Large flange top flatness detection method and system based on multi-point sensing

Abstract

The invention provides a large flange top flatness detection method and system based on multi-point sensing, and relates to the technical field of flange detection.The method comprises the steps that the lowest point and the highest point of a flange are determined, and the lowest point serves as a height reference; recording radius and angle information of each measurement point; acquiring actual height information of each measuring point, and calculating the relative height of each measuring point relative to the lowest point; processing the acquired actual height data, and fitting a theoretical reference plane of a flange by combining radius and angle information of a measurement point; and comparing the relative height with the height of the theoretical reference plane to obtain and compare the flatness deviation of the flange surface, and automatically outputting an abnormal position and flatness correction suggestion. The technical problems that in the prior art, large flange top flatness detection precision is low, measurement efficiency is not high, and abnormal areas are difficult to accurately recognize are solved. The technical effects of improving the flange flatness detection precision and efficiency and accurately positioning the abnormal area are achieved.

Description

Technical Field

[0001] This invention relates to the field of flange inspection technology, and specifically to a method and system for detecting the flatness of the top of large flanges based on multi-point sensing. Background Technology

[0002] Large flanges, as crucial components connecting pipelines and pressure vessels, directly impact the safety and reliability of equipment operation through their sealing performance. The flatness of the flange sealing surface is a key indicator for ensuring sealing performance; deviations in flatness can lead to media leakage, stress concentration, and reduced equipment lifespan. Current methods for inspecting the flatness of large flanges largely rely on manual measurement or single-point measuring tools such as height gauges, levels, or laser levels. These methods have significant shortcomings in terms of measurement accuracy, efficiency, and repeatability. Single-point measurements cannot comprehensively reflect the overall morphology of the flange surface, easily overlooking local unevenness or deformation areas, especially around bolt holes, sealing lines, and load transfer paths, making it difficult to accurately identify potential anomalies. Furthermore, flanges may tilt or deviate in posture during installation, and existing methods typically only focus on the flange's height change relative to the horizontal plane, failing to distinguish between flange flatness and levelness, resulting in flange top flatness test results deviating from the actual flatness condition.

[0003] Existing technologies suffer from low accuracy in detecting the flatness of the top of large flanges, low measurement efficiency, and difficulty in accurately identifying abnormal areas. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for detecting the flatness of the top of large flanges based on multi-point sensing, in order to solve the technical problems of low accuracy, low measurement efficiency and difficulty in accurately identifying abnormal areas in the existing technology for detecting the flatness of the top of large flanges.

[0005] In view of the above problems, this application provides a method and system for detecting the flatness of the top of a large flange based on multi-point sensing.

[0006] The first aspect of this application provides a method for detecting the top flatness of a large flange based on multi-point sensing. This method includes: determining the lowest and highest points of the flange using a field level, calculating the overall height difference of the flange surface, and using the height measurement data of the lowest point as a height reference benchmark; arranging measurement points on the flange circumference at preset equal angles, and recording the radius and angle information of each measurement point; collecting the actual height information of each measurement point using a multi-point sensor, and calculating the relative height of each measurement point relative to the lowest point based on the height reference benchmark; performing noise filtering and attitude correction on the collected actual height data, and fitting a theoretical reference plane of the flange by combining the radius and angle information of the measurement points; comparing the relative height of each measurement point with the height of the theoretical reference plane, and obtaining the flatness deviation of the flange surface through comparison; if the flatness deviation exceeds a preset threshold, automatically outputting abnormal locations and flatness correction suggestions.

[0007] Optionally, a basic layout density template can be retrieved from a standard library based on the flange pressure rating and sealing surface type; the density of measurement points can be increased on the load transfer path on both sides of the bolt holes based on historical data analysis; the measurement points can be set according to the distribution angle calculated by using the basic layout density template as a reference point and the pre-increased measurement point density, and sensors can be deployed accordingly.

[0008] Optionally, the system acquires a historical equipment database, extracts leak points and maintenance records, identifies abnormal point distributions and analyzes abnormal probabilities in the records, and establishes abnormal distribution characteristics. Based on the abnormal distribution characteristics, it performs a sensor measurement requirement analysis to determine the need to increase the density of measurement points.

[0009] Optionally, the collected actual height data is subjected to multi-layer noise filtering to obtain filtered height data; based on the filtered height data, full-parameter attitude correction is performed to obtain attitude angle correction data; the main reference point is selected from the attitude angle correction data, and the main reference point is used as the forced passage point. Combining the three-dimensional distribution relationship of radius-angle-height of the measurement points, the optimal theoretical reference plane is fitted using the constrained least squares method, wherein the measurement points in the sealing line area are given a higher fitting weight than other measurement points.

[0010] Optionally, the measured height data is initially filtered based on the median height of a time-series sliding window to eliminate impulse noise. Each measurement point is sampled N times consecutively, and the height value sequence is filtered using a time-series sliding window, where N is not less than 5. Then, outlier measured height data is removed based on spatial neighborhood consistency to obtain intermediate-level filtered data. This involves finding K nearest neighbor points within a circular region of radius r centered on the current measurement point, calculating the neighborhood standard deviation, identifying outliers, where r is 2-5 times the distance between measurement points, and K is not less than 4. Finally, considering the elastic modulus constraint of the flange material, the height gradient and maximum curvature between adjacent measurement points are calculated to correct unreasonable measurement values, resulting in the final filtered height data.

[0011] Optionally, a transformation matrix from the sensor coordinate system to the flange coordinate system is established, and a correction factor for the coordinate transformation is calculated in real time based on the acquired attitude angles. Using the correction factor, a six-degree-of-freedom spatial transformation is performed on the height of each measurement point, and the three-dimensional coordinates in the flange coordinate system are recalculated to obtain the attitude angle correction data.

[0012] Optionally, based on the height reference benchmark, points with attitude angle correction data within the P% range are selected from all measurement points to form an initial candidate set, wherein P% is no greater than 15%; spatial density analysis is performed on the initial candidate set to remove isolated abnormal low points, obtaining a secondary candidate set; for each candidate point in the secondary candidate set, a comprehensive evaluation value is calculated in the height dimension, position dimension, and stability dimension; based on the comprehensive evaluation value, the candidate point with the highest evaluation value is selected as the final master benchmark point, wherein the master benchmark point satisfies the following conditions: its height and the height reference benchmark satisfy a distance range, its position is close to the theoretical sealing line and theoretical geometric center of the flange, and its height value has the smallest coefficient of variation in multiple measurements.

[0013] Optionally, a predetermined proportion of measurement points are randomly selected as a validation set, wherein the measurement points in the validation set are those that did not participate in the fitting of the reference plane; based on the theoretical reference plane, the average deviation and maximum deviation of the points in the validation set are calculated; when the deviation exceeds a predetermined limit, an iterative reweighted least squares method is used to dynamically adjust the weights according to the residuals of the previous fitting round, and the fitting is repeated until convergence; during the iterative optimization process, the reference plane obtained in each round and its corresponding validation set evaluation score are retained, and after the iteration is completed, the plane with the highest validation set evaluation score is selected as the final theoretical reference plane.

[0014] Optionally, each test data is bound to a unique flange identifier to establish a three-dimensional digital archive; time-series analysis is performed on the test data of the same flange to fit the degradation curve of key indicators; using the degradation curve of the key indicators in combination with the three-dimensional digital archive, predictive maintenance information is generated before the planned shutdown is predicted.

