Wire insulation layer sample measurement method and system

Abstract

The invention relates to the technical field of wire insulation layer measurement and evaluation, in particular to a wire insulation layer sample measurement method and system. The method comprises the following steps: scanning and measuring the thickness of an insulating layer of a numbered sample by using a laser thickness gauge to generate two-dimensional thickness distribution diagram data; performing thickness uniformity analysis according to the two-dimensional thickness distribution diagram data to obtain thickness uniformity data; selecting a tensile strength test sample according to the thickness uniformity data to obtain tensile strength test sample selection data; performing tensile strength test according to the tensile strength test sample selection data to obtain tensile strength test data; conducting wire insulation layer performance evaluation according to the two-dimensional thickness distribution diagram data, the thickness uniformity data and the tensile strength test data to obtain wire insulation layer performance evaluation data so as to conduct wire insulation layer sample measurement auxiliary operation. Through accurate thickness measurement and uniformity analysis, material waste in the production process can be effectively reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of measuring and evaluating wire insulation layers, and in particular to a wire insulation layer measuring sample method and system. Background Art

[0002] The quality of the conductor insulation layer is crucial to the safety, stability and reliability of the cable. With the continuous development of power systems and electronic equipment, the performance requirements for conductor insulation are also increasing. Especially in the fields of high-voltage power transmission, electronic equipment connection and communication lines, the thickness uniformity and tensile strength of the conductor insulation layer directly affect its durability and safety. Therefore, how to efficiently and accurately detect and evaluate the conductor insulation layer has become an important research topic in materials science, engineering technology and quality control.

[0003] Traditional methods for measuring wire insulation rely on manual inspection or contact measurement equipment. These methods have problems such as low accuracy, low efficiency, and cumbersome operation. For example, when using tools such as vernier calipers and micrometers for measurement, only limited measurement point data can be obtained, and the uniformity analysis requirements of a wide range of samples cannot be met. In addition, these traditional methods are mostly manual operations, with high human errors, and are easily affected by environmental factors during the detection process, and the accuracy and reliability of the data cannot be guaranteed. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a wire insulation layer measurement sample method and system to solve at least one of the above technical problems.

[0005] The present application provides a method for measuring a wire insulation layer sample, comprising the following steps: Step S1: Scan and measure the thickness of the insulation layer of the numbered sample using a laser thickness gauge to generate two-dimensional thickness distribution map data; Step S2: performing thickness uniformity analysis according to the two-dimensional thickness distribution map data to obtain thickness uniformity data; Step S3: Selecting tensile strength test samples according to the thickness uniformity data to obtain tensile strength test sample selection data; Step S4: performing a tensile strength test according to the tensile strength test sample selection data to obtain tensile strength test data; Step S5: Conducting wire insulation layer performance evaluation based on the two-dimensional thickness distribution map data, thickness uniformity data and tensile strength test data to obtain wire insulation layer performance evaluation data for conducting wire insulation layer sample measurement auxiliary operations.

[0006] In the present invention, by scanning the sample with a laser thickness gauge, high-precision, non-contact insulation layer thickness measurement can be achieved, which has higher accuracy and efficiency than the traditional physical contact measurement method, and will not cause damage to the sample surface. By generating a thickness distribution map, the thickness change of the insulation layer at different positions can be more intuitively presented. Thickness uniformity is a key factor affecting the performance of the wire insulation layer. By performing uniformity analysis on the two-dimensional thickness distribution map data, it is possible to identify uneven areas, which cause the insulation layer performance to be unstable, thereby affecting the electrical performance and safety of the wire. The selection of tensile strength test samples based on the thickness uniformity data can effectively reduce unnecessary sample testing, while ensuring that representative samples with potential problems are selected for in-depth testing. Doing so not only improves test efficiency, but also saves time and cost. The performance evaluation of the wire insulation layer is performed by combining the two-dimensional thickness distribution map data, thickness uniformity data and tensile strength test data. Multi-dimensional data fusion provides a more comprehensive and accurate performance evaluation, which can not only understand the thickness distribution of the insulation layer, but also master its mechanical properties, thereby comprehensively evaluating its quality.

