Online detection method for sealing performance of sealing element

Abstract

The invention discloses an online detection method for the sealing performance of a sealing element, particularly relates to the technical field of quality detection of industrial sealing elements, and is used for solving the problems of reference value deviation and misjudgment caused by neglecting the influence of mounting stress in an existing detection method. According to the method, a sealed contact interface phase field model is constructed by collecting installation stress data in real time, and crack network permeability is analyzed to predict a leakage diffusion path; high / low stress sensitive areas are divided according to the spatial correlation between the stress gradient and the leakage density, and differential signal acquisition is implemented; extracting a bar code entropy value through topological data analysis, generating a compensation priority sequence, and dynamically correcting a detection reference; and finally, based on the regional weight fusion compensation signal, outputting a high-precision pressure distribution and leakage flow trend result. Cross-field coupling analysis of installation stress and leakage signals is achieved, and the accuracy and reliability of sealing performance detection under complex working conditions are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial seal quality inspection, and more specifically, the present invention relates to an online inspection method for the sealing performance of seals. Background Art

[0002] In the field of industrial equipment sealing performance inspection, the online inspection technology of seals has been widely applied to the quality control link of production lines; existing technologies usually monitor pressure, flow rate or medium leakage parameters under the operating state of the sealing system to judge whether the seals meet the design requirements. Such methods rely on preset inspection reference values, such as initial pressure thresholds or leakage rate standards, as the basis for judging whether the sealing performance is qualified. However, the external mechanical forces (such as bolt tightening, flange assembly, etc.) received by the seals during the installation process may cause uneven internal stress distribution, and this stress state may still persist after the sealing system starts to work and affect the actual contact state of the sealing interface.

[0003] The limitation of the current online inspection method for sealing performance is that the setting of the inspection reference value does not fully consider the potential impact of the installation process on the initial state of the seals. Due to the microscopic deformation or local stress residue of the sealing interface caused by the installation stress, the initial parameters obtained by the online inspection system may deviate from the true working state of the seals, thereby causing a systematic deviation between the inspection reference value and the actual sealing performance. This deviation will reduce the accuracy of leakage detection, especially in the scenario of online inspection immediately after the seals are installed, which may lead to misjudgment or missed inspection and affect the reliability of quality control. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an online inspection method for the sealing performance of seals to solve the problems raised in the above background art.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] An online inspection method for the sealing performance of seals, comprising the following steps:

[0007] S1. During the installation process of the seal, obtain the real-time installation stress data between the seal and the installation structure;

[0008] S2. Based on the real-time installation stress data, construct a phase field model of the sealing contact interface, and determine the diffusion path of the leakage medium along the sealing contact interface through crack network permeability analysis;

[0009] S3. According to the phase field model and the diffusion path, divide the sealing contact interface into a high stress sensitive area and a low stress sensitive area;

[0010] S4. Implement high-frequency signal acquisition for high-stress sensitive areas and low-frequency signal acquisition for low-stress sensitive areas to obtain the pressure signal of the sealed contact interface and the flow signal of the leakage medium.

[0011] S5. Map the pressure signal and the flow signal into high-dimensional point cloud data, calculate the persistent homology barcode of the high-dimensional point cloud data and extract the barcode entropy value, and assign regional weights according to the spatio-temporal distribution correlation of the barcode entropy value in the high-stress sensitive area and the low-stress sensitive area to generate a compensation priority sequence.

[0012] S6. Perform dynamic compensation on high-weight areas based on the compensation priority sequence and output the corrected sealed performance detection result.

[0013] In a preferred embodiment, during the installation process of the seal, obtain the real-time installation stress data between the seal and the installation structure, including:

[0014] Arrange multiple strain gauges on the contact surface between the seal and the installation structure to form a stress sensing network.

[0015] Real-time collect the dynamic strain data during the seal installation process through a distributed fiber optic sensor.

[0016] Based on the preset stress-strain relationship curve, perform dynamic calibration on the dynamic strain data to generate real-time installation stress data.

[0017] In a preferred embodiment, construct a phase field model of the sealed contact interface based on the real-time installation stress data, and determine the diffusion path of the leakage medium along the sealed contact interface through crack network permeability analysis, including:

[0018] Input the three-dimensional stress field components of the real-time installation stress data as the stress driving term of the phase field model. The control equations of the phase field model include a coupled stress balance equation and a phase field order parameter evolution equation. The crack surface energy density function in the phase field order parameter evolution equation is defined based on the fracture toughness parameters of the sealed contact interface material.

[0019] Use an explicit time integration algorithm to iteratively solve the phase field model to generate the crack propagation path and the crack network topology structure of the sealed contact interface. The crack network topology structure includes geometric characteristic parameters such as crack branch angles and curvature radii.

[0020] Based on the geometric characteristic parameters of the crack branch angles and curvature radii, calculate the equivalent permeability tensor of the crack network. The calculation of the equivalent permeability tensor includes weighted statistical averaging of the crack branch angles and curvature radii.

[0021] Determine the diffusion path of the leakage medium along the sealed contact interface according to the matching degree between the maximum principal value direction of the equivalent permeability tensor and the main branch direction in the crack network topology structure.

[0022] In a preferred embodiment, according to the phase field model and the diffusion path, the sealed contact interface is divided into a high stress sensitive region and a low stress sensitive region, including:

[0023] Calculate the stress gradient distribution of the sealed contact interface based on the phase field model, and the stress gradient distribution includes the normal stress gradient and the shear stress gradient components;

[0024] Extract the spatial density distribution of the diffusion path, and mark the overlapping region of the stress gradient distribution and the diffusion path density distribution as the candidate region of the high stress sensitive region;

[0025] Generate the boundary division rule of the high stress sensitive region and the low stress sensitive region according to the geometric relationship between the stress gradient component and the diffusion path density in the candidate region;

[0026] Output the coordinate sets of the high stress sensitive region and the low stress sensitive region based on the boundary division rule, and the generation of the coordinate sets uses a spatial clustering algorithm to aggregate the regions that meet the conditions.

[0027] In a preferred embodiment, the boundary division rule is that the region where the stress gradient component exceeds the preset gradient threshold and the diffusion path density exceeds the preset density threshold is defined as the high stress sensitive region.