[0015] A second aspect of this application provides a multi-point sensing-based system for detecting the flatness of the top of a large flange. This system includes: a height difference calculation module, used to determine the lowest and highest points of the flange using a field level, calculate the overall height difference of the flange surface, and use the height measurement data of the lowest point as a height reference benchmark; a data recording module, used to arrange measurement points on the flange circumference at preset equal angles and record the radius and angle information of each measurement point; a relative height calculation module, used to collect the actual height information of each measurement point using multi-point sensors, and calculate the relative height of each measurement point relative to the lowest point based on the height reference benchmark; a theoretical reference plane fitting module, used to perform noise filtering and attitude correction on the collected actual height data, and fit the theoretical reference plane of the flange by combining the radius and angle information of the measurement points; and a height comparison module, used to compare the relative height of each measurement point with the height of the theoretical reference plane, obtain the flatness deviation of the flange surface through comparison, and automatically output abnormal locations and flatness correction suggestions if the flatness deviation exceeds a preset threshold.

[0016] One or more technical solutions provided in this application have at least the following technical effects or advantages: The method provided in this application embodiment determines the lowest and highest points of a flange using a field level, calculates the overall height difference of the flange surface, and uses the height measurement data of the lowest point as a height reference benchmark. Measurement points are arranged on the flange circumference at preset equal angles, and the radius and angle information of each measurement point are recorded. Multi-point sensors are used to collect the actual height information of each measurement point, and the relative height of each measurement point relative to the lowest point is calculated based on the height reference benchmark. Noise filtering and attitude correction are performed on the collected actual height data, and the theoretical reference plane of the flange is fitted by combining the radius and angle information of the measurement points. The relative height of each measurement point is compared with the height of the theoretical reference plane, and the flatness deviation of the flange surface is obtained through comparison. If the flatness deviation exceeds a preset threshold, the abnormal location and flatness correction suggestions are automatically output. This achieves the technical effect of improving the accuracy and efficiency of flatness detection on the top of large flanges while accurately locating abnormal areas.

[0017] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the method for detecting the top flatness of a large flange based on multi-point sensing provided in this application.

[0020] Figure 2 This is a schematic diagram of the structure of the large flange top flatness detection system based on multi-point sensing provided in this application.

[0021] Explanation of reference numerals in the attached diagram: Height difference calculation module 11, data recording module 12, relative height calculation module 13, theoretical reference plane fitting module 14, height comparison module 15. Detailed Implementation

[0022] This application provides a method and system for detecting the flatness of the top of large flanges based on multi-point sensing, which addresses the technical problems of low accuracy, low measurement efficiency, and difficulty in accurately identifying abnormal areas in existing technologies. It achieves the technical effect of improving the accuracy and efficiency of large flange top flatness detection while accurately locating abnormal areas.

[0023] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0024] Example 1, as Figure 1 As shown, this application provides a method for detecting the flatness of the top of a large flange based on multi-point sensing. The method for detecting the flatness of the top of a large flange based on multi-point sensing includes: The lowest and highest points of the flange are determined using an on-site level, and the overall height difference of the flange surface is calculated. The height measurement data of the lowest point is used as the height reference benchmark.

[0025] Specifically, a field level is used to measure the height of multiple points on the top area of ​​the flange. This field level is an instrument capable of quickly and accurately measuring the height differences of an object's surface on-site, possessing high-precision leveling and data recording functions. The level scans the flange surface using its on-site measurement function, automatically identifying and recording the positions and heights of the lowest and highest points within the measurement range. By obtaining the lowest and highest points, the overall height difference of the flange surface is calculated using the formula: Overall Height Difference = Highest Point Height - Lowest Point Height. This overall height difference reflects the maximum vertical undulation of the flange surface. The height measurement data of the lowest point is used as a height reference benchmark, a fixed standard set for subsequent calculations of the relative heights of each measurement point.

[0026] By acquiring multi-point height measurement data using an on-site level, determining the lowest and highest points, and calculating the overall height difference, the degree of undulation on the flange surface can be intuitively understood, and its flatness condition can be preliminarily judged. Using the lowest point as a height reference benchmark, this lays the foundation for subsequent operations such as calculating the relative height of each measurement point, fitting the theoretical reference plane, and calculating flatness deviation. This ensures that the entire inspection process can be carried out under a unified reference standard, guaranteeing the accuracy and consistency of the inspection results.

[0027] Measurement points are set up on the flange circumference at preset equal angles, and the radius and angle information of each measurement point are recorded.

[0028] Furthermore, measuring points are arranged on the flange circumference at preset equal intervals, including: calling a basic layout density template from the standard library according to the flange pressure rating and sealing surface type; increasing the density of measuring points on the load transmission path on both sides of the bolt holes based on historical data analysis; using the basic layout density template as a reference point, calculating the distribution angle according to the equal intervals and the pre-increased measuring point density, setting the measuring points, and correspondingly arranging sensors.

[0029] Specifically, the pressure rating and sealing surface type of the flange are obtained by consulting the flange's design drawings and product manuals. The pressure rating reflects the pressure the flange withstands during operation, while the sealing surface type determines the flange's sealing structure, such as a flat seal or a raised face seal. Based on the flange's pressure rating and sealing surface type, the corresponding basic measurement point density template is retrieved from a pre-established standard library. This standard library is pre-established based on common flange design specifications, such as ASME B 16.5 and DIN 2632, and practical experience data. It contains a database of various flange design parameters, operating conditions, and common measurement point layout templates. Its purpose is to provide standardized references for flange measurement point layout, ensuring the scientific and uniform nature of the layout. The basic measurement point density template defines the basic layout density and distribution method of the flange measurement points.

[0030] The bolt holes are critical load transfer paths. Based on historical data, which includes inspection information of similar flanges under the same or similar operating conditions (such as maintenance records, leak points, and the condition of the sealing surface), areas prone to stress concentration or sealing problems on both sides of the bolt holes are identified, and the density of measurement points is increased in these areas. Using the obtained basic layout density template as a reference point, and considering the increased measurement point density, the distribution angle is calculated according to the principle of equal angle division. For example, the basic layout interval angle is first determined, and then the distribution angle of measurement points on the entire flange circumference is replanned based on the increased measurement points on both sides of the bolt holes. Measurement points are accurately set on the flange circumference according to the calculated distribution angle, and a sensor is placed at each measurement point. The sensor uses laser, ultrasonic, or capacitive technologies, and height data is obtained by measuring the distance between the sensor and the flange surface. After each measurement point is set up, a professional measuring tool, such as a laser rangefinder, is used to accurately measure the radius of the measurement point from the flange center, and the angle of the measurement point relative to a fixed reference direction of the flange, such as the line connecting the centers of a bolt hole, is recorded.

[0031] For example, for a flange with a pressure rating of PN16 and a flat seal type, the basic measurement point density template retrieved from the standard library might specify a measurement point every 30° on the flange circumference. Analysis of historical data reveals that a 15° interval on either side of the bolt holes is a problem-prone area, so measurement points are added in this area. That is, while the original basic measurement point interval is 30°, after adding measurement points within the 15° interval on either side of the bolt holes, the 30° interval is maintained in areas outside the bolt holes, while measurement points are set at 15° intervals within the 15° interval on either side of the bolt holes. After the sensor is installed at each measurement point, the measurement is recorded, revealing that the radius of measurement point A from the flange center is 500 mm, and the angle relative to the reference direction is 45°. In this way, measurement points are rationally and accurately distributed on the flange circumference, and key location information for each point is obtained.