[0007] Preferably, step S1 specifically includes: Perform spatial positioning and optical axis calibration on the laser thickness gauge to generate laser positioning data; Reconstruct the sample reference surface through a laser thickness gauge according to the laser positioning data to obtain reference surface data; According to the reference surface data, multi-point thickness sampling is performed by a laser thickness gauge to obtain preliminary thickness sampling matrix data; Perform thickness distribution interpolation calculation according to the thickness preliminary sampling matrix data to obtain thickness distribution interpolation data; A dynamic distribution map is generated according to the thickness distribution interpolation data to obtain two-dimensional thickness distribution map data.

[0008] In the present invention, a laser thickness gauge is used to scan and collect data of the sample, thereby avoiding the contact error generated in the traditional measurement method. By reconstructing the sample reference plane through a laser thickness gauge, it can be ensured that a suitable reference plane is selected on the irregular sample surface. The generated reference plane data provides a reliable reference framework for thickness measurement, avoiding errors caused by the irregular shape of the sample surface. Through interpolation calculation, multi-point sampling data is converted into continuous thickness distribution data. This interpolation method can help fill the spatial data gaps existing in the sampling process, generate smoother and more accurate thickness distribution data, and avoid the influence of rough discrete data on subsequent analysis. From laser positioning to sample reference plane reconstruction, and then to thickness data sampling, interpolation calculation and dynamic distribution map generation, the entire process realizes hierarchical measurement and analysis from coarse to fine.

[0009] Preferably, the spatial positioning and optical axis calibration are specifically: By using the laser thickness gauge’s integrated inertial sensors and external vision sensors, the spatial position data of the equipment is captured in real time; Perform posture calculation based on spatial position data to obtain device posture data; Perform posture error analysis based on the equipment posture data to obtain posture error data; Perform attitude calibration according to attitude error data to obtain attitude calibration data; Perform laser optical axis alignment on the posture calibration data to obtain optical axis alignment data; Perform error correction according to the optical axis alignment data to obtain error correction data; Laser positioning data is generated according to the spatial position data, attitude calibration data, optical axis alignment data and error correction data to obtain laser positioning data.

[0010] In the present invention, by using integrated inertial sensors and external visual sensors to capture real-time spatial position data, the spatial position and posture changes of the device can be effectively tracked. Through posture error analysis and posture calibration, the posture deviation of the device can be accurately corrected, and the error caused by instability or external interference during the scanning process of the device can be eliminated. The device after posture calibration can work stably in a more complex or dynamic environment, thereby improving the accuracy of the measurement data. The optical axis calibration step eliminates the error caused by misalignment of the light beam during the laser measurement process by accurately aligning the laser optical axis. Through fine optical axis alignment, the vertical relationship between the laser beam and the sample surface can be ensured, thereby minimizing the laser measurement error and ensuring the accuracy and consistency of the scanning data. By correcting the optical axis error, the measurement accuracy can be further improved, especially in complex environments (such as under the influence of factors such as equipment vibration and thermal expansion). Error correction can dynamically adjust the error of the equipment to ensure the reliability of each measurement result.

[0011] Preferably, the sample reference plane reconstruction is specifically: Perform a preliminary scan of the sample using a laser thickness gauge according to the laser positioning data to obtain sample scanning data; Reconstructing a local reference plane according to the sample scanning data to obtain first reference plane data; Perform minimum convex hull reconstruction according to the sample scanning data to obtain the second reference surface data; Performing shape feature fusion on the first reference plane data and the second reference plane data to obtain reference plane data; The preliminary scan of the sample is as follows: Scan the sample surface with a laser thickness gauge to obtain point cloud data of the sample surface; Perform particle iterative filtering on the sample surface point cloud data to obtain scanning noise reduction data; Perform KD tree region division according to the scanned denoising data to obtain scanned division data; voxelize and resample the scanned divided data to obtain sample resampled data; The spatial coordinates are aligned according to the sample scanning data and the sample resampling data to obtain the sample scanning data.

[0012] In the present invention, the surface of the sample is scanned by a laser thickness gauge, and particle iterative filtering is used for denoising, which can effectively remove the noise in the scanned data and ensure the high quality of the data. According to the complexity of the sample surface, the accuracy of the regional division is flexibly adjusted to adapt to samples of different shapes, sizes and complexities. It can effectively process samples with different surface features and avoid the overly simple or overly complex division methods affecting the final data quality. Through local reference plane reconstruction, the sample surface can be processed more finely, and the local changes of the sample surface are taken into account when the reference plane is aligned. The minimum convex hull algorithm can effectively find the most compact outer surrounding area of ​​the sample surface and use it as the reference plane. This method can provide a stable and relatively robust reference plane solution when the sample surface has an irregular shape or convexity. Through the minimum convex hull reconstruction, the influence of surface noise on the result can be avoided, and the stability and reliability of the reference plane data can be improved. By fusing the first reference plane data and the second reference plane data with shape features, the advantages of local reference plane reconstruction and minimum convex hull reconstruction can be comprehensively considered to obtain more accurate reference plane data, and the fusion can reduce the deviation caused by a single method, while improving the accuracy and applicability of the entire reference plane reconstruction.