[0028] In a preferred embodiment, high-frequency signal acquisition is performed on the high stress sensitive region, and low-frequency signal acquisition is performed on the low stress sensitive region to obtain the pressure signal of the sealed contact interface and the flow signal of the leakage medium, including:

[0029] Configure the deployment positions of the pressure sensor array and the flow sensor based on the coordinate sets of the high stress sensitive region and the low stress sensitive region;

[0030] Allocate the high-frequency signal acquisition frequency according to the risk level of the high stress sensitive region, and the risk level is quantified by the product of the stress gradient component and the diffusion path density;

[0031] Allocate the low-frequency signal acquisition frequency to the low stress sensitive region;

[0032] Synchronize the acquisition timestamps of the pressure sensor array and the flow sensor to obtain the pressure signal of the sealed contact interface and the flow signal of the leakage medium in real time.

[0033] In a preferred embodiment, map the pressure signal and the flow signal into high-dimensional point cloud data, calculate the persistent homology barcode of the high-dimensional point cloud data and extract the barcode entropy value, and allocate the region weights according to the spatio-temporal distribution correlation of the barcode entropy value in the high stress sensitive region and the low stress sensitive region to generate the compensation priority sequence, including:

[0034] Slice the pressure signal and the flow signal according to a time window, and map the signal sequence within each time window into high-dimensional point cloud data by the time-delay embedding method;

[0035] Perform persistent homology analysis on the high-dimensional point cloud data to generate a persistent homology barcode. The generation of the persistent homology barcode adopts the Vietoris-Rips complex construction method;

[0036] Calculate the barcode entropy value based on the length distribution of the persistent homology barcode. The barcode entropy value is the Shannon entropy, and a regional sensitivity weight coefficient is introduced during the calculation. The regional sensitivity weight coefficient is the ratio of the average barcode lifetimes of the high-stress sensitive region and the low-stress sensitive region;

[0037] Allocate regional weights according to the spatio-temporal distribution correlation of the barcode entropy value in the high-stress sensitive region and the low-stress sensitive region;

[0038] Generate a compensation priority sequence in descending order of the regional weights. The generation rule of the compensation priority sequence is that regions with the same weight are sorted preferentially according to the high-frequency signal acquisition regions.

[0039] In a preferred embodiment, the embedding dimension and the delay parameter of the time-delay embedding method are adaptively adjusted according to the dynamic ratio of the high-frequency signal acquisition frequency in the high-stress sensitive region to the low-frequency signal acquisition frequency in the low-stress sensitive region.

[0040] In a preferred embodiment, the spatio-temporal distribution correlation is calculated through the correlation between spatial grid division and time window sliding. The division of the spatial grid is aligned with the clustering boundary of the high-stress sensitive region, and the length of the time window is inversely proportional to the signal acquisition frequency.

[0041] In a preferred embodiment, perform dynamic compensation on the high-weight regions based on the compensation priority sequence, and output the corrected seal performance detection results, including:

[0042] Extract the high-weight regions with higher rankings from the compensation priority sequence. The number of high-weight regions is determined according to a preset proportion range of the total number of regions in the compensation priority sequence;

[0043] Perform dynamic compensation on the pressure signals of the high-weight regions. The dynamic compensation method is to adjust the compensation coefficient based on the pressure signal deviation value;

[0044] Perform dynamic compensation on the flow signals of the high-weight regions. The dynamic compensation method is to adjust the compensation coefficient based on the flow signal deviation value;

[0045] Fuse the compensated pressure signals and flow signals with the original signals in the uncompensated regions by weighting according to the product of the regional weights and the compensation coefficients;

[0046] Output the corrected seal performance detection results after weighted fusion, including the pressure distribution contour map and the leakage flow trend curve.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. By obtaining the installation stress data in real time and constructing a phase field model, it is possible to accurately quantify the microscopic deformation and crack propagation path of the seal interface under the action of installation stress. Based on the differential signal acquisition strategy for stress-sensitive area division and the dynamic adaptation mechanism combining high-frequency and low-frequency monitoring, the problem of detection benchmark distortion caused by residual installation stress is effectively solved, and the real-time performance and spatial resolution of leakage detection are significantly improved. It is especially suitable for the scenario of immediate detection after installation, avoiding misjudgment or missed detection caused by local stress concentration.

[0049] 2. By generating a compensation priority sequence through topological data analysis and dynamically associating the spatio-temporal distribution characteristics of leakage signals with regional weights, the adaptive correction of the detection benchmark is realized. Through multi-physical field coupling modeling and dynamic signal fusion, the strong consistency between the detection results and the actual working conditions of the seal is ensured, providing high-reliability support for the seal performance evaluation in complex installation environments, while reducing the amount of redundant data processing and optimizing the detection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a flowchart of an on-line detection method for the seal performance of a seal of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0052] Embodiment: Figure 1 An on-line detection method for the seal performance of a seal of the present invention is given, including the following steps:

[0053] S1. During the installation process of the seal, obtain the real-time installation stress data between the seal and the installation structure;

[0054] S2. Based on the real-time installation stress data, construct a phase field model of the seal contact interface, and determine the diffusion path of the leakage medium along the seal contact interface through crack network permeability analysis;

[0055] S3. According to the phase field model and the diffusion path, divide the seal contact interface into a high stress-sensitive area and a low stress-sensitive area;

[0056] S4. Implement high-frequency signal acquisition for high-stress sensitive areas and low-frequency signal acquisition for low-stress sensitive areas to obtain the pressure signal of the sealed contact interface and the flow signal of the leakage medium.

[0057] S5. Map the pressure signal and the flow signal into high-dimensional point cloud data, calculate the persistent homology barcode of the high-dimensional point cloud data and extract the barcode entropy value, and assign regional weights according to the spatio-temporal distribution correlation of the barcode entropy value in the high-stress sensitive area and the low-stress sensitive area to generate a compensation priority sequence.

[0058] S6. Perform dynamic compensation on the high-weight areas based on the compensation priority sequence and output the corrected sealed performance detection result.

[0059] During the installation process of the seal, the specific implementation method for obtaining the real-time installation stress data between the seal and the installation structure is as follows: Arrange multiple resistive strain gauges on the contact surface between the seal and the installation structure to form a stress sensing network. The resistive strain gauges cover the key areas of the sealed contact interface in an equally spaced grid layout. The key areas include the area around the bolt hole, the flange edge contact area, and the area where the geometry of the seal changes. The area around the bolt hole is defined as an annular area with a radius of 10 mm centered on the bolt hole center. The flange edge contact area is defined as a strip area where the flange and the seal actually contact with a width of 5 mm. The area where the geometry changes includes the groove and boss structures of the seal. The layout density of the resistive strain gauges is dynamically adjusted according to the stress concentration degree of the sealed contact interface. The stress concentration degree is determined by pre-judging through finite element analysis. The finite element analysis uses ANSYS software to establish a three-dimensional model of the sealed contact interface and apply the bolt pre-tightening force load, and outputs the stress distribution cloud map to guide the strain gauge layout.

