentration sensor continuously collect data during the operation of the high-pressure shut-off valve and are not affected by the switching of the appropriate monitoring parameter group.

6. The intelligent monitoring method for hydrogen valve leakage based on multiple sensor parameters according to claim 5, characterized in that, An adaptation monitoring scheme is generated by combining the aforementioned adaptation weight ratios, including: Extract the indicator parameter weights corresponding to each adaptation monitoring parameter in the adaptation monitoring parameter group from the adaptation weight ratio; The weights of the mean index and the fluctuation coefficient index corresponding to each adaptive monitoring parameter are summed to obtain the combined weight of each adaptive monitoring parameter. The combined weights of the various adaptive monitoring parameters are normalized so that the sum of the normalized weights of the various adaptive monitoring parameters is 1, thus obtaining the parameter weight distribution. The parameter weight distribution and the adaptation weight ratio are combined to generate an adaptation monitoring scheme.

7. A hydrogen valve leakage intelligent monitoring system based on multiple sensor parameters, characterized in that, The method for performing the intelligent monitoring of hydrogen valve leakage based on multiple sensing parameters as described in any one of claims 1-6 includes: The model simulation module is used to simulate and build a hydrogen valve twin model for predicting the current hydrogen concentration in the stuffing gland dynamic sealing area based on pressure and temperature parameters. The data acquisition module is used to collect the pressure, temperature and hydrogen concentration of the packing gland moving sealing area in real time through sensors during the operation of the high-pressure shut-off valve, and obtain the pressure sequence, temperature sequence and hydrogen concentration sequence within the historical monitoring and adjustment time window; The sequence calculation module is used to calculate the pressure-temperature joint change rate sequence based on the pressure sequence and temperature sequence, and to calculate the hydrogen concentration difference sequence based on the hydrogen concentration sequence and the predicted hydrogen concentration sequence obtained by simulation through the hydrogen valve twin model. The emergency assessment module is used to assess the urgency of hydrogen leak monitoring based on the pressure sequence, temperature sequence, hydrogen concentration sequence, pressure-temperature combined rate of change sequence, and hydrogen concentration difference sequence, and to determine the leak monitoring urgency index. Specifically, the module adjusts the baseline weights set in advance using the analytic hierarchy process based on the hydrogen concentration difference sequence to obtain the appropriate weight ratio, and performs the urgency assessment based on the appropriate weight ratio. The scheme generation module is used to query the pre-built emergency index-monitoring parameter group association table according to the leakage monitoring emergency index, determine the appropriate monitoring parameter group, generate an appropriate monitoring scheme in combination with the appropriate weight ratio, and execute hydrogen leakage monitoring of the stuffing gland dynamic sealing area in the next monitoring adjustment time window.

Images