[0013] Preferably, the local reference plane reconstruction is specifically: Extracting sample morphological features from the sample scanning data to obtain sample morphological feature data; Performing preliminary area division on the sample scanning data according to the sample morphological characteristic data to obtain first sample division data; Performing morphological clustering calculation according to the first sample partitioning data to obtain sample partitioning clustering data; Constructing a local connection graph according to the first sample partitioning data and the sample partitioning clustering data to obtain sample partitioning graph data; Finely dividing the sample partition graph data according to the sample partition cluster data to obtain second sample partition data; Performing feature detection on the second sample division data to obtain sample division feature data; Performing weighted screening on the second sample partitioning data according to the sample partitioning characteristic data to obtain local area partitioning data; Performing local reference surface fitting on the local area division data to obtain local reference surface fitting data; Perform local plane error correction based on local reference surface fitting data to obtain reference surface smoothing correction data; The local reference plane fusion is performed according to the reference plane smoothing correction data to obtain the first reference plane data.

[0014] In the present invention, by extracting morphological features from sample scanning data, the slight morphological changes on the sample surface can be captured in detail, thereby providing rich detail information for datum reconstruction, and detailed processing helps to eliminate measurement errors caused by different subtle morphologies on the sample surface, and ensure the accuracy of the reconstruction results. By performing preliminary regional division of the scanning data and performing clustering calculations according to the morphological features of the sample, the properties of different regions on the sample surface can be automatically distinguished. The local morphological features of each region are clearly reflected, avoiding the errors caused by global assumptions on datum reconstruction, thereby improving the accuracy. The preliminary regional division and clustering algorithms based on morphological features can automatically identify the heterogeneity of the sample surface in shape and structure. Regardless of whether the sample surface is flat or there are complex curved surfaces, the system can flexibly divide it to ensure the applicability and flexibility of the datum reconstruction method. The system can finely adjust and fit the morphology of each local area to ensure that the entire sample surface can be accurately reconstructed. Through automated feature detection and weighted screening steps, manual intervention can be greatly reduced and work efficiency can be improved. The generation of sample division feature data and weighted screening data is the result of the execution of the automated system, which greatly reduces the impact of manual errors and saves time and labor costs. By constructing a local connection graph for sample partitioning data and clustering data, the data can be processed in a more structured manner, and at the same time, the deep data connection relationship of the data can be reflected, such as the similarity and distance correlation of adjacent data. Through local reference surface fitting and smoothing correction, the influence of local errors on the overall reference surface reconstruction can be further eliminated, ensuring the stability and accuracy of the reference surface data.

[0015] Preferably, the minimum convex hull reconstruction is specifically: Perform point cloud screening and simplification on sample scanning data to obtain sample screening data; Perform point cloud space optimization based on sample screening data to obtain point cloud optimization data; Construct an initial convex hull set based on point cloud optimization data to obtain preliminary convex hull data; Perform local adjustments on the preliminary convex hull data to obtain convex hull boundary optimization data; Performing surface reconstruction on the convex hull boundary optimization data to obtain convex hull optimization data; The preliminary convex hull data and the convex hull optimized data are fully and locally fused to obtain the second reference surface data.

[0016] In the present invention, by screening and streamlining the sample scanning data, noise data and irrelevant points can be eliminated, and the most representative and effective point data can be retained, which can effectively reduce the redundancy of point cloud data, reduce the computational burden, and improve the processing accuracy. Point cloud space optimization is to ensure the optimal distribution of point cloud data in space and eliminate deviation and redundancy through precise spatial distribution analysis and adjustment. By constructing preliminary convex hull data based on the optimized point cloud data, the basic shape framework of the sample can be quickly obtained in the early stage, laying the foundation for fine optimization. The local adjustment step can better adapt to the local changes on the sample surface by finely adjusting the convex hull boundary, especially when processing samples of complex shapes, it can ensure the optimization of local accuracy in the global range. By globally fusing the preliminary convex hull data with the optimized convex hull boundary, the global consistency of the data can be further improved and the deviation between local data can be eliminated. The convex hull data after global fusion is more accurate and stable, and can reflect the true geometric characteristics of the entire sample.