[0060] Real-time collect the dynamic strain data during the installation process of the seal through a distributed fiber optic sensor. The distributed fiber optic sensor is a sensor array based on fiber Bragg gratings. The fiber Bragg grating sensor array is laid crosswise along the circumferential and radial directions of the sealed contact interface. The circumferential laying path is to arrange a fiber every 30 degrees, and the radial laying path is to arrange a fiber every 5 mm along the radius direction of the seal. The laying path of the fiber Bragg grating sensor array is complementary to the grid layout of the resistive strain gauges to achieve spatial redundancy verification of the strain data. The acquisition frequency of the dynamic strain data is synchronized with the loading rate of the seal installation process. The loading rate of the installation process includes the application rate of the bolt tightening torque and the pressing speed of the flange assembly. The acquisition frequency of the dynamic strain data is set to 1000 to 5000 times per second, and the specific value is adaptively matched according to the real-time adjustment of the installation process. For example, when the application rate of the bolt tightening torque exceeds 5 N·m per second, the acquisition frequency is increased to 5000 times per second.

[0061] Based on the preset stress-strain relationship curve, the dynamic strain data is dynamically calibrated to generate real-time installation stress data. The stress-strain relationship curve is obtained by jointly calibrating the uniaxial tensile test and shear test of the sealing material. The uniaxial tensile test applies axial tension to the sealing material specimen on a universal testing machine, and the tension range is 0 to 120% of the material yield strength. The shear test uses a double shear fixture to apply an in-plane shear load with a loading rate of 0.5 mm per second. The test environment temperature covers the operating temperature range of the seal (-40°C to 150°C), and the humidity range is 30% to 90% relative humidity.

[0062] The stress-strain relationship curve is stored in the non-volatile memory of the signal processing terminal in the form of a two-dimensional interpolation table. The dynamic calibration process includes the following steps: converting the original wavelength offset collected by the distributed optical fiber sensor into a strain value. The conversion formula between the wavelength offset and the strain value is Δλ / λ=K·ε, where Δλ is the wavelength offset, λ is the initial center wavelength of the fiber Bragg grating, K is the strain sensitivity coefficient of the fiber Bragg grating, and ε is the strain value.

[0063] The validity of the strain value is verified according to the resistance change rate of the resistive strain gauge. The relationship between the resistance change rate ΔR / R and the strain value is ΔR / R=G·ε, where G is the strain gauge sensitivity coefficient. If the strain value deviation calculated by the two sensors exceeds 5%, it is determined to be an abnormal data point; the abnormal data points exceeding the preset strain threshold range are eliminated, and the preset strain threshold range is 50% to 80% of the yield strain of the sealing material. The rule for eliminating abnormal data points is that the strain value exceeds the threshold within three consecutive sampling periods and the spatial correlation with the adjacent sensor data is lower than the preset correlation coefficient. The preset correlation coefficient is 0.7 to 0.9, and the spatial correlation is determined by calculating the Pearson correlation coefficient between the abnormal data point and the adjacent 8 sensor data.

[0064] The verified strain data is matched with the stress-strain relationship curve through the interpolation algorithm to generate the real-time installation stress data of the sealing contact interface. The interpolation algorithm is a bilinear interpolation algorithm. The input of the bilinear interpolation algorithm is the spatial coordinates and timestamp of the strain value, and the output is the stress component of the corresponding coordinate point. The stress component includes normal stress and shear stress.

[0065] The specific steps of the bilinear interpolation algorithm are as follows: for a certain coordinate point (x, y), select the strain values ​​ε1, ε2, ε3, ε4 of the four nearest neighboring resistive strain gauges around it, and calculate the interpolation strain by distance weighting: ε(x, y) = (w1ε1+w2ε2+w3ε3+w4ε4) / (w1+w2+w3+w4); ε(x, y) represents the interpolation strain; the weight is: wi = 1 / di 2 ; wi represents the weight of the i-th resistive strain gauge, and di is the Euclidean distance between the coordinate point and the i-th resistive strain gauge.

[0066] The real-time installed stress data is stored in the form of a three-dimensional stress field with a spatial resolution of 1 mm to 5 mm and a time resolution of 1 ms to 10 ms. The coordinate system of the three-dimensional stress field is aligned with the local coordinate system of the seal contact interface. The origin of the local coordinate system is located at the geometric center of the seal contact interface, and the axial direction coincides with the axis of symmetry of the seal.

[0067] The following is the specific implementation method for constructing a phase field model of the seal contact interface based on the real-time installed stress data and determining the diffusion path of the leakage medium along the seal contact interface through crack network permeability analysis: The three-dimensional stress field components of the real-time installed stress data are input as the stress driving term of the phase field model. The control equations of the phase field model include a coupled stress balance equation and a phase field order parameter evolution equation. The stress balance equation is the equilibrium equation of linear elasticity mechanics, which is used to calculate the displacement field and stress field distribution of the seal contact interface under the action of the installed stress. The phase field order parameter evolution equation is a diffusion-type equation based on the Ginzburg-Landau theory, which is used to describe the process of crack initiation and propagation. The crack surface energy density function in the phase field order parameter evolution equation is defined based on the fracture toughness parameters of the seal contact interface material. The fracture toughness parameters are obtained by measuring through a standard three-point bending test. The specimen size of the three-point bending test is 50 mm in length, 10 mm in width, 10 mm in height, the span is 40 mm, the loading rate is 0.5 mm per minute. The form of the crack surface energy density function is a quadratic polynomial function, and the coefficient of the quadratic term is proportional to the fracture toughness parameter. The stress driving term is calculated through the three-dimensional stress field components of the real-time installed stress data. The three-dimensional stress field components include normal stress and shear stress. The calculation formula of the stress driving term is a linear combination of the stress field components, and the weight coefficients of the linear combination are adjusted according to the anisotropic characteristics of the seal contact interface material. The anisotropic characteristics are determined by the stress-strain curve differences between the uniaxial tensile test and the shear test of the seal material. The loading rate of the uniaxial tensile test is 1 mm per minute, and the shear test uses a double-shear fixture to apply in-plane shear load with a loading rate of 0.5 mm per minute.