[0032] By arranging measurement points at preset equal angles on the flange circumference and combining historical data with the guidance of standard library templates, the scientific and rational distribution of measurement points is ensured. This ensures comprehensive coverage of all areas on the flange circumference and records the radius and angle information of each measurement point, providing accurate positioning basis for subsequent data processing and analysis. This enables accurate correlation between the data of each measurement point and the specific location on the flange, thereby more accurately assessing the overall condition of the flange.

[0033] Furthermore, the density of measurement points is increased based on historical data analysis, including: acquiring the equipment historical data database, extracting leakage points and maintenance record information; identifying abnormal point distributions and analyzing abnormal probabilities in the record information to establish abnormal distribution characteristics; and conducting sensor measurement demand analysis based on the abnormal distribution characteristics to determine the need to increase the density of measurement points.

[0034] Specifically, the equipment management system acquires a complete historical database related to the target flange, containing various data about the flange since it was put into use, such as long-term operating data, maintenance history, leak points, and fault records. Leak point and maintenance record information are extracted from the equipment historical database. A leak point refers to the area where gas or liquid leaks at the flange connection surface, usually related to factors such as the flange's sealing performance and flatness. The leak point information records key information such as the specific location of the leak during past use, the time of the leak, and the severity of the leak. The maintenance record information includes various maintenance operations performed on the flange, such as maintenance time, maintenance method, and replaced parts.

[0035] Anomaly distribution identification is performed on the extracted leak points and maintenance records. By categorizing the specific locations of leak points and maintenance records on the flange surface, areas on the flange surface that frequently experience leaks or maintenance events are identified. Spatial data analysis methods, such as heat maps, are used to visualize the locations of leak points and maintenance records on the flange surface, visually displaying the fault density in each area. Furthermore, density analysis and outlier detection algorithms, such as Gaussian distribution-based anomaly detection and local outlier analysis, are used to identify areas where leaks or maintenance frequently occur as high-risk areas. For example, analysis revealed a higher probability of faults in the load transfer paths on both sides of the flange bolt holes.

[0036] Based on the identified high-risk areas, the probability of anomalies occurring in each area is calculated. The anomaly probability is obtained by statistically analyzing the ratio of the number of anomalies occurring in each area to the total operating time. Anomaly distribution characteristics are then formed based on the distribution of identified anomaly points and their probabilities. Sensing measurement requirements are analyzed based on these characteristics, and optimization is performed using both spatial uniform distribution and density adaptive distribution methods. Increasing the density of measurement points is determined to ensure more sensors are deployed in critical areas. By combining historical data with adjustments to the distribution of measurement points, high-risk areas are more densely covered, reducing omissions and improving the accuracy of flange monitoring.

[0037] The actual height information of each measurement point is collected using multi-point sensors, and the relative height of each measurement point relative to the lowest point is calculated based on the height reference benchmark.

[0038] Specifically, a multi-point sensor refers to a combination of devices capable of simultaneously or sequentially measuring the height of multiple preset measurement points on the circumference of a flange, characterized by high precision and rapid response. Based on the pre-defined measurement point positions arranged at preset equal angles on the flange circumference, the multi-point sensor is accurately installed at the corresponding positions to ensure stable and accurate measurement of the height of each point. Data is collected through the multi-point sensor to obtain the actual height information of each measurement point. For example, a laser displacement sensor emits a laser beam to the surface of the measurement point; the reflected laser beam is received by the sensor. Based on the time difference and speed of light propagation, the distance between the sensor and the surface of the measurement point is calculated, thus obtaining the actual height of the measurement point. During the data acquisition process, it is crucial to ensure the stable operation of the sensor and avoid external interference affecting measurement accuracy. Multiple measurements can be performed and averaged to improve data accuracy. For example, five measurements can be taken for each measurement point, and the average of these five measurements can be taken as the actual height of that measurement point. Based on a height reference benchmark, the relative height of each measurement point relative to the lowest point is calculated: Relative height = Actual height - Height reference benchmark.

[0039] For example, for a large flange with a diameter of 2 meters, eight measuring points are arranged on its top according to a layout plan. Laser displacement sensors are used as multi-point sensors, with each sensor aligned with one measuring point. The height of that point relative to the sensor's initial position is accurately measured. The height data of these eight points are 102.5, 103.2, 101.8, 104.1, 102.9, 103.6, 101.5, and 104.5 mm. After measuring with a level, the lowest point height is determined to be 101.5 mm, and the highest point height is 104.5 mm. The overall height difference is 104.5 - 101.5 = 3 mm. Using the lowest point height measurement of 101.5 mm as a baseline, the relative height is calculated by subtracting 101.5 mm from the actual height of each point. For example, if a measuring point has a height of 103.2 mm, its relative height to the lowest point is 103.2 - 101.5 = 1.7 mm.

[0040] By using multi-point sensors to collect the actual height information of each measurement point and calculate the relative height, accurate basic data is provided for subsequent flatness testing. By setting a consistent height benchmark, the data consistency and reliability of the entire testing process are ensured, thereby improving the accuracy and reliability of flange flatness analysis.

[0041] The actual height data collected is subjected to noise filtering and attitude correction. Combined with the radius and angle information of the measurement points, the theoretical reference plane of the flange is fitted.

[0042] Furthermore, noise filtering and attitude correction are performed on the collected actual height data. Combined with the radius and angle information of the measurement points, the theoretical reference plane of the flange is fitted. This includes: performing multi-layer noise filtering on the collected actual height data to obtain filtered height data; performing full-parameter attitude correction based on the filtered height data to obtain attitude angle correction data; selecting the main reference point from the attitude angle correction data, using the main reference point as the forced passage point, and combining the three-dimensional distribution relationship of the radius-angle-height of the measurement points, using the constrained least squares method to fit the optimal theoretical reference plane, wherein the measurement points in the sealing line area are given a higher fitting weight than other measurement points.

[0043] Specifically, during actual measurement, sensors are affected by environmental factors such as electromagnetic interference, mechanical vibration, and limitations in sensor accuracy, resulting in noise in the acquired height data. Therefore, multi-layer noise filtering is performed on the acquired height data. This multi-layer noise filtering uses various filtering algorithms and techniques to remove noise interference from the height data at multiple levels. The initial filtering stage uses median filtering to suppress impulse noise in the height data, replacing each height data point with the median of its neighborhood. The intermediate filtering stage uses mean filtering, taking the neighborhood average of the height data after the first layer of filtering to further smooth the data and reduce the impact of random noise. Finally, by combining the elastic modulus of the flange material and calculating the height gradient and maximum curvature between adjacent measurement points, unreasonable height measurements can be further corrected. By combining various filtering methods, multiple types of noise can be effectively removed, resulting in more accurate filtered height data.

[0044] Then, full-parameter attitude correction is performed on the filter height data. Full-parameter attitude correction comprehensively considers various attitude parameters of the flange in space, such as pitch angle, yaw angle, and roll angle, to adjust the filter height data. Since the flange may not be in an ideal horizontal state during installation, its attitude will affect the accuracy of the height data. In this application, the flatness detection refers to the flatness detection of the flange sealing surface, not the horizontality relative to the horizontal plane. Flatness is a form tolerance, indicating the degree to which the surface deviates from an ideal plane, while horizontality is a directional tolerance, indicating the degree of inclination of the surface relative to the horizontal plane. The attitude angle information of the flange is acquired using professional attitude measurement equipment, such as an inertial measurement unit (IMU). For example, if the measured pitch angle of the flange is 5°, the yaw angle is 3°, and the roll angle is 1°, the filter height data is transformed using six degrees of freedom in space based on the attitude angle information. Through this transformation, the position of each measurement point is converted into the true three-dimensional coordinates in the flange coordinate system, thereby eliminating errors caused by tilt, rotation, etc., and obtaining the attitude angle correction data.