[0017] Preferably, step S2 specifically includes: Performing regional calibration according to the two-dimensional thickness distribution map data to obtain regional calibration data; Calculate the thickness variance of the regional calibration data to obtain regional variance data; Calculate thickness non-uniformity according to regional variance data to obtain thickness non-uniformity data; Perform critical area detection on thickness non-uniformity data to obtain critical area data; The thickness non-uniformity data and the critical area data are integrated to obtain the thickness uniformity data.

[0018] In the present invention, by performing regional calibration on the two-dimensional thickness distribution map data, the thickness characteristics of different regions can be identified, providing clear regional division for subsequent uniformity analysis. The calibration data can accurately reflect the thickness differences in different regions, ensuring the accuracy and reliability of the calculation. Calculating the variance of the thickness in the region can effectively quantify the discrete degree of thickness in each region. Variance is a standard indicator for describing data changes, which can reflect the thickness stability of the local area and help to quantify the uniformity of the entire sample. Calculating the thickness non-uniformity based on the regional variance data and further evaluating the thickness change trend of each area can effectively reveal the fluctuation of the surface thickness. By performing critical area detection on the thickness non-uniformity data, those areas with significant thickness changes can be automatically identified. The areas often correspond to potential problems in the production process, such as raw material quality problems, improper equipment calibration or irregular operation.

[0019] Preferably, step S3 specifically includes: Formulate sample screening rules based on thickness uniformity data to obtain sample screening rule data; Scoring the samples according to the sample screening rule data to obtain sample scoring data; Classifying and sorting the sample scoring data to obtain sample classification and sorting data; Monte Carlo simulation performance prediction is performed based on sample classification data to obtain sample performance distribution simulation data; Key samples are marked according to the sample performance distribution simulation data to obtain the tensile strength test sample selection data.

[0020] In the present invention, scientific and reasonable screening rules are formulated according to thickness uniformity data to ensure the systematicness and standardization of the screening process. The screening rules can clearly define which samples meet the requirements of the tensile strength test, avoid the interference of human subjective factors, and improve the reliability of the screening results. The formulation of the screening rules not only relies on experience, but also combines data analysis, and determines the key screening parameters through the quantitative results of the thickness uniformity data, ensuring that the screening process is more in line with the actual performance requirements of the samples. By scoring and sorting the samples, the samples can be prioritized according to certain standards. The sorting not only takes into account the thickness uniformity, but also combines other performance parameters to provide a basis for the selection of tensile strength tests, avoid unnecessary testing work, and improve the utilization efficiency of test resources. Monte Carlo simulation provides a statistical distribution of sample performance based on known thickness data and other factors affecting performance, which can predict the performance of different samples in actual tensile strength tests and help select samples that best represent overall performance. Marking key samples according to performance simulation results can ensure that test resources are used for the most representative samples with the most extreme performance, avoid a large number of unnecessary tests, and focus on testing the samples that best reflect the true performance of the samples, thereby improving test efficiency.

[0021] Preferably, step S4 is specifically: According to the tensile strength test sample selection data, the tensile strength test operation is carried out through the tensile strength test equipment to obtain stress time series data; Perform curve fitting on stress time series data to obtain stress-strain curve data; Calculate the extreme strength value according to the stress-strain curve data to obtain the extreme strength value data; The tensile strength is evaluated based on the stress-strain curve data and the extreme strength data to obtain the tensile strength test data.

[0022] In the present invention, the stress changes of the sample during the stretching process are accurately recorded by the tensile strength testing equipment, and the generated stress time series data provides detailed time series data support for the analysis. The stress time series data is converted into stress-strain curve data by curve fitting, and the fitting not only improves the smoothness and accuracy of the data, but also can effectively remove noise. By calculating the extreme strength of the stress-strain curve, key performance indicators such as the breaking point and yield point of the material under extreme stress can be obtained. By combining the extreme strength data with the stress-strain curve data for comprehensive analysis, not only can the single tensile strength value of the material be understood, but also the performance of the material under different stress states can be fully grasped, such as yield strength, tensile strength, elastic modulus and other key physical performance indicators. The tensile strength evaluation results provide data basis for problems in the production process. For example, if the test data shows that some batches of materials have abnormalities in the performance of the stress-strain curve, the production line can be adjusted immediately to optimize the material ratio and production process to ensure the consistency and quality of production.