[0068] The phase field model is iteratively solved using an explicit time integration algorithm. The explicit time integration algorithm is the central difference method. The time step is adaptively adjusted according to the stability condition of the phase field order parameter evolution equation. The stability condition is that the time step is less than or equal to the ratio of the square of the grid size to the material diffusion coefficient. The grid size ranges from 1 mm to 5 mm, and the material diffusion coefficient is obtained by experimental calibration of the thermal diffusion coefficient of the seal material. The crack propagation path and the crack network topology structure of the sealed contact interface are generated. The crack network topology structure includes geometric characteristic parameters such as crack branch angle and curvature radius. The crack branch angle is defined as the angle between two branches at the crack bifurcation, and the curvature radius is defined as the fitting circular arc radius of the curved section of the crack path. The geometric characteristic parameters of the crack branch angle and curvature radius are extracted from the crack propagation path through an image processing algorithm. The image processing algorithm is a skeletonization algorithm based on Canny edge detection. After the crack centerline is extracted by the skeletonization algorithm, the crack branch angle and curvature radius are fitted by the least squares method. The residual threshold of the least squares fitting is set to 0.1 mm, and the fitting results exceeding the threshold are judged as invalid and recalculated.

[0069] Based on the geometric characteristic parameters of the crack branch angle and curvature radius, the equivalent permeability tensor of the crack network is calculated. The calculation of the equivalent permeability tensor includes weighted statistical averaging of the crack branch angle and curvature radius. The weighting coefficients are assigned according to the contribution ratio of the crack branch angle and curvature radius to the fluid flow resistance. The contribution ratio is determined by hydrodynamic simulation. The specific method is as follows: crack element models with different branch angles and curvature radii are established. The size of the crack element model is 5 mm in length and 0.1 mm in width. A constant pressure difference of 1 MPa is applied and the flow rate is calculated. The flow rate is linearly related to the reciprocal of the branch angle and curvature radius. The slope of the linear relationship is the contribution ratio. For example, when the branch angle is 90 degrees, the contribution ratio coefficient is 1.0. For every 10-degree decrease in the branch angle, the coefficient increases by 0.1. When the curvature radius is 1 mm, the contribution ratio coefficient is 1.0. For every 0.5-mm increase in the curvature radius, the coefficient decreases by 0.2. The calculation formula for weighted statistical averaging is the sum of the permeability components of each crack element multiplied by their contribution ratio coefficients, and then divided by the sum of the total contribution ratio coefficients. The unit of the permeability component is Darcy, and the sum of the total contribution ratio coefficients is obtained by accumulating the contribution ratio coefficients of all crack elements.

[0070] According to the matching degree between the direction of the maximum principal value of the equivalent permeability tensor and the main branch direction in the crack network topology, determine the diffusion path of the leakage medium along the sealing contact interface. The matching degree is quantified by the cosine value of the direction angle. The cosine value of the direction angle is the cosine of the angle between the direction of the maximum principal value of the permeability tensor and the main branch direction of the crack. When the cosine value of the direction angle is greater than or equal to 0.9, it is determined as a match, and the diffusion path extends along this direction. The main branch direction is the extension direction of the branch path with the highest connectivity in the crack network. The connectivity is calculated by the product of the crack branch length and the number of adjacent crack units. For example, if the length of a branch path is 5 millimeters and it connects 3 adjacent crack units, the connectivity is 15. Select the branch path with the maximum connectivity as the main branch direction. The main branch direction is obtained by fitting the extension trend line of the branch path using the least squares method. The fitting residual threshold of the trend line is set to 0.2 millimeters.

[0071] In step S2, a phase field model is constructed based on the real-time installation stress data, and the leakage path is predicted based on the crack network permeability analysis. Its rationality lies in that the phase field model can couple the stress distribution and crack evolution. Compared with the prior art, in this step, the permeability calculation is dynamically corrected through the correlation between the stress driving term and the crack geometric parameters (branch angle, curvature radius), solving the problem of leakage path prediction deviation caused by ignoring the installation stress. The parameter linkage mechanism across the mechanical and seepage fields improves the prediction accuracy.

[0072] The specific implementation of dividing the sealing contact interface into high-stress sensitive areas and low-stress sensitive areas according to the phase field model and the diffusion path is as follows: Calculate the stress gradient distribution of the sealing contact interface based on the phase field model. The stress gradient distribution includes the normal stress gradient and shear stress gradient components. The normal stress gradient is the change rate of the normal stress of the sealing contact interface with respect to the spatial coordinates, and the shear stress gradient is the change rate of the shear stress with respect to the spatial coordinates. The calculation method of the stress gradient distribution is to perform a spatial partial derivative operation on the three-dimensional stress field output by the phase field model. The spatial partial derivative operation uses the central difference method, and the grid size is from 1 millimeter to 5 millimeters. The step size of the central difference method is the same as the grid size. For example, when the grid size is 2 millimeters, the step size is set to 2 millimeters. Extract the spatial density distribution of the diffusion path. The spatial density distribution of the diffusion path is defined as the proportion of the length of the diffusion path in the unit area of the sealing contact interface. The diffusion path length is measured from the diffusion path diagram through an image processing algorithm. The diffusion path diagram is generated by the crack network permeability analysis. The image processing algorithm is the skeletonization and length statistics method based on the OpenCV library. The value of the unit area ranges from 1 square millimeter to 5 square millimeters. For example, when the unit area is set to 3 square millimeters, the diffusion path density is the total path length divided by 3 square millimeters.

[0073] The overlapping region of the stress gradient distribution and the diffusion path density distribution is marked as the candidate area of the high stress sensitive region. The determination condition for the overlapping region is that the absolute value of the stress gradient component is greater than or equal to the preset gradient threshold and the diffusion path density is greater than or equal to the preset density threshold. The preset gradient threshold is obtained through the yield stress gradient experiment of the sealing contact interface material. The specimen size of the yield stress gradient experiment is 50 mm in length, 10 mm in width, 10 mm in height, the span is 40 mm, and the loading rate is 1 mm per minute. The preset gradient threshold is determined according to the stress gradient value at the specimen fracture. For example, if the stress gradient value at the specimen fracture is 100 MPa / mm, the preset gradient threshold is set to 80 MPa / mm. The preset density threshold is determined through the diffusion path density simulation of the leakage medium at the standard leakage rate. The standard leakage rate is 1 cm³ per minute. The simulation uses the finite volume method to calculate the diffusion path length distribution. For example, if the median of the diffusion path density obtained by simulation at the standard leakage rate is 0.5 mm / mm², the preset density threshold is set to 0.4 mm / mm².