[0045] A primary reference point is selected from the attitude angle correction data. This primary reference point is a representative, stable point on the flange with accurate and reliable height measurement. It can be a location on the flange with symmetrical structure, uniform stress, and easy-to-measure location. The radius, angle, and height information of multiple measurement points are integrated to form a three-dimensional distribution relationship of radius-angle-height for each measurement point. Using the primary reference point as a forced pass-through point, and combining the three-dimensional distribution relationship of radius-angle-height, a constrained least squares method is used to fit the optimal theoretical reference plane. The three-dimensional distribution relationship of radius-angle-height reflects the position of each measurement point on the flange in the polar coordinate system, i.e., radius, angle, and corresponding height information. The constrained least squares method requires that the primary reference point must be located on the fitted plane during the fitting process, while minimizing the sum of squared distances from other measurement points to this plane, thus obtaining the optimal theoretical reference plane. Furthermore, the fitting weight of measurement points in the sealing line area is higher than that of other measurement points because the sealing performance of the flange mainly depends on the flatness of the sealing line area. Increasing the weight of measurement points in the sealing line area during the fitting process allows the fitted theoretical reference plane to be closer to the actual shape of the sealing line area, thus more accurately reflecting the sealing performance of the flange. For example, firstly, a differentiated weight allocation is performed for functional areas, assigning different weights to different functional areas. Among them, the sealing contact area, such as the annular zone between bolt holes, is crucial for sealing performance evaluation and is given the highest weight w=1.5. The bolt bearing area, such as directly below the bolt holes, is given a lower weight w=0.5 to reduce the impact of bolt preload deformation. Non-critical areas, such as the outer edge of the flange, are given a basic weight w=1.0. Physical constraints are set, including forced passage constraints, trend constraints, and symmetry constraints. Forced passage constraints require the fitted plane to pass through the main reference point. Trend constraints are based on the flange design parameters to ensure that the plane's tilt direction matches the expected stress deformation trend of the flange. Symmetry constraints ensure that the symmetry error of the fitted plane is less than a preset value for flanges that are theoretically symmetrical. Iterative reweighted least squares method is used to dynamically adjust the weights of each point based on the residuals of the previous fitting round. The fitting operation is repeated until the result converges, thereby obtaining the optimal theoretical reference plane that meets the actual working conditions and functional requirements of the flange.

[0046] Multi-layer noise filtering removes noise interference from the data, improving the accuracy of height data. Full-parameter attitude correction eliminates the influence of flange installation attitude deviations on height measurement, making the height data more reflective of the flange's flatness. By selecting the main reference point and using constrained least squares to fit the theoretical reference plane, combined with assigning higher weights to measurement points in the sealing line area, an ideal plane that meets the actual sealing performance requirements of the flange can be more accurately fitted. This allows for accurate calculation of the flatness deviation at each measurement point, improving the effectiveness and accuracy of flatness detection at the top of large flanges.

[0047] Furthermore, the collected actual height data undergoes multi-layer noise filtering to obtain filtered height data. This includes: primary filtering of the measured height data based on the median height of a time-series sliding window to eliminate impulse noise, wherein each measurement point is sampled continuously N times, and the height value sequence is filtered using a time-series sliding window, with N not less than 5; secondary filtering of the measured height data based on spatial neighborhood consistency to remove outlier measured height data, wherein K nearest neighbor points are found within a circular region of radius r centered on the current measurement point, the neighborhood standard deviation is calculated, outlier points are identified, r is 2-5 times the distance between measurement points, and K is not less than 4; and finally, the height gradient and maximum curvature between adjacent measurement points are calculated in conjunction with the elastic modulus constraint of the flange material to correct unreasonable measurement values, thus obtaining the final filtered height data.

[0048] Specifically, the median height of a time-series sliding window is used for primary filtering to eliminate impulse noise. Impulse noise is usually caused by strong interference to the sensor momentarily, such as electromagnetic interference or mechanical shock, resulting in abnormally high or low measurement values. For each measurement point, N consecutive samples are taken, where N is an integer not less than 5, to obtain a height value sequence. The median of the height value sequence is then filtered using a time-series sliding window, that is, the median of all height values ​​within the time-series sliding window is taken as the filtered height value at the center of the window.

[0049] Outlier height data is removed based on spatial neighborhood consistency. Spatial neighborhood consistency refers to the similarity and continuity of height values ​​of measurement points within a local spatial range. Taking the current measurement point as the center, K nearest neighbor points are found within a circular area of ​​radius r. The standard deviation of the height values ​​of the K points is calculated. If the difference between the height value of the current measurement point and the average height value of the neighboring points exceeds a certain multiple, such as 3 times the standard deviation, the point is identified as an outlier and removed, thus obtaining intermediate filtered data. Here, r is 2-5 times the distance between measurement points, and K is an integer not less than 4.

[0050] The elastic modulus of flange materials is obtained through material handbooks and experimental testing. For example, the elastic modulus of carbon steel is typically 200-210 GPa, while that of alloy steel varies from 190-210 GPa depending on its composition. For non-standard materials or when precise measurements are required, tensile testing can be used. The specific procedure involves applying an axial tensile force to a flange material specimen, measuring the ratio of the specimen's elongation to its original length (strain), and simultaneously recording the ratio of the tensile force to the specimen's cross-sectional area (stress). The result is calculated using the formula E=σ / ε, where E is the elastic modulus, σ is the stress, and ε is the strain. Using the elastic modulus of the flange material as a constraint, combined with the flange's stress conditions and geometric parameters (such as bolt preload and medium pressure) and geometric parameters (such as flange thickness and diameter), elasticity mechanics formulas, such as beam-plate theory or thin-plate theory, are used to calculate the theoretical deformation range of the flange under normal stress conditions. For example, according to thin-plate bending theory, the flange deflection w satisfies the biharmonic equation. 4 w = q / D, where q is the load per unit area and D is the bending stiffness, calculated using the formula D = Eh. 3 / 12(1 v 2 In this equation, E is the elastic modulus, h is the flange thickness, and v is Poisson's ratio. Based on boundary conditions, such as fixing the edge, the deflection equation is solved to obtain the deflection distribution w(x,y) on the flange surface. The height gradient can be approximated by the ratio of the deflection difference between two points to the distance. Curvature is the second partial derivative of the deflection. The maximum curvature occurs at the position of maximum deflection curvature. Based on this, the theoretical deformation range of the flange under normal stress is obtained, including the height gradient range and the maximum curvature. For any adjacent measurement points i and j, the height difference ΔH = H... j -H i Distance d between two points ij Calculate the height gradient G ij =ΔH / d ij The system combines the height values ​​of multiple adjacent points to fit a local surface, and calculates the local maximum curvature using second-order differences. The calculated height gradient and curvature are then compared with the maximum allowable deformation or curvature range of the material's elasticity. If the height gradient or curvature between adjacent measurement points exceeds this range, it indicates that the measurement data may be unreasonable. This unreasonable measurement value is then corrected based on surrounding reasonable data, such as using linear interpolation, to ensure that the height gradient conforms to the material's elastic modulus constraint, thus obtaining the final filtered height data.