[0023] Preferably, the present application also provides a wire insulation layer measurement sample system, which is used to perform the wire insulation layer measurement sample method as described above, and the wire insulation layer measurement sample system includes: A two-dimensional thickness distribution map generation module is used to scan and measure the thickness of the insulation layer of the numbered sample using a laser thickness gauge to generate two-dimensional thickness distribution map data; A thickness uniformity analysis module is used to perform thickness uniformity analysis based on two-dimensional thickness distribution map data to obtain thickness uniformity data; A tensile strength test sample selection module is used to select tensile strength test samples according to thickness uniformity data to obtain tensile strength test sample selection data; A tensile strength test module is used to perform a tensile strength test according to the tensile strength test sample selection data to obtain tensile strength test data; The wire insulation layer performance evaluation module is used to evaluate the wire insulation layer performance according to the two-dimensional thickness distribution map data, thickness uniformity data and tensile strength test data, and obtain the wire insulation layer performance evaluation data to perform auxiliary work for wire insulation layer sample measurement.

[0024] The beneficial effect of the present invention is that the thickness of the insulating layer of the sample is scanned by a laser thickness gauge to generate high-precision two-dimensional thickness distribution map data. This non-contact detection technology based on laser measurement can accurately capture the thickness change of the insulating layer, avoid the error caused by traditional contact measurement, and is not affected by surface microstructure or foreign matter, ensuring the accuracy of the data. By performing thickness uniformity analysis on the two-dimensional thickness distribution map data, the thickness uniformity of the wire insulation layer can be detected in real time to find potential weak links. In particular, for wire products in large-scale production, unqualified products can be efficiently screened out, and the production process can be adjusted in time to ensure the quality of each wire. By screening samples for tensile strength testing based on thickness uniformity data, and then combining thickness distribution and strength data for comprehensive performance evaluation, multi-dimensional performance evaluation can fully reflect the quality of the wire insulation layer, especially in actual use. The tensile strength test can provide the performance of the material under stress conditions, including its rupture strength, ductility and other characteristics under various extreme conditions, thereby providing data support for reliability in applications. Through data-based analysis and evaluation, quality problems that occur during the production process, such as uneven thickness and insufficient tensile strength, can be identified in real time, and production process parameters can be adjusted in a timely manner, such as changing coating thickness, changing material selection, adjusting production speed, etc., to ensure the quality stability of each batch of products from the source. Combining the tensile strength test sample selection data and thickness uniformity data, representative samples can be intelligently screened for in-depth testing, thereby reducing unnecessary testing expenses and time consumption. In the large-scale production process, it can not only reduce the errors caused by manual operation, but also greatly shorten the test cycle and improve the overall operating efficiency of the production line. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting implementations made with reference to the following drawings: Figure 1 A flow chart showing the steps of a method for measuring a wire insulation layer sample according to an embodiment is shown; Figure 2 A flowchart showing a method for generating a two-dimensional thickness distribution map according to an embodiment is shown; Figure 3 A flowchart showing a thickness uniformity analysis method according to an embodiment is shown; Figure 4 A flowchart showing a method for selecting a tensile strength test sample according to an embodiment of the present invention is provided; Figure 5 A flowchart of a tensile strength testing method according to an embodiment is shown. DETAILED DESCRIPTION

[0026] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0027] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0028] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0029] See also Figures 1 to 5 The present application provides a method for measuring a wire insulation layer sample, comprising the following steps: Step S1: Scan and measure the thickness of the insulation layer of the numbered sample using a laser thickness gauge to generate two-dimensional thickness distribution map data; Specifically, a laser thickness gauge is used, which irradiates the surface of the insulating layer with a laser beam, and the reflected laser signal is received by the receiver. The distance from the surface to the instrument is calculated based on the time delay of the laser beam (time flight principle) or the change in the intensity of the reflected light. The sample to be measured is placed within the scanning range of the laser thickness gauge. The laser thickness gauge scans along the surface of the sample, automatically records the thickness information of each point, and generates a two-dimensional thickness distribution map. During scanning, the resolution of the instrument is set to a certain value to ensure that the thickness measurement accuracy of each point meets the requirements. Through the calculation function of the laser thickness gauge, the thickness information of each point is generated into a two-dimensional thickness distribution map data, and the data is in the form of a matrix, and each element of the matrix represents the thickness of the corresponding point.