[0074] According to the geometric relationship between the stress gradient component and the diffusion path density in the candidate area, the boundary division rule for the high stress sensitive region and the low stress sensitive region is generated. The boundary division rule is that the region where the stress gradient component exceeds the preset gradient threshold and the diffusion path density exceeds the preset density threshold is defined as the high stress sensitive region, and the remaining regions are defined as the low stress sensitive regions. The threshold determination uses the pixel-by-pixel comparison method, and the pixel resolution is 0.1 mm. For example, when the normal stress gradient of a certain pixel point is 90 MPa / mm and the diffusion path density is 0.6 mm / mm², it is determined as the high stress sensitive region. Based on the boundary division rule, the coordinate sets of the high stress sensitive region and the low stress sensitive region are output. The generation of the coordinate sets uses the spatial clustering algorithm to aggregate the regions that meet the conditions. The spatial clustering algorithm is the density-based DBSCAN algorithm. The neighborhood radius of the DBSCAN algorithm is set to 2 mm, and the minimum number of neighborhood points is set to 5. During the clustering process, isolated noise points are excluded. The determination condition for the isolated noise points is that the number of pixels in the clustering region is less than 10 and the distance from the adjacent clustering region is greater than 5 mm. For example, if a certain region contains 8 pixels and is 6 mm away from the nearest neighbor clustering region, it is determined as a noise point and excluded. The storage format of the coordinate sets is the polygon vertex sequence in the local coordinate system of the sealing contact interface. The polygon vertex sequence is extracted from the clustering region boundary through the Alpha Shape algorithm. The radius parameter of the Alpha Shape algorithm is set to 1 mm. For example, the boundary point cloud of a certain clustering region generates a polygon with 20 vertices through the Alpha Shape algorithm.

[0075] Step S3 divides the high / low stress-sensitive regions according to the double thresholds of stress gradient and diffusion path density, and at the same time considers the spatial superposition effect of mechanical load and leakage medium diffusion. Compared with the single stress threshold division method in the prior art, this step accurately locates the high-risk leakage regions through the spatial clustering algorithm and geometric relationship determination, solves the problem of detection blind spots caused by misjudgment due to local stress concentration in traditional methods, and the multi-parameter collaborative regional division logic reduces the false detection rate.

[0076] The specific implementation of high-frequency signal acquisition for high stress-sensitive regions and low-frequency signal acquisition for low stress-sensitive regions, and obtaining the pressure signal and leakage medium flow signal of the seal contact interface is as follows: Based on the coordinate sets of high stress-sensitive regions and low stress-sensitive regions, the deployment positions of the pressure sensor array and the flow sensor are configured. The pressure sensor array consists of multiple piezoelectric pressure sensors. The deployment positions of the piezoelectric pressure sensors are determined according to the polygon vertex sequence in the coordinate set of the high stress-sensitive region. One piezoelectric pressure sensor is deployed at each polygon vertex, and the distance between adjacent vertices is 1 mm to 5 mm. The deployment interval of the piezoelectric pressure sensors in the low stress-sensitive region is twice that of the high stress-sensitive region. For example, when the vertex spacing in the high stress-sensitive region is 2 mm, the deployment interval in the low stress-sensitive region is 4 mm. The flow sensor is a vortex flowmeter, and the deployment position of the vortex flowmeter is determined according to the main branch direction of the diffusion path. The main branch direction is parallel to the boundary of the clustering region of the high stress-sensitive region, and the distance from the boundary line is 1 mm to 3 mm. For example, when the boundary of the clustering region is a 10-mm straight line, the vortex flowmeters are deployed at intervals of 1 mm along the parallel direction of this straight line.

[0077] The high-frequency signal acquisition frequency is allocated according to the risk level of the high stress-sensitive region. The risk level is quantified by the product of the stress gradient component and the diffusion path density. The stress gradient component is the vector sum of the normal stress gradient and the shear stress gradient, which are extracted from the stress gradient distribution of the seal contact interface, and the diffusion path density is extracted from the diffusion path density distribution. The product quantification formula is risk level = stress gradient component × diffusion path density. For example, if the stress gradient component is 120 MPa / mm and the diffusion path density is 0.6 mm / mm², then the risk level is 72. The high-frequency signal acquisition frequency is linearly mapped to 1000 to 5000 times per second according to the risk level. The linear mapping relationship is frequency = 1000 + (risk level / maximum risk level) × 4000. The maximum risk level is determined by statistical analysis of historical leakage accident data. The historical leakage accident data includes the leakage rate records of the seal under different combinations of stress gradient and diffusion path density. For example, when the historical maximum risk level is 100, the frequency corresponding to the risk level of 72 is 1000 + (72 / 100) × 4000 = 3880 times per second.

[0078] Allocate a low-frequency signal acquisition frequency to the low-stress sensitive area. The low-frequency signal acquisition frequency is fixed at 100 to 500 times per second, and the specific value is adjusted according to the area ratio between the low-stress sensitive area and the high-stress sensitive area. The area ratio is calculated by the ratio of the polygon area of the low-stress sensitive area coordinate set to the total area of the sealed contact interface. When the area ratio is less than 10%, 500 times per second is adopted; when the area ratio is greater than or equal to 10% and less than 30%, 300 times per second is adopted; when the area ratio is greater than or equal to 30%, 100 times per second is adopted. For example, if the area of the low-stress sensitive area is 15% of the total area, the acquisition frequency is 300 times per second. Synchronize the acquisition timestamps of the pressure sensor array and the flow sensor to obtain the pressure signal of the sealed contact interface and the flow signal of the leakage medium in real time. The timestamp synchronization adopts a method combining GPS timing module and internal clock calibration. The synchronization accuracy of the GPS timing module is 1 microsecond, and the internal clock calibration period is once per minute. The calibration method is to compare the internal clock with the GPS timestamp and correct the deviation. The timestamp deviation between the pressure signal and the flow signal is less than or equal to 1 millisecond. The pressure signal and the flow signal are transmitted to the signal processing terminal through the Modbus protocol. The data buffer capacity of the signal processing terminal is 10,000 data points per second, and the data buffer uses a circular queue structure to store real-time signals.

