A method and system for non-destructive testing of internal defects of a bearing seal structure

Abstract

The application discloses a bearing sealing structure internal defect nondestructive testing method and system, and relates to the technical field of nondestructive testing.The method comprises the following steps: before and after the start and end of a measurement task, calibrating a physical reference built in the system to calibrate and quantify system drift; during the scanning process, through a coaxial reverse differential measurement mechanism, real-time immunity to dynamic errors is generated to generate pretreated three-dimensional point cloud data flow; finally, through vector projection analysis of actual geometric distortion topological morphology and a predefined topological fingerprint representing a systematic aging mode, the diagnosis and compensation of the health state of the system itself are realized.Through an internal and full-cycle error defense system, the application can generate a three-dimensional model highly consistent with a physical entity, and the fidelity and long-term reliability of the detection result are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of nondestructive testing, in particular to a bearing sealing structure internal defect nondestructive testing method and system. BACKGROUND

[0002] At present, for the internal defect detection of ring-shaped workpieces such as bearing sealing structures, a nondestructive testing system based on laser scanning is generally used. This kind of system obtains the three-dimensional profile data of the inner wall through a sensor moving and rotating inside the workpiece. Although the prior art can obtain high measurement accuracy under ideal conditions, in actual and variable industrial environments and long-term use cycles, there is a fundamental technical defect: the measurement results ultimately output lack credibility.

[0003] The inconsistency of this measurement result with the physical true state, i.e. the lack of fidelity, is rooted in the fact that the existing system cannot systematically resist error erosion from multiple dimensions. First, changes in environmental temperature and heat generated by the device itself during long-term operation will cause environmental drift in the entire measurement chain, making the measurement data benchmarks at different times or under different conditions inconsistent, losing the value of accurate comparison. Second, when the detection system moves inside the workpiece, due to the inherent deviation of mechanical positioning and the unevenness of the inner wall surface, real-time attitude shaking will inevitably occur, introducing dynamic misalignment errors and directly polluting the original collected data. More deeply, the mechanical parts of the detection system, such as positioning rollers and bearings, will wear and age over time, and this system aging will introduce a slow, cumulative and difficult-to-detect systematic deviation, and the prior art generally lacks online self-diagnosis and compensation capabilities for this. SUMMARY

[0004] The present application provides a bearing sealing structure internal defect nondestructive testing method and system to solve the problem in the prior art that the measurement results lack credibility due to the comprehensive influence of multiple error sources for the internal defect detection of ring-shaped workpieces such as bearing sealing structures.

[0005] In view of the above problems, in a first aspect, the present application provides a bearing sealing structure internal defect nondestructive testing method, comprising the following steps:

[0006] Before the measurement task starts, the physical reference built-in the detection system is calibrated to generate an initial reference calibration parameter set;

[0007] Based on the initial reference calibration parameter set, the inner wall of the bearing sealing structure is scanned through an error immunity mechanism to generate a preprocessed three-dimensional point cloud data stream;

[0008] After the measurement task ends, the physical reference is calibrated again and compared with the initial reference calibration parameter set to quantify the total system drift amount during the measurement task;

[0009] Based on the pre-processed three-dimensional point cloud data stream and the total system drift amount, the mechanical health status of the detection system is diagnosed, and a high-fidelity three-dimensional model of the bearing sealing structure is generated.

[0010] In a second aspect, the application also provides a bearing sealing structure internal defect nondestructive detection system.

[0011] The technical scheme provided by the application has the following technical effects: the fidelity and long-term reliability of the detection result are improved. Specifically, first, by calibrating the built-in physical reference and quantifying the drift, it is ensured that the measurement data at different times and spaces have a unified reference, and accurate comparison and wear trend analysis across cycles are realized; second, by the error immunity mechanism, the influence of dynamic disturbance is eliminated from the data acquisition source, ensuring the cleanliness and robustness of the original data; finally, by the self-diagnosis algorithm based on the topological fingerprint, online monitoring and compensation of the health status of the system itself are realized, ensuring the credibility of the measurement system throughout its life cycle, so that the finally generated three-dimensional model can truly and accurately reflect the physical state of the measured object. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 A flowchart of a bearing sealing structure internal defect nondestructive detection method provided by an embodiment of the application is shown.