[0051] For example, a measurement area is detected, and each measurement point is sampled N=5 times consecutively. If the sequence of 5 height values ​​obtained from measurement point A is [102.5, 10.3.2, 1000.0, 101.8, 104.1], where 1000.0 is simulated impulse noise, a sliding window with a window size of 3 is used. The median value of the first window [102.5, 10.3.2, 1000.0] is 103.2, the median value of the second window [10.3.2, 1000.0, 101.8] is 103.2, and the median value of the third window [1000.0, 101.8, 104.1] is 101.8. After median filtering, the primary filtered data is [103.2, 103.2, 101.8], which effectively eliminates the impulse noise 1000.0. Then, we set r=3mm and K=5. Assuming the current measurement point height is 104.5, the heights of its 5 neighboring points are [102.5, 103.2, 101.8, 104.1, 102.9], with an average of 102.9 and a standard deviation of approximately 0.89. We set the threshold to 3 times the standard deviation, i.e., 2.67. The difference between 104.5 and 102.9 is 1.6, which is less than 2.67, so this point is not an outlier. If another measurement point has a height of 98.0mm, and the difference between it and the neighborhood average exceeds 3 times the standard deviation, then this point is an outlier and is removed to obtain the intermediate filtered data. Given that the maximum height gradient between adjacent measurement points corresponding to the elastic modulus of the flange material is 0.1, if the distance between two adjacent measurement points is 2mm, and the height of one measurement point is 101.5 and the other is 103.7, the height difference is 2.2, and the height gradient is 2.2÷2=1.1, which exceeds the maximum height gradient of 0.1. At this time, the unreasonable measurement value should be corrected based on reasonable data from the surrounding area.

[0052] By using multi-layer noise filtering, impulse noise, outliers, and unreasonable measurement values ​​in the actual height data are effectively removed, greatly improving the accuracy and reliability of the height data. This ensures that the final flange flatness assessment results are true and credible, providing reliable and effective data support for flange flatness detection and correction.

[0053] Furthermore, based on the filtered height data, full-parameter attitude correction is performed to obtain attitude angle correction data, including: establishing a transformation matrix from the sensor coordinate system to the flange coordinate system; calculating the coordinate transformation correction factor in real time based on the collected attitude angles; using the correction factor, performing a six-degree-of-freedom spatial transformation on the height of each measurement point, recalculating the three-dimensional coordinates in the flange coordinate system, and obtaining the attitude angle correction data.

[0054] Specifically, a transformation matrix is ​​established from the sensor coordinate system to the flange coordinate system. The sensor coordinate system is a coordinate system defined by the sensor itself, and its origin and axis are usually related to the sensor's installation position and orientation. The flange coordinate system is a coordinate system established with the geometric center of the flange as the origin and a certain characteristic direction of the flange, such as the arrangement direction of the flange bolt holes, as the axis. The installation position and attitude of the sensor in the flange coordinate system are measured by an inertial measurement unit (IMU), such as the three translational components x0, y0, z0 and the three rotational components α, β, γ, which correspond to the rotation angles around the x, y, and z axes, respectively. The transformation matrix is ​​constructed using the homogeneous coordinate transformation principle. The transformation matrix is ​​a 4×4 square matrix, reflecting the spatial transformation relationship from the sensor coordinate system to the flange coordinate system.

[0055] The sensor's attitude angles are acquired in real time. Based on these acquired attitude angles, a coordinate transformation correction factor is calculated in real time, updating the rotation component in the transformation matrix to obtain the real-time transformation matrix. The rotation component is recalculated based on the new attitude angles, while the translation component remains unchanged. This real-time transformation matrix serves as the coordinate transformation correction factor, reflecting the accurate relationship of the coordinate transformation under the sensor's current attitude. Furthermore, the correction factor is used to perform a six-degree-of-freedom spatial transformation on the height of each measurement point. A six-degree-of-freedom spatial transformation refers to an object's ability to translate along the x, y, and z axes and rotate around these axes in three-dimensional space, totaling six degrees of freedom. For each measurement point, its coordinates in the sensor coordinate system are (x... s ,y s ,z s Expand it to homogeneous coordinates (x) s ,y s ,z s ,1), and then multiply with the real-time transformation matrix to obtain the coordinates (x,y) in the flange coordinate system. f ,y f ,z f ), where z f This refers to the height data of the measurement points after attitude correction. The corrected height data of all measurement points constitute the attitude angle correction data.

[0056] For example, the initial installation position of a sensor in the flange coordinate system is x0=10mm, y0=5mm, z0=0mm, and the initial attitude angles are α=0°, β=0°, γ=0°. An initial transformation matrix T is obtained. During the detection process, the sensor's attitude angle is acquired in real time and changes to α. real =1°, β real =0.5°, γ real =0.2°, and based on these real-time attitude angles, the rotation part of the transformation matrix is ​​updated to obtain the real-time transformation matrix T. realThe coordinates of a certain measurement point in the sensor coordinate system are (20, 15, 102.5). Expanding these coordinates to homogeneous coordinates (20, 15, 102.5, 1), and relating them to T... real Multiplying these values ​​yields the coordinates (29.98, 19.97, 102.48) in the flange coordinate system, where 102.48 mm represents the height of the measurement point after attitude correction. By processing all measurement points in this way, complete attitude angle correction data can be obtained.

[0057] By establishing a transformation matrix from the sensor coordinate system to the flange coordinate system and updating it in real time, the influence of sensor installation attitude deviation and real-time attitude changes on the measurement data can be accurately eliminated, ensuring that the height data of all measurement points are unified under the flange coordinate system, thereby further improving the accuracy and consistency of the height data.

[0058] Furthermore, the primary reference point is selected from the attitude angle correction data, including: based on the height reference reference, selecting points whose attitude angle correction data are within the P% range from all measurement points to form an initial candidate set, wherein P% is no greater than 15%; performing spatial density analysis on the initial candidate set to remove isolated abnormal low points and obtain a secondary candidate set; calculating the comprehensive evaluation value of each candidate point in the secondary candidate set in the height dimension, position dimension, and stability dimension; based on the comprehensive evaluation value, selecting the candidate point with the highest evaluation value as the final primary reference point, wherein the primary reference point satisfies the following conditions: its height and the height reference reference satisfy a distance range, its position is close to the theoretical sealing line and theoretical geometric center of the flange, and its height value has the smallest coefficient of variation in multiple measurements.

[0059] Specifically, based on a height reference benchmark, measurement points whose height values ​​fall within a range of P% above the benchmark are selected from all attitude angle correction data to form an initial candidate set. P% is a proportional parameter, not exceeding 15%, to avoid introducing outliers that are significantly too high or too low. Spatial density analysis is then performed on the initial candidate set. Combining the angle and radius distribution of the measurement points on the flange circumference, isolated, unsupported low-lying points are identified and eliminated. For example, by analyzing the distribution of other points within a 3mm radius of each point, if a point has almost no other points around it (i.e., very low spatial density), then this point is likely an isolated low-lying point and is eliminated to obtain a secondary candidate set, thus improving the quality of the candidate points.

[0060] For each candidate point in the secondary candidate set, a comprehensive evaluation value is calculated in the dimensions of height, position, and stability. The height dimension primarily considers the closeness of the candidate point's height to the height reference benchmark. The position dimension primarily considers whether the candidate point is close to the theoretical sealing line and theoretical geometric center of the flange. The stability dimension primarily considers the coefficient of variation (COP) of the candidate point's height value across multiple measurements. The COP is determined by calculating the ratio of the standard deviation to the mean of the height values ​​obtained from multiple repeated measurements of the same measurement point; that is, dividing the standard deviation of the height data of that measurement point by its average height value. A smaller COP indicates better stability. For example, for a point in the secondary candidate set, the difference between its height and the height reference benchmark is 2mm. In the position dimension, it is 1mm from the theoretical sealing line of the flange and 3mm from the theoretical geometric center. The COP of its height value across multiple measurements is 0.05. Appropriate weights are assigned to each dimension: 0.4 for height, 0.3 for position, and 0.3 for stability. Then, based on the scoring criteria for each dimension—such as a smaller height difference, a closer location, and a smaller coefficient of variation—the score for each point is calculated. These scores are then multiplied by their respective weights and summed to obtain the point's overall evaluation value. For example, if the point scores 80 points in the height dimension, 90 points in the location dimension, and 95 points in the stability dimension, the overall evaluation value would be 80 × 0.4 + 90 × 0.3 + 95 × 0.3 = 88.5 points.