[0079] Map the pressure signal and the flow signal into high-dimensional point cloud data, calculate the persistent homology barcode of the high-dimensional point cloud data and extract the barcode entropy value. The specific implementation of allocating region weights according to the spatio-temporal distribution correlation of the barcode entropy value in the high-stress sensitive area and the low-stress sensitive area to generate a compensation priority sequence is as follows: Slice the pressure signal and the flow signal according to time windows. The length of each time window is 1 second to 5 seconds, and the sliding step of the time window is 10% to 50% of the window length. For example, when the time window length is 3 seconds, the sliding step is 0.3 seconds to 1.5 seconds. The signal sequence within the time window is mapped into high-dimensional point cloud data through the time-delay embedding method. The embedding dimension and delay parameter of the time-delay embedding method are adaptively adjusted according to the dynamic ratio of the high-frequency signal acquisition frequency in the high-stress sensitive area to the low-frequency signal acquisition frequency in the low-stress sensitive area. The dynamic ratio is the quotient of the high-frequency signal acquisition frequency divided by the low-frequency signal acquisition frequency. The embedding dimension is the integer part of the dynamic ratio, and the delay parameter is the decimal part of the dynamic ratio rounded to two decimal places. For example, if the high-frequency signal acquisition frequency is 5000 times per second and the low-frequency signal acquisition frequency is 500 times per second, the dynamic ratio is 10.0. At this time, the embedding dimension is set to 10 and the delay parameter is 0.00. The mapping dimension of the time-delay embedding method is 10 dimensions, and the delay time is 0 milliseconds.

[0080] Perform persistent homology analysis on high-dimensional point cloud data to generate persistent homology barcodes. The generation of persistent homology barcodes adopts the Vietoris-Rips complex construction method. The range of filtering parameters for the Vietoris-Rips complex construction method is dynamically set according to the statistical distribution of the pressure signal amplitude. The statistical distribution of the pressure signal amplitude includes the mean and variance. The mean is the average value of the pressure signal within each time window, and the variance is the square of the standard deviation of the pressure signal amplitude. The standard deviation is calculated using an unbiased estimation formula. The lower limit of the filtering parameter is the mean minus three times the variance, and the upper limit is the mean plus three times the variance. For example, if the mean of the pressure signal is 10 MPa and the variance is 4 (the standard deviation is 2 MPa), then the lower limit of the filtering parameter is 10 - 3×2 = 4 MPa, and the upper limit is 10 + 3×2 = 16 MPa. The range of the filtering parameter is from 4 MPa to 16 MPa. The filtering step size is adjusted according to the dynamic range of the pressure signal amplitude. The dynamic range is the difference between the upper limit and the lower limit. For example, when the dynamic range is 12 MPa, the filtering step size is set to 0.3 MPa.

[0081] Calculate the barcode entropy value based on the length distribution of the persistent homology barcodes. The barcode entropy value is the Shannon entropy. The calculation formula of the Shannon entropy is the negative logarithm weighted sum of the probability distribution of the barcode lengths. The probability distribution is the proportion of each barcode length in the total length. When calculating, a regional sensitivity weight coefficient is introduced. The regional sensitivity weight coefficient is the ratio of the average lifetimes of the barcodes in the high-stress sensitive region and the low-stress sensitive region. The average barcode lifetime is the arithmetic mean of all barcode lengths in the same region. For example, if the average barcode lifetime in the high-stress sensitive region is 10 seconds and that in the low-stress sensitive region is 5 seconds, then the regional sensitivity weight coefficient is 2.0. The determination threshold for short-lived barcodes is dynamically adjusted according to the regional sensitivity weight coefficient. The determination threshold for short-lived barcodes is the reciprocal of the regional sensitivity weight coefficient multiplied by the reference threshold. The reference threshold is 1 second. For example, if the regional sensitivity weight coefficient is 2.0, then the short-lived determination threshold is 0.5 seconds. Barcodes with a length less than 0.5 seconds are ignored in the entropy value calculation, and the proportion of the ignored barcodes in the total number of barcodes does not exceed 10%.

[0082] Allocate regional weights according to the spatio-temporal distribution correlation of barcode entropy values in high-stress sensitive areas and low-stress sensitive areas. The spatio-temporal distribution correlation is calculated through spatial grid division and time-window sliding correlation. The division of spatial grids is aligned with the clustering boundaries of high-stress sensitive areas. The spatial grid size ranges from 1 mm to 5 mm, which is consistent with the polygon vertex spacing of the coordinate set of high-stress sensitive areas. For example, when the polygon vertex spacing is 2 mm, the spatial grid size is set to 2 mm. The length of the time window is inversely proportional to the signal acquisition frequency. The signal acquisition frequency ranges from 1000 to 5000 times per second, and the time window length ranges from 1 second to 5 seconds. For example, when the signal acquisition frequency is 5000 times per second, the time window length is 1 second; when the signal acquisition frequency is 1000 times per second, the time window length is 5 seconds. The spatio-temporal sliding correlation is calculated using the Pearson correlation coefficient. Areas with a correlation coefficient greater than or equal to 0.7 are determined to have a high correlation. The regional weight is the square of the correlation coefficient multiplied by the regional sensitivity weight coefficient. For example, if the correlation coefficient is 0.8 and the regional sensitivity weight coefficient is 2.0, then the regional weight is 0.8 2 ×2.0 = 1.28.

[0083] Generate a compensation priority sequence in descending order of regional weights. The generation rule of the compensation priority sequence is to prioritize the high-frequency signal acquisition areas for areas with the same weight. The determination condition for high-frequency signal acquisition areas is that the signal acquisition frequency is greater than or equal to 3000 times per second. The storage format of the compensation priority sequence is a list of regional numbers in the local coordinate system of the sealed contact interface. The regional numbers correspond one-to-one with the coordinate set indexes of high-stress sensitive areas and low-stress sensitive areas. For example, the compensation priority sequence is [Area A1, Area B3, Area C2], indicating that Area A1 has the highest compensation priority. The list of regional numbers is stored in JSON format, and each entry contains the regional number, weight value, and coordinate range.

[0084] In step S5, calculate the barcode entropy value by dynamically adjusting the time-delay embedding parameter and the regional sensitivity weight, and combine the high-frequency / low-frequency signal characteristics and regional risk differences to solve the limitation that traditional fixed topology parameters cannot adapt to dynamic leakage signals. Compared with the existing technology, this step introduces the regional sensitivity weight coefficient (the ratio of barcode lifetimes in high / low-stress areas) and a filtering rule driven by statistical distribution. By dynamically suppressing noise interference and strengthening the characteristics of high-risk areas, the sensitivity of leakage signal recognition is improved. The dynamic adaptation mechanism of signal characteristics and regional weights directly targets the complexity of local leakage paths caused by installation stress, ensuring strong consistency between the detection benchmark correction accuracy and the spatial distribution of leakage risks.