[0013] Figure 2 A structural schematic diagram of a bearing sealing structure internal defect nondestructive detection system provided by an embodiment of the application is shown. DETAILED DESCRIPTION

[0014] The above technical scheme will be described in detail below in combination with the drawings and specific embodiments, so that the above technical scheme can be better understood. Obviously, the described embodiments are only part of the embodiments of the application, not all embodiments of the application, and it should be understood that the application is not limited to the example embodiments for explaining the application. Based on the embodiments of the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application. In addition, it should be noted that, for convenience of description, only parts related to the application are shown in the drawings, not all.

[0015] Please refer to Figure 1 A bearing sealing structure internal defect nondestructive detection method, the method comprising the following steps:

[0016] Before the measurement task starts, calibrate the physical reference built-in the detection system to generate an initial reference calibration parameter set;

[0017] Based on the initial reference calibration parameter set, scan the inner wall of the bearing seal structure through an error-immune mechanism to generate a pre-processed three-dimensional point cloud data stream;

[0018] After the measurement task ends, calibrate the physical reference again and compare it with the initial reference calibration parameter set to quantify the total system drift during the measurement task;

[0019] Based on the pre-processed three-dimensional point cloud data stream and the total system drift, diagnose the mechanical health status of the detection system and generate a high-fidelity three-dimensional model of the bearing seal structure.

[0020] The present application provides a bearing seal structure internal defect non-destructive testing method and system, the overall workflow of which is uniformly scheduled and executed by a central processor and control firmware integrated in the detection system. The central processor and control firmware are responsible for analyzing external instructions, controlling the mechanical movement of the detection system, and executing all data acquisition, calibration, and diagnosis algorithms disclosed in the present application to ultimately generate a three-dimensional model that can truly reflect the internal defect status of the bearing seal structure.

[0021] The implementation of the entire detection method begins with the triggering and initial calibration of the measurement task. After the central processor and control firmware receive the start measurement task instruction sent externally, the central processor and control firmware first enter the reference calibration and system drift quantification stage. In the initial step of this stage, the central processor and control firmware control the measurement module inside the high-fidelity inertial measurement and calibration unit (HIMCU) set at the front end of the robot to perform a complete 360-degree rotation scan on a physical reference integrated at the front end of the HIMCU module.

[0022] This physical reference, in the preferred embodiment of the present application, is a ring-shaped workpiece, i.e., a reference ring. In order to fundamentally eliminate the influence of environmental temperature changes on the dimensional stability of the measurement reference itself, the reference ring is precisely machined from a material with ultra-low thermal expansion coefficient. A feasible material selection is Invar alloy (Invar 36), which has a thermal expansion coefficient lower than , which can ensure that the geometric dimensions of the reference ring remain highly stable under different temperature environments, thereby providing a reliable, environmentally invariant physical reference for the entire detection method.

[0023] After the initial scan of the physical reference is completed, the on-board processor within the HIMCU module immediately executes an initial reference calibration parameter set generation algorithm. The on-board processor has a digitized reference ring true value model pre-stored within it, which precisely defines the ideal inner diameter and profile of the reference ring under standard conditions. The on-board processor compares the just-acquired set of scan data points with the pre-stored digital true value model point by point, and through least square fitting, calculates the systematic deviations existing in the entire optical measurement chain under the current working conditions. These deviations are quantified as a set of calibration parameters, mainly including the measurement zero point deviation and the measurement gain coefficient. This set of parameters collectively constitutes an initial reference calibration parameter set (BCP0) with a task start time stamp, and is stored in the on-board processor memory as the basis for the first layer correction of all subsequent measurement data.

[0024] After the scan task of the entire bearing seal structure inner wall is completed, but before the detection system is completely removed from the structure, the central processor and control firmware again enter the final step of the reference calibration and system drift quantification phase. The central processor and control firmware instruct the HIMCU module to perform a second complete 360-degree rotation scan of the physical reference ring with exactly the same parameters and manner as the initial calibration. The on-board processor then generates a final reference calibration parameter set (BCP1) from the second scan data. Immediately after, the on-board processor accurately calculates the total drift of the entire measurement system during the entire measurement task period by vector subtraction of BCP1 from BCP0 stored at the task start time, due to factors such as device self-heating, slight environmental temperature fluctuations, etc. This total drift is constructed as a task period drift correction vector, which is used for the final, global drift compensation of the entire three-dimensional point cloud data collected.