[0061] Based on the comprehensive evaluation value, the candidate point with the highest evaluation value is selected as the final master reference point. This master reference point must also meet the following requirements: its height must be within a certain distance range from the height reference datum (e.g., ±3mm); its location must be close to the theoretical sealing line and theoretical geometric center of the flange (e.g., no more than 2mm from the theoretical sealing line and no more than 3mm from the theoretical geometric center); and its height value must have the smallest coefficient of variation across multiple measurements. For example, in the secondary candidate set, after calculating the comprehensive evaluation values ​​of each candidate point, one point has the highest comprehensive evaluation value of 92 points. This point's height difference from the height reference datum is 1mm, its distance from the theoretical sealing line is 0.5mm, its distance from the theoretical geometric center is 2mm, and its height value has a coefficient of variation of 0.03 across multiple measurements. Meeting all conditions, this point is selected as the final master reference point.

[0062] By determining the main reference point through a multi-layered screening and comprehensive evaluation mechanism, the fitting of the theoretical reference plane no longer depends on any single point or the lowest point in a purely mathematical sense, but is based on measurement points that are highly reasonable, spatially critical, and stable over a long period of time. This improves the stability and physical consistency of the plane fitting using the constrained least squares method, thereby ensuring that the calculated flatness deviation truly reflects the flatness condition of the flange sealing surface, and thus improving the reliability and accuracy of flatness assessment.

[0063] Furthermore, fitting the theoretical reference plane of the flange also includes: randomly selecting a predetermined proportion of measurement points as a validation set, wherein the measurement points in the validation set are those that did not participate in the fitting of the reference plane; calculating the average deviation and maximum deviation of the points in the validation set based on the theoretical reference plane; when the deviation exceeds a predetermined limit, using an iterative reweighted least squares method, dynamically adjusting the weights according to the residuals of the previous fitting, and repeating the fitting until convergence; during the iterative optimization process, retaining the reference plane obtained in each round and its corresponding validation set evaluation score, and after the iteration is completed, selecting the plane with the highest validation set evaluation score as the final theoretical reference plane.

[0064] Specifically, after initially fitting the theoretical reference plane of the flange, to avoid the fitting results being overly affected by the distribution of local measurement points or abnormal data, an iterative optimization mechanism based on a validation set is further introduced. A preset proportion of measurement points is randomly selected from the measurement points that have not participated in the fitting of the reference plane as the validation set. The preset proportion is a pre-set value, such as 5%, and the measurement points in the validation set have not participated in the fitting of the theoretical reference plane and are only used for evaluating the fitting effect.

[0065] After obtaining the fitted theoretical reference plane, substitute each measurement point in the validation set into the theoretical reference plane and calculate its average and maximum deviations from the plane. The average deviation is the average distance of all points in the validation set from the theoretical reference plane, reflecting the overall deviation of the validation points from the reference plane. The maximum deviation is the distance from the point in the validation set farthest from the theoretical reference plane, reflecting the extreme case of the validation point deviating from the reference plane. Determine whether the deviation exceeds a preset limit. The preset limit is a standard value set in advance based on the flange machining accuracy requirements, used to determine whether the fitted reference plane meets the accuracy requirements. If the average deviation or maximum deviation exceeds the preset limit, it indicates that the initially fitted reference plane is not accurate enough and needs further optimization. For example, assuming the preset limit is an average deviation of no more than 0.03 mm and a maximum deviation of no more than 0.15 mm, the calculated average deviation of 0.05 mm and maximum deviation of 0.2 mm both exceed the limit. Optimization is then performed using an iterative reweighted least squares method.

[0066] When using the iterative reweighted least squares method, the weights are dynamically adjusted based on the residuals from the previous fitting round. The residual refers to the distance from the measurement point to the fitting reference plane. By adjusting the weights according to the magnitude of the residuals, points with larger residuals receive more attention in subsequent fittings, thus gradually reducing the bias of these points. For example, after the first fitting round, the weights of points with larger residuals are appropriately increased. In the second fitting round, these points will have a greater impact on the fitting results, causing the fitted reference plane to be closer to these points. This process is repeated until convergence. Convergence means that as the number of iterations increases, the fitted reference plane no longer changes significantly, or the average bias and maximum bias no longer decrease significantly. During the iterative optimization process, the reference plane obtained in each round and its corresponding validation set evaluation score are retained. The validation set evaluation score comprehensively considers the average bias, maximum bias, and convergence stability. After the iteration is completed, the plane with the highest validation set evaluation score is selected as the final theoretical reference plane.

[0067] By introducing a validation set evaluation and iterative reweighting optimization mechanism, this step effectively avoids overfitting or local distortion problems that may occur with single fitting. This ensures that the theoretical reference plane not only performs well on the measurement points involved in the fitting, but also has high consistency and generalization ability on the data not involved in the fitting. This significantly improves the objectivity, stability and anti-anomaly ability of the theoretical reference plane, thereby ensuring the accuracy and reliability of the overall flange flatness evaluation results.

[0068] The relative height of each measurement point is compared with the height of the theoretical reference plane. The flatness deviation of the flange surface is obtained by comparison. If the flatness deviation exceeds the preset threshold, the abnormal location and flatness correction suggestions are automatically output.

[0069] Specifically, after obtaining the theoretical reference plane of the flange, the flatness deviation is calculated for each measurement point using the theoretical reference plane as an ideal geometric reference. Based on the known radius and angle information of the measurement point, its spatial position is substituted into the plane equation of the theoretical reference plane to calculate the theoretical height value of that position on the theoretical reference plane. The relative height of each measurement point is compared with the theoretical height to obtain the height difference of each measurement point relative to the theoretical reference plane. The height difference reflects the flatness deviation at the location of the measurement point. The deviation essentially reflects the degree of deviation of the flange sealing surface from the ideal plane at that position, and is a quantitative description of the flatness tolerance.

[0070] The calculated flatness deviation at each measurement point is compared with a preset threshold. The preset threshold is an upper limit of the allowable deviation range determined by factors such as the actual use requirements of the flange, processing accuracy standards, and industry specifications. For example, if the preset threshold is set to 0.1mm, when the flatness deviation at a certain measurement point exceeds the preset threshold, the system automatically identifies that location as an abnormal location and outputs corresponding flatness correction suggestions based on its spatial location, deviation direction, and magnitude. For example, it may suggest that local grinding, overall leveling, or a focused inspection of the sealing line area is required.

[0071] By comparing the measurement points with the theoretical reference plane, the flatness of the flange surface can be assessed intuitively and accurately, and potential quality problems can be detected in a timely manner, such as local bulges, depressions, or overall uneven deformation. The system can also automatically output the abnormal location and correction suggestions, which not only improves the efficiency and accuracy of flange flatness anomaly detection, but also provides clear direction and guidance for flange quality improvement and processing adjustment.