[0085] The specific implementation method of dynamically compensating high-weight regions based on the compensation priority sequence and outputting the corrected seal performance detection results is as follows: Extract high-weight regions with a higher ranking from the compensation priority sequence. The number of high-weight regions is determined according to a preset proportion range of the total number of regions in the compensation priority sequence. The preset proportion range is 10% to 30% of the total number of regions. For example, when the compensation priority sequence contains 100 regions, the first 10 to 30 regions are extracted as high-weight regions. The extraction rule is to continuously select from the head of the sequence. If there are regions with the same weight in the sequence, the region corresponding to the high-frequency signal acquisition frequency is preferentially selected. The determination condition for the high-frequency signal acquisition frequency is that the signal acquisition frequency is greater than or equal to 3000 times per second.

[0086] Dynamically compensate the pressure signals of high-weight regions. The dynamic compensation method is to adjust the compensation coefficient based on the pressure signal deviation value. The pressure signal deviation value is the difference between the current pressure signal and the reference pressure signal. The reference pressure signal is the moving average of the historical pressure signal mean. The window length of the moving average is adaptively adjusted according to the time span of historical data. The time span is the total duration of historical data, and the window length is 10% to 20% of the time span. For example, when the historical data duration is 100 seconds, the window length is 10 seconds to 20 seconds. The calculation of the moving average is to take the arithmetic mean of the historical pressure signals within the window. The current pressure signal deviation value is the difference between the current pressure signal value and the moving average. The compensation coefficient is 1 plus the ratio of the deviation value to the moving average. For example, when the moving average is 10 MPa, the current pressure signal is 12 MPa, the deviation value is 2 MPa, and the compensation coefficient is 1 + 2 / 10 = 1.2.

[0087] Dynamically compensate the flow signals of high-weight regions. The dynamic compensation method is to adjust the compensation coefficient based on the flow signal deviation value. The flow signal deviation value is the difference between the current flow signal and the reference flow signal. The reference flow signal is the exponential smoothing value of the historical flow signal mean. The smoothing coefficient of the exponential smoothing value is dynamically set according to the flow signal change rate. The flow signal change rate is the absolute change amount of the flow signal per unit time. The smoothing coefficient is the reciprocal of the change rate multiplied by the reference coefficient. The reference coefficient is 0.1 to 0.3. For example, when the flow signal change rate is 0.5 liters per second, the smoothing coefficient is 0.2. The calculation formula for the exponential smoothing value is the current flow signal value multiplied by the smoothing coefficient plus the exponential smoothing value at the previous moment multiplied by (1 - the smoothing coefficient). The current flow signal deviation value is the difference between the current flow signal value and the exponential smoothing value. The compensation coefficient is 1 minus the ratio of the deviation value to the exponential smoothing value. For example, when the exponential smoothing value is 5 liters per minute, the current flow signal is 6 liters per minute, the deviation value is 1 liter per minute, and the compensation coefficient is 1 - 1 / 5 = 0.8.

[0088] The compensated pressure signal and flow signal are weighted and fused with the original signals in the uncompensated area according to the product of the area weight and the compensation coefficient. The area weight is the weight value defined in the compensation priority sequence, and the compensation coefficient is the dynamic compensation coefficient of the pressure or flow signal. The weighted fusion formula is the compensated signal value multiplied by the product of the area weight and the compensation coefficient, plus the original signal value in the uncompensated area multiplied by (1 - the product of the area weight and the compensation coefficient). For example, if the area weight is 1.2, the compensation coefficient is 0.8, and the product is 0.96, the compensated pressure signal is 12 MPa, and the original pressure signal in the uncompensated area is 10 MPa, then the fused pressure signal value is 12×0.96 + 10×(1 - 0.96) = 11.52 + 0.4 = 11.92 MPa.

[0089] Output the corrected seal performance detection results after fusion. The corrected results include the pressure distribution contour map and the leakage flow trend curve. The pressure distribution contour map is generated by mapping the fused pressure signal to the local coordinate system of the seal contact interface through an interpolation algorithm, and the interpolation algorithm is the bilinear interpolation method. The leakage flow trend curve is smoothed by the time series analysis method according to the fused flow signal in a time window. The time series analysis method is the moving average method, and the moving average window length is from 1 second to 5 seconds, which is the same as the time window length in step S5. The corrected results are stored in the form of two-dimensional images and curve graphs. The image resolution is from 10 pixels to 20 pixels per millimeter. The horizontal axis of the curve graph is the time axis, and the vertical axis is the flow value. The scale of the time axis is aligned with the signal acquisition timestamp in step S4.

[0090] The calculations involved in the embodiments are all dimensionless numerical calculations. The preset parameters and threshold selections in the calculations are set by those skilled in the art according to the actual situation.

[0091] It should be noted that the present invention can be deployed on the device itself to achieve embedded applications, or can also run on a PC or other terminals with a user interface, so as to meet various hardware environments and usage requirements.

[0092] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more collections of available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0093] Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices, and modules described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0094] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the modules is only a logical function division, and there can be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the couplings or direct couplings or communication connections shown or discussed with each other can be through some interfaces. The indirect couplings or communication connections of the devices or modules can be in electrical, mechanical, or other forms.

[0095] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules. They can be located in one place or distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0096] In addition, in each embodiment of the present application, each functional module can be integrated into a processing module, can exist physically alone for each module, or two or more modules can be integrated into one module.

[0097] If the above-mentioned function is implemented in the form of a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0098] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present application, and all of them should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0099] Finally: The above is only the preferred embodiment of the present invention and is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An on-line detection method for the sealing performance of a seal, characterized in that It includes the following steps: S1. During the installation of the seal, obtain the real-time installation stress data between the seal and the installation structure; S2. Based on the real-time installation stress data, construct a phase field model of the seal contact interface, and determine the diffusion path of the leakage medium along the seal contact interface through crack network permeability analysis; S3. According to the phase field model and the diffusion path, divide the seal contact interface into a high stress sensitive area and a low stress sensitive area; S4. Implement high-frequency signal acquisition for the high stress sensitive area and low-frequency signal acquisition for the low stress sensitive area to obtain the pressure signal of the seal contact interface and the flow signal of the leakage medium; S5. Map the pressure signal and the flow signal into high-dimensional point cloud data, calculate the persistent homology barcode of the high-dimensional point cloud data and extract the barcode entropy value, and allocate regional weights according to the spatio-temporal distribution correlation of the barcode entropy value in the high stress sensitive area and the low stress sensitive area to generate a compensation priority sequence; S6. Based on the compensation priority sequence, perform dynamic compensation on the high-weight area and output the corrected seal performance detection result.