[0025] After generating the initial reference calibration parameter set, the central processor and control firmware immediately control the detection system to enter the bearing seal structure and begin the error-immune three-dimensional profile generation process. The core of this process is that the measurement unit within the HIMCU module adopts a coaxial reverse differential measurement mechanism to actively immunize dynamic errors introduced by unstable robot posture from the physical source of data acquisition. In this mechanism, the measurement unit contains two laser displacement sensors, sensor A and sensor B, which are rigidly installed in the same radial direction of the same rotating shaft in a back-to-back manner, so that their measurement beams are emitted along the same straight line.

[0026] When the HIMCU module is scanning the structure inner wall, if its rotation center produces a radial offset amount Δx relative to the geometric center of the structure inner wall, then the measurement reading of sensor A will be approximately equal to the "true radius ", while at the same time, the measurement reading of sensor B approximately equal to the "true radius (where R is the true radius). The HIMCU's on-board processor sums these two readings at each sampling instant to obtain the current cross-sectional diameter In this way, the radial offset Ax is mathematically directly cancelled out, ensuring that the measured cross-sectional diameter remains highly accurate even in the presence of minor robotic wobble.

[0027] In the specific real-time data acquisition process, the HIMCU's on-board processor will first call the initial baseline calibration parameter set (BCP0) stored in memory to correct the raw distance readings output by sensor A and sensor B for zero and gain, respectively, before performing the above-mentioned differential summation operation. This ensures that each data point participating in the differential operation is already clean data that has undergone first-level baseline calibration under the current working conditions. This real-time data processing flow ensures that the data input to the subsequent link eliminates both the system's static baseline deviation and the dynamic offset error in the measurement process.

[0028] After the above real-time correction and fusion processing, the on-board processor will combine the cross-sectional diameter value calculated at each sampling instant with the angular encoder reading at that instant and the linear displacement encoder reading along the axial movement to jointly organize a data point with accurate three-dimensional spatial coordinates (usually r, 0, z in cylindrical coordinates) and a collection time stamp. As the detection system continues to move and scan the inner wall of the bearing seal structure, a large number of data points are continuously generated and constructed by the on-board processor into a pre-processed three-dimensional point cloud data stream, which is transmitted in real time to the central processor and control firmware through the internal bus for subsequent final processing.

[0029] After completing the entire scanning task and calculating the task period drift correction vector, the central processor and control firmware enter the core algorithm stage of system health diagnosis and final model correction. The first step in this stage is the extraction of geometric distortion topology. The central processor and control firmware first apply the task period drift correction vector to the complete pre-processed three-dimensional point cloud data stream to perform global drift compensation and obtain a set of final calibrated three-dimensional point clouds for analysis. Subsequently, the algorithm extracts coordinate point sets belonging to each sun apex from the calibrated three-dimensional point cloud according to pre-set geometric feature recognition rules (e.g., identifying local maximum points of radial distance).

[0030] For each set of sunline vertex coordinate points on a cross section, the algorithm uses the least square method to fit a reference circle that best approximates these vertices. The center and radius of this reference circle are considered as the ideal geometric center and radius of the current cross section. Next, the algorithm calculates the radial deviation of each actually measured sunline vertex relative to the fitted reference circle (i.e., the distance from the actual vertex to the center of the circle minus the radius of the fitted circle). For a complete 360-degree scanning cross section, the radial deviations of all N sunline vertices are arranged in order of their rotation angles, thereby constructing an N-dimensional radial deviation distribution vector (R) ). This vector completely and topologically describes the mathematical expression of the cross section's geometric distortion pattern.

[0031] In order to accurately identify the systematic deviation introduced by the detection system's own mechanical aging from the complex geometric distortion topological pattern, the present application adopts a matching method based on pre-defined aging pattern topological fingerprints. The core of this method is that one or more pre-defined aging pattern base vectors (B ) are stored in the memory of the central processor and control firmware. Each base vector is an N-dimensional, normalized standard vector that mathematically describes an ideal radial deviation distribution pattern caused by the aging of a specific mechanical component (e.g., 0.1 mm wear of the positioning roller in the Y-axis direction).

[0032] The generation of these base vectors is obtained by simulating the accurate kinematic model of the detection system before the system is shipped. The kinematic simulation models the geometric relationships and physical properties of the robot's mechanical links, joints, rollers, and springs. By applying a virtual wear amount to a specific component in the model, the simulation software can accurately calculate the regular and periodic offset of the HIMCU's rotation center during the 360-degree scanning process caused by this wear. This offset ultimately reflects in the radial deviation of the sunline vertices, forming a deviation distribution with a specific mathematical pattern (e.g., a standard cosine or sine function pattern). After sampling and normalizing this ideal deviation distribution pattern, a topological fingerprint, i.e., a pre-defined aging pattern base vector, is obtained.