[0030] Step S2: performing thickness uniformity analysis according to the two-dimensional thickness distribution map data to obtain thickness uniformity data; Specifically, the obtained two-dimensional thickness distribution map data is processed. First, the noise is removed and the data is smoothed to ensure the accuracy of the thickness measurement value. The thickness data is smoothed by a moving average filter (such as a 3x3 or 5x5 window). The thickness uniformity of the entire sample is calculated, and the thickness variation is quantified by statistical methods, including calculating the standard deviation of the thickness, the coefficient of variation, etc. If the standard deviation is small, it means that the thickness distribution is relatively uniform; if the standard deviation is large, it means that there is a situation of uneven thickness. At this point, the location and size of the uneven area are further analyzed.

[0031] Step S3: Selecting tensile strength test samples according to the thickness uniformity data to obtain tensile strength test sample selection data; Specifically, based on the thickness uniformity data obtained in step S2, areas with poor thickness uniformity are selected as candidate areas for tensile strength testing. In general, areas with large thickness standard deviations are selected for subsequent testing. Based on the results of the uniformity analysis, the size and location of the test area are determined. If there are areas where the thickness is significantly lower or higher, these areas are selected as test samples. When selecting test samples, a certain number of test points must be determined based on the total area of ​​the sample and the test requirements. Select k test points to ensure that the test points cover different areas of the entire sample.

[0032] Step S4: performing a tensile strength test according to the tensile strength test sample selection data to obtain tensile strength test data; Specifically, from the test sample selected in step S3, a sample with a specific size is cut out to ensure that the boundary of the sample meets the requirements of the tensile test. The tensile strength test is performed using a standard tensile testing instrument. The tensile testing instrument generally includes a loading device, and the sample is fixed at both ends, and the tensile strength of the sample is obtained by the stress-strain curve generated by stretching. The stress-strain curve (applied tensile force divided by the cross-sectional area) is recorded during the test. The tensile strength of the sample is obtained by measuring the maximum tensile force of the sample. The tensile strength value obtained by the test is recorded, and this value can be compared with the thickness data to analyze the influence of thickness on the performance of the wire insulation layer.

[0033] Step S5: Conducting wire insulation layer performance evaluation based on the two-dimensional thickness distribution map data, thickness uniformity data and tensile strength test data to obtain wire insulation layer performance evaluation data for conducting wire insulation layer sample measurement auxiliary operations.

[0034] Specifically, performance evaluation is performed based on two-dimensional thickness distribution data, thickness uniformity data, and tensile strength test data. Based on various performance indicators of the insulation layer (such as thickness uniformity, tensile strength, etc.), performance scores are obtained through weighted average method, regression analysis or other data fusion methods. Based on the performance evaluation data, the evaluation results of the insulation layer are output and decision support is provided for production and quality control.

[0035] Preferably, step S1 specifically includes: Step S11: spatially positioning and calibrating the optical axis of the laser thickness gauge to generate laser positioning data; Specifically, the laser thickness gauge is spatially positioned by setting up its positioning system. The commonly used positioning method is to use three-dimensional coordinate measurement technology to ensure that the laser thickness gauge can accurately track the spatial position of the sample during the scanning process. The coordinate system of the laser thickness gauge is calibrated to a specific reference point on the sample surface to ensure that the data collected during the scanning process conforms to the actual three-dimensional geometry of the sample surface. In order to ensure that the laser beam is perpendicular to the sample surface and avoid measurement errors caused by the tilt of the optical axis, the optical axis of the laser thickness gauge needs to be calibrated. By calibrating the target plate and the reflector, it is detected whether the laser beam is completely vertical, and the position of the laser is fine-tuned to ensure the consistency of the emission angle and the receiving angle of the laser beam. The error correction of the optical axis is performed using precision optical equipment. According to the spatial positioning and optical axis calibration, a series of reference data based on the initial position of the laser is generated, which is called laser positioning data. This data records all laser emission and receiving positions in the measurement area.

[0036] Step S12: reconstructing the sample reference surface through a laser thickness gauge according to the laser positioning data to obtain reference surface data; Specifically, during the initial scanning stage of the sample, the laser thickness