[0072] Furthermore, the method also includes: binding each test data with a unique flange identifier to establish a three-dimensional digital archive; performing time-series analysis on the test data of the same flange to fit the degradation curve of key indicators; and using the degradation curve of the key indicators in combination with the three-dimensional digital archive to generate predictive maintenance information before the planned shutdown.

[0073] Specifically, after completing a single flange flatness inspection and obtaining complete inspection results, the inspection data is bound to the flange's unique identifier. This unique identifier can be composed of equipment number, installation location code, specifications, and QR code / electronic tag information, ensuring a unique correspondence of data throughout its entire lifecycle. The spatial coordinates of the measurement points, relative height, flatness deviation, theoretical reference plane parameters, abnormal area markings, and correction suggestions obtained during the inspection are also uniformly stored to construct a three-dimensional digital profile of the flange. This three-dimensional digital profile can intuitively reflect the true data of the flange sealing surface and its changes over time.

[0074] For multiple inspection records of the same flange at different time points, time-series analysis is performed in chronological order. Representative key indicators, such as maximum flatness deviation, average deviation, and deviation change rate, are selected. Statistical regression methods, such as linear regression, exponential decay fitting, or trend fitting based on least squares, are used to establish degradation curves of these key indicators over time to reflect the deterioration process of flange flatness performance. Based on the degradation curves of the key indicators and historical maintenance and correction measures recorded in the 3D digital archive, the degradation curve parameters are corrected to account for the impact of human intervention or abnormal operating conditions on the degradation rate. Based on the corrected degradation curve parameters, time extrapolation methods, such as linear extrapolation, exponential decay prediction, or polynomial fitting extrapolation, are used to predict the values ​​of future key indicators before the next planned shutdown. When the prediction results show that a certain key indicator will approach or exceed the allowable threshold before the next planned shutdown time, corresponding predictive maintenance information is automatically generated, prompting the need for targeted maintenance, leveling, or replacement measures during the planned shutdown period.

[0075] By establishing a three-dimensional digital archive bound to the unique identifier of the flange and introducing degradation analysis based on time-series data, the single flatness inspection is transformed into a state management method for the entire life cycle of the equipment. This realizes the transformation from post-event inspection to pre-event prediction. It can not only continuously track the evolution of the flatness of the flange sealing surface, but also generate maintenance decision suggestions in advance without affecting the production plan. This effectively reduces the risk of sudden leakage and unplanned downtime, and significantly improves the safety, operation and maintenance efficiency and management intelligence of large flanges.

[0076] Example 2, based on the same inventive concept as the multi-point sensing-based method for detecting the flatness of the top of a large flange in the preceding examples, such as... Figure 2 As shown, this application provides a large flange top flatness detection system based on multi-point sensing, wherein the large flange top flatness detection system based on multi-point sensing includes: The height difference calculation module 11 is used to determine the lowest and highest points of the flange using a field level, calculate the overall height difference of the flange surface, and use the height measurement data of the lowest point as the height reference benchmark. The data recording module 12 is used to set up measurement points on the flange circumference at preset equal angles and record the radius and angle information of each measurement point. The relative height calculation module 13 is used to collect the actual height information of each measurement point using a multi-point sensor, and calculate the relative height of each measurement point relative to the lowest point according to the height reference benchmark. The theoretical reference plane fitting module 14 is used to perform noise filtering and attitude correction on the collected actual height data, and fit the theoretical reference plane of the flange by combining the radius and angle information of the measurement points. The height comparison module 15 is used to compare the relative height of each measurement point with the height of the theoretical reference plane, and obtain the flatness deviation of the flange surface by comparison. If the flatness deviation exceeds a preset threshold, the abnormal position and flatness correction suggestions are automatically output.

[0077] Furthermore, the data recording module 12 is also used to: retrieve the basic layout density template from the standard library according to the flange pressure rating and sealing surface type; increase the density of measurement points on the load transmission path on both sides of the bolt hole based on historical data analysis; and set the measurement points and correspondingly deploy sensors by calculating the distribution angle according to the equal division angle and the pre-increased measurement point density, using the basic layout density template as the reference point.

[0078] Furthermore, the data recording module 12 is also used to: acquire the equipment historical record database, extract leakage points and maintenance record information; identify abnormal point distribution and analyze abnormal probability of the recorded information to establish abnormal distribution characteristics; and perform sensor measurement demand analysis based on the abnormal distribution characteristics to determine the need to increase the density of measurement points.

[0079] Furthermore, the theoretical reference plane fitting module 14 is also used to: perform multi-layer noise filtering on the collected actual height data to obtain filtered height data; perform full-parameter attitude correction based on the filtered height data to obtain attitude angle correction data; select the main reference point from the attitude angle correction data, use the main reference point as the forced passage point, and combine the radius-angle-height three-dimensional distribution relationship of the measurement points to fit the optimal theoretical reference plane using the constrained least squares method, wherein the measurement points in the sealing line area are given a higher fitting weight than other measurement points.

[0080] Furthermore, the theoretical reference plane fitting module 14 is also used for: performing primary filtering on the measured height data based on the median height of a time-series sliding window to eliminate impulse noise, wherein each measurement point position is sampled continuously N times, and the height value sequence is filtered by median using a time-series sliding window, where N is not less than 5; removing outlier measured height data based on spatial neighborhood consistency to obtain intermediate filtered data, wherein, with the current measurement point as the center, K nearest neighbor points are found in a circular region of radius r, the neighborhood standard deviation is calculated, outlier points are identified, where r is 2-5 times the distance between measurement points, and K is not less than 4; and combining the elastic modulus constraint of the flange material, calculating the height gradient and maximum curvature between adjacent measurement points, correcting unreasonable measurement values ​​in the height data, and obtaining the final filtered height data.

[0081] Furthermore, the theoretical reference plane fitting module 14 is also used to: establish a transformation matrix from the sensor coordinate system to the flange coordinate system, calculate the coordinate transformation correction factor in real time based on the collected attitude angles; and use the correction factor to perform a six-degree-of-freedom spatial transformation on the height of each measurement point, recalculate the three-dimensional coordinates in the flange coordinate system, and obtain the attitude angle correction data.

[0082] Furthermore, the theoretical reference plane fitting module 14 is also used to: based on the height reference reference, select points from all measurement points whose attitude angle correction data are within the range of P% to form an initial candidate set, wherein P% is not greater than 15%; perform spatial density analysis on the initial candidate set, remove isolated abnormal low points, and obtain a secondary candidate set; calculate the comprehensive evaluation value of each candidate point in the secondary candidate set in the height dimension, position dimension, and stability dimension; based on the comprehensive evaluation value, select the candidate point with the highest evaluation value as the final main reference point, wherein the main reference point satisfies the following conditions: its height and the height reference reference satisfy a distance range, its position is close to the theoretical sealing line and theoretical geometric center of the flange, and its height value has the smallest coefficient of variation in multiple measurements.

[0083] Furthermore, the theoretical reference plane fitting module 14 is also used to: randomly select a preset proportion of measurement points as a validation set, wherein the measurement points in the validation set are measurement points that did not participate in the reference plane fitting; calculate the average deviation and maximum deviation of the points in the validation set based on the theoretical reference plane; when the deviation exceeds a preset limit, adopt the iterative reweighted least squares method, dynamically adjust the weights according to the residuals of the previous fitting, and repeat the fitting until convergence; during the iterative optimization process, retain the reference plane obtained in each round and its corresponding validation set evaluation score, and after the iteration is completed, select the plane with the highest validation set evaluation score as the final theoretical reference plane.