[0033] After obtaining the actually measured radial deviation distribution vector (R ) and the pre-stored topological fingerprint (B ), the central processor and control firmware begin to execute the pattern matching measurement algorithm based on vector projection analysis. The purpose of this algorithm is to quantify to what extent the actually observed distortion pattern can be attributed to this known, systematic aging pattern. The core principle is to project R onto B defined "direction". The length of this projection represents the component in the direction of the component in the direction of the

[0034] To eliminate the influence of the total distortion energy magnitude and focus only on the shape similarity purity, the algorithm further calculates a normalized projection ratio. The calculation of this ratio is as follows: divide the projection vector in the direction of the component by the magnitude of the component itself (i.e. the ). This ratio is a scalar between 0 and 1: if the ratio is close to 1, it means that the actual observed distortion shape is almost perfectly consistent with the predefined "aging fingerprint", and it can be highly confident that this is caused by the aging of this particular component; if the ratio is close to 0, it means that the actual distortion shape is not related to this aging pattern, and it is likely to be random noise or other unmodeled factors. In this way, the algorithm successfully quantitatively decouples the systematic aging distortion from the complex total distortion.

[0035] To further improve the reliability of the diagnosis result, before finally determining the aging degree, the algorithm also introduces an evaluation of the reliability of the measurement process itself, i.e. a confidence-based weighted system aging index calculation. In this step, the central processor and control firmware will call the attitude data recorded synchronously by the inertial measurement unit (IMU) inside the HIMCU module during the inner wall scanning. The algorithm quantifies the degree of shaking of the detection system during the journey by calculating the variance or standard deviation of the gyroscope angular velocity data throughout the journey. A smaller variance value represents a smoother scanning process and higher data acquisition quality. This variance value is normalized and functionally mapped (e.g. through a sigmoid function) to be converted into an attitude stability factor between 0 and 1, which is used as the confidence weight for the final calculation ( ).

[0036] Subsequently, the central processor and control firmware use the following formula to calculate the final system aging index ( ): wherein, is the system aging index; is the confidence weight just calculated; is the normalized projection ratio calculated in the previous step; γ is a preset nonlinear amplification factor (e.g. γ = 2) to amplify the influence of high matching ratio.

[0037] In a specific numerical example, suppose the algorithm calculates that the normalized projection ratio is 0.9 (indicating high shape matching); the scanning process is relatively smooth, and the calculated confidence weight The non-linear amplification factor γ is set to 2. Then the system aging index is calculated as follows: .

[0038] After the system aging index is calculated, the algorithm enters the final compensation correction and report generation stage of the model. On the one hand, the central processor and control firmware will execute a compensation algorithm that will subtract from the original point cloud data the part of the radial deviation distribution vector (R) that is identified as belonging to systematic aging (i.e. its projection component on the R axis). In this way, the measurement errors introduced by the detection system itself are accurately removed, and a three-dimensional model is finally generated that has been corrected for all errors and can truly reflect the state of the inner wall of the bearing seal structure. On the other hand, the central processor and control firmware will compare the calculated system aging index (e.g. 0.7695) with the pre-set health status threshold (e.g. <0.3 is healthy, 0.3-0.7 is warning, >0.7 is serious). Based on the comparison result, the system will automatically generate a system health status report containing the current aging index value and giving clear maintenance recommendations, such as: "System aging index is 0.7695, status is serious, it is recommended to check and maintain the Y-axis positioning roller". The report will be presented to the user together with the final three-dimensional model.

[0039]

[0040] In order to realize the above detection method, the present application also provides a non-destructive testing system for internal defects of a bearing seal structure. Referring to the attached Figure 2 , the system is preferably a robot that can move autonomously inside the bearing seal structure.

[0041] In terms of specific system hardware architecture, the detection system includes a direct driving unit and a passive centering positioning unit. The direct driving unit can be composed of a motor, a reducer and a driving wheel, and is responsible for driving the entire system to move forward or backward in the axial direction of the inner wall of the structure. The passive centering positioning unit makes the system roughly aligned with the geometric center of the inner wall of the structure during the movement process through multiple groups of circumferentially distributed spring-loaded roller mechanisms.