2. The online detection method for the sealing performance of a seal according to claim 1, characterized in that During the installation of the seal, obtaining the real-time installation stress data between the seal and the installation structure includes: Arrange a plurality of strain gauges on the contact surface between the seal and the installation structure to form a stress sensing network; Real-time collect the dynamic strain data during the installation of the seal through a distributed fiber optic sensor; Based on a preset stress-strain relationship curve, perform dynamic calibration on the dynamic strain data to generate real-time installation stress data.

3. An on-line detection method for the sealing performance of a seal according to claim 1, characterized in that, Based on the real-time installation stress data, constructing a phase field model of the seal contact interface and determining the diffusion path of the leakage medium along the seal contact interface includes: Input the three-dimensional stress field components of the real-time installation stress data as the stress driving term of the phase field model. The control equations of the phase field model include a coupled stress balance equation and a phase field order parameter evolution equation. The crack surface energy density function in the phase field order parameter evolution equation is defined based on the fracture toughness parameters of the seal contact interface material; Use an explicit time integration algorithm to iteratively solve the phase field model to generate the crack propagation path and the crack network topology of the seal contact interface. The crack network topology includes geometric characteristic parameters such as crack branch angle and curvature radius; Based on the geometric characteristic parameters of the crack branch angle and the curvature radius, calculate the equivalent permeability tensor of the crack network. The calculation of the equivalent permeability tensor includes weighted statistical averaging of the crack branch angle and the curvature radius; According to the matching degree between the maximum principal value direction of the equivalent permeability tensor and the main branch direction in the crack network topology, determine the diffusion path of the leakage medium along the seal contact interface.

4. An on-line detection method for the sealing performance of a seal according to claim 1, characterized in that, According to the phase field model and the diffusion path, dividing the seal contact interface into a high stress sensitive area and a low stress sensitive area includes: Calculate the stress gradient distribution of the seal contact interface based on the phase field model. The stress gradient distribution includes normal stress gradient and shear stress gradient components; Extract the spatial density distribution of the diffusion path, and mark the overlapping area of the stress gradient distribution and the diffusion path density distribution as the high stress sensitive area candidate area; Generate the boundary division rules for high-stress sensitive areas and low-stress sensitive areas according to the geometric relationship between the in-stress gradient components and the diffusion path density in the candidate area; Based on the boundary division rules, output the coordinate sets of high-stress sensitive areas and low-stress sensitive areas. The generation of the coordinate sets uses a spatial clustering algorithm to aggregate the areas that meet the conditions.

5. The on-line detection method for the sealing performance of a seal according to claim 4, characterized in that The boundary division rule defines the area where the stress gradient component exceeds the preset gradient threshold and the diffusion path density exceeds the preset density threshold as a high-stress sensitive area.

6. The on-line detection method for the sealing performance of a seal according to claim 1, characterized in that Implement high-frequency signal acquisition for high-stress sensitive areas and low-frequency signal acquisition for low-stress sensitive areas to obtain the pressure signal of the sealed contact interface and the flow signal of the leakage medium, including: Based on the coordinate sets of high-stress sensitive areas and low-stress sensitive areas, configure the deployment positions of the pressure sensor array and the flow sensor; Allocate the high-frequency signal acquisition frequency according to the risk level of the high-stress sensitive area, and the risk level is quantified by the product of the stress gradient component and the diffusion path density; Allocate the low-frequency signal acquisition frequency to the low-stress sensitive area; Synchronize the acquisition timestamps of the pressure sensor array and the flow sensor to obtain the pressure signal of the sealed contact interface and the flow signal of the leakage medium in real time.

7. The on-line detection method for the sealing performance of a seal according to claim 1, characterized in that Map the pressure signal and the flow signal to high-dimensional point cloud data, calculate the persistent homology barcode of the high-dimensional point cloud data and extract the barcode entropy value, and allocate regional weights according to the spatio-temporal distribution correlation of the barcode entropy value in high-stress sensitive areas and low-stress sensitive areas to generate a compensation priority sequence, including: Slice the pressure signal and the flow signal by time window, and map the signal sequence in each time window to high-dimensional point cloud data through the time delay embedding method; Perform persistent homology analysis on the high-dimensional point cloud data to generate a persistent homology barcode. The generation of the persistent homology barcode uses the Vietoris-Rips complex construction method; Calculate the barcode entropy value based on the length distribution of the persistent homology barcode. The barcode entropy value is the Shannon entropy, and a regional sensitivity weight coefficient is introduced during the calculation. The regional sensitivity weight coefficient is the ratio of the average barcode lifetimes of high-stress sensitive areas and low-stress sensitive areas; Allocate regional weights according to the spatio-temporal distribution correlation of the barcode entropy value in high-stress sensitive areas and low-stress sensitive areas; Generate a compensation priority sequence in descending order of regional weights. The generation rule of the compensation priority sequence is that areas with the same weight are sorted by high-frequency signal acquisition areas first.

8. The on-line detection method for the sealing performance of a seal according to claim 7, characterized in that, The embedding dimension and delay parameter of the time delay embedding method are adaptively adjusted according to the dynamic ratio of the high-frequency signal acquisition frequency of high-stress sensitive areas to the low-frequency signal acquisition frequency of low-stress sensitive areas.

9. An on-line detection method for the sealing performance of a seal according to claim 7, characterized in that, The spatio-temporal distribution correlation is calculated through spatial grid division and time window sliding correlation. The division of the spatial grid is aligned with the clustering boundary of the high-stress sensitive area, and the length of the time window is inversely proportional to the signal acquisition frequency.

10. The on-line detection method for the sealing performance of a seal according to claim 1, characterized in that Perform dynamic compensation on high-weight areas based on the compensation priority sequence and output the corrected sealed performance detection results, including: Extract the high-weight areas with higher rankings from the compensation priority sequence. The number of high-weight areas is determined according to a preset ratio range based on the total number of areas in the compensation priority sequence; Dynamically compensate the pressure signal in the high-weight area, and the dynamic compensation method is to adjust the compensation coefficient based on the pressure signal deviation value; Dynamically compensate the flow signal in the high-weight area, and the dynamic compensation method is to adjust the compensation coefficient based on the flow signal deviation value; Weightedly fuse the compensated pressure signal and flow signal with the original signals in the uncompensated area according to the product of the area weight and the compensation coefficient; Output the corrected seal performance detection results after weighted fusion, including the pressure distribution contour map and the leakage flow trend curve.

Images