[0084] Furthermore, the system is also used to: bind each inspection data with a unique flange identifier to establish a three-dimensional digital archive; perform time-series analysis on the inspection data of the same flange to fit the degradation curve of key indicators; and use the degradation curve of the key indicators in combination with the three-dimensional digital archive to generate predictive maintenance information before the planned shutdown.

[0085] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The method and specific examples of large flange top flatness detection based on multi-point sensing in the foregoing embodiment one are also applicable to the large flange top flatness detection system based on multi-point sensing in this embodiment. Through the foregoing detailed description of the large flange top flatness detection method based on multi-point sensing, those skilled in the art can clearly understand the large flange top flatness detection system based on multi-point sensing in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0086] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0087] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A large flange top flatness detection method based on multi-point sensing, characterized in that, The method comprises the following steps: Determine the lowest point and the highest point of the flange by using a field level meter, calculate the overall height difference of the flange surface, and take the height measurement data of the lowest point as the height reference datum; Distribute the measurement points on the circumference of the flange according to the preset equal angle, and record the radius and angle information of each measurement point; Use a multi-point sensor to collect the actual height information of each measurement point, and calculate the relative height of each measurement point relative to the lowest point according to the height reference datum; Filter the collected actual height data for noise, correct the posture, combine the radius and angle information of the measurement points, and fit the theoretical reference plane of the flange; Compare the relative height of each measurement point with the height of the theoretical reference plane, and obtain the flatness deviation of the flange surface by comparison. If the flatness deviation exceeds the preset threshold, automatically output the abnormal position and flatness correction suggestion.

2. The large flange top flatness detection method based on multi-point sensing according to claim 1, characterized in that, Distribute the measurement points on the circumference of the flange according to the preset equal angle, comprising: Call the basic point distribution density template from the standard library according to the flange pressure rating and the sealing surface type; Based on historical record data analysis, increase the measurement point density on the load transmission path on both sides of the bolt hole; Take the basic point distribution density template as the reference point, calculate the distribution angle according to the equal angle and the pre-increased measurement point density, set the measurement points, and correspondingly distribute the sensors.

3. The large flange top flatness detection method based on multi-point sensing according to claim 2, characterized in that, Based on historical record data analysis, increase the measurement point density, comprising: Obtain the equipment historical record database, and extract the leakage point and maintenance record information; Identify the abnormal point distribution and analyze the abnormal probability of the record information, and establish the abnormal distribution characteristics; Based on the abnormal distribution characteristics, analyze the sensing measurement demand, and determine the increased measurement point density.

4. The large flange top flatness detection method based on multi-point sensing according to claim 1, characterized in that, Filter the collected actual height data for noise, correct the posture, combine the radius and angle information of the measurement points, and fit the theoretical reference plane of the flange, comprising: Filter the collected actual height data for noise, obtain filtered height data; Based on the filtered height data, perform full-parameter posture correction to obtain posture angle correction data; From the posture angle correction data, select the main reference point, take the main reference point as the forced passing point, combine the radius-angle-height three-dimensional distribution relationship of the measurement points, and use the least squares method with constraints to fit the optimal theoretical reference plane, wherein the fitting weight of the sealing line area measurement point is higher than that of other measurement points.

5. The large flange top flatness detection method based on multi-point sensing according to claim 4, characterized in that, Filter the collected actual height data for noise, obtain filtered height data, comprising: Based on the time sequence sliding window, filter the measured height data for primary filtering to eliminate impulse noise, wherein each measurement point position is continuously sampled for N times, the time sequence sliding window is used to filter the height value sequence for median filtering, and N is not less than 5; Based on the spatial neighborhood consistency, remove the outlier measurement height data to obtain intermediate filtered data, wherein the current measurement point is taken as the center, K nearest neighborhood points are found in the circular area with a radius r, the neighborhood standard deviation is calculated, the outlier points are identified, r is 2-5 times the measurement point spacing, and K is not less than 4; Combine the flange material elastic modulus constraint to calculate the height gradient and maximum curvature between adjacent measurement points, correct the unreasonable measurement values of the height data, and obtain the final filtered height data.

6. The large flange top flatness detection method based on multi-point sensing according to claim 4, characterized in that, Based on the filtering height data, full-parameter attitude correction is performed to obtain attitude angle correction data, including: A transformation matrix of the sensor coordinate system to the flange coordinate system is established, and a correction factor for coordinate conversion is calculated in real time according to the collected attitude angle; Using the correction factor, six-degree-of-freedom spatial transformation is performed on the height of each measurement point, and three-dimensional coordinates in the flange coordinate system are recalculated to obtain the attitude angle correction data.

7. The large flange top flatness detection method based on multi-point sensing according to claim 4, characterized in that, Screening a main reference point from the attitude angle correction data, including: Based on the height reference reference, points with attitude angle correction data within P% range are screened from all measurement points to form an initial candidate set, wherein P% is not greater than 15%; The spatial density analysis is performed on the initial candidate set to remove isolated abnormal low points to obtain a secondary candidate set; For each candidate point in the secondary candidate set, a comprehensive evaluation value in the height dimension, position dimension and stability dimension is calculated; Based on the comprehensive evaluation value, the candidate point with the highest evaluation value is selected as the final main reference point, wherein the main reference point satisfies the distance interval with the height reference reference, is close to the theoretical sealing line and the theoretical geometric center of the flange, and the height value has the minimum coefficient of variation in multiple measurements.

8. The large flange top flatness detection method based on multi-point sensing according to claim 4, characterized in that, Fitting the theoretical reference plane of the flange, further including: A predetermined proportion of measurement points are randomly selected as a verification set, and the measurement points in the verification set are measurement points that do not participate in the fitting of the reference plane; Based on the theoretical reference plane, the average deviation and the maximum deviation of the points in the verification set are calculated; When the deviation exceeds the preset limit value, the iterative reweighted least squares method is used to dynamically adjust the weight according to the residual error of the last round of fitting, and the fitting is repeated until convergence; During the iterative optimization process, the reference plane obtained in each round and the corresponding verification set evaluation score are retained, and after the iteration is completed, the plane with the highest verification set evaluation score is selected as the final theoretical reference plane.

9. The multi-point sensing based large flange top flatness detection method according to claim 1, wherein, Further including: Binding each detection data with the unique identification of the flange to establish a three-dimensional digital archive; Performing time series analysis on the detection data of the same flange over time to fit the degradation curve of the key indicators; Using the degradation curve of the key indicators in combination with the three-dimensional digital archive, predictive maintenance information is generated before predicting the planned shutdown.

10. A large flange top flatness detection system based on multi-point sensing, characterized in that, Steps for implementing the large flange top flatness detection method based on multi-point sensing in any one of claims 1 to 9, the large flange top flatness detection system based on multi-point sensing comprising: A height difference calculation module for determining the lowest point and the highest point of the flange using a field level, calculating the overall height difference of the flange surface, and taking the height measurement data of the lowest point as the height reference reference; A data recording module for arranging measurement points on the flange circumference at a predetermined equal angle and recording the radius and angle information of each measurement point; A relative height calculation module for collecting actual height information of each measurement point using a multi-point sensor, and calculating the relative height of each measurement point relative to the lowest point according to the height reference reference; A theoretical reference plane fitting module for performing noise filtering and attitude correction on the collected actual height data, and fitting the theoretical reference plane of the flange in combination with the radius and angle information of the measurement points; The height comparison module is used for comparing the relative height of each measuring point with the height of the theoretical reference plane, obtaining the flatness deviation of the flange surface by comparison, and automatically outputting the abnormal position and flatness correction suggestion if the flatness deviation exceeds the preset threshold.

Images