[0042] ​​​The core hardware of the present application is integrated in a high fidelity inertial measurement and calibration unit (HIMCU) module, which is mounted at the front end of the entire detection system. The physical structure of the HIMCU module is designed as a highly integrated compact unit, which encapsulates a reference ring made of ultra-low thermal expansion coefficient material as a physical reference, a co-axial differential measurement module composed of two back-to-back coaxially mounted laser displacement sensors, and an on-board processing and diagnostic unit integrated with a microprocessor, memory and an inertial measurement unit (IMU).

[0043] In terms of system function module division, the detection system contains a central processor and control firmware as the system master. The central processor is configured to execute all the method steps described in the present application. Specifically, the functional modules inside the central processor are divided into:

[0044] Task scheduling and control module: responsible for receiving external instructions and controlling the movement of the direct drive unit and the rotation scanning of the HIMCU module.

[0045] Data processing and algorithm engine module: this module is configured to execute the reference calibration and system drift quantification algorithm; execute the error immune three-dimensional profile generation algorithm; and finally execute the system health diagnosis and final model correction algorithm including aging topology fingerprint matching and system aging index calculation. Through the cooperative work of these functional modules, the detection system can completely realize all the technical steps described, thereby outputting a three-dimensional model with multiple source errors removed and a system health status report.

[0046] The present detailed embodiment elaborates the internal operation logic of each main functional module, aiming to make the detailed basis and explanation for understanding and implementation by those skilled in the art. It needs to be emphasized that the above description constitutes a specific, preferred embodiment, but the concept of the present application is not limited to this. Any equivalent transformation, modification or improvement based on the core spirit of the present application, without departing from the technical principles and scope disclosed in the specification, as long as the same or similar technical effects can be achieved, should be considered to fall within the scope of the present application.

Claims

1. A non-destructive testing method for internal defects in a bearing sealing structure, characterized in that, Includes the following steps: Before the measurement task begins, the physical reference built into the detection system is calibrated to generate an initial reference calibration parameter set; Based on the initial benchmark calibration parameter set, the inner wall of the bearing sealing structure is scanned through an error immunity mechanism to generate a preprocessed three-dimensional point cloud data stream; After the measurement task is completed, the physical reference is recalibrated and compared with the initial reference calibration parameter set to quantify the total system drift during the measurement task. Based on the preprocessed 3D point cloud data stream and the total system drift, the mechanical health status of the detection system is diagnosed, and a high-fidelity 3D model of the bearing sealing structure is generated. The steps for diagnosing the mechanical health status of the detection system include: From the preprocessed 3D point cloud data stream, extract the geometric distortion topology that represents the deviation between the actual measured contour and the ideal geometric shape; The geometric distortion topology is matched with at least one predefined topological fingerprint that characterizes the radial deviation distribution pattern caused by the aging of the mechanical components of the detection system to determine the degree of aging of the detection system. The step of performing morphological matching measurement includes performing vector projection analysis on the geometrically distorted topological morphology in one or more of the topological fingerprint directions; The vector projection analysis includes: The geometrically distorted topology is constructed as a radial deviation distribution vector; The topological fingerprint is constructed as a predefined aging mode basis vector; The proportion of the deviation caused by the aging of the mechanical component in the total geometric distortion is quantified by calculating the normalized projection ratio of the radial deviation distribution vector onto the predefined aging mode basis vector, thereby determining the degree of aging.

2. The method for non-destructive testing of internal defects in a bearing sealing structure according to claim 1, characterized in that, The physical reference is made of a material with an ultra-low coefficient of thermal expansion; and the error immunity mechanism includes synchronous ranging using coaxial reverse differential measurement.

3. The non-destructive testing method for internal defects in a bearing sealing structure according to claim 1, characterized in that, The method further employs a system aging index to quantify the degree of aging; and the calculation of the system aging index includes weighting the normalized projection ratio based on a confidence weight characterizing the attitude stability of the detection system during the scanning process.

4. The non-destructive testing method for internal defects in a bearing sealing structure according to claim 3, characterized in that, The system aging index Determined by the following formula: in, The confidence weight is... Let the radial deviation distribution vector be... Let be the predefined aging mode basis vector, proj be the vector projection operation, and γ be the nonlinear amplification factor.

5. The non-destructive testing method for internal defects in a bearing sealing structure according to claim 1, characterized in that, The predefined topological fingerprint is generated by simulating the kinematic model of the detection system.

6. A non-destructive testing system for internal defects in a bearing sealing structure, characterized in that, It includes multiple functional units configured to perform a non-destructive testing method for internal defects in a bearing seal structure as described in any one of claims 1 to 5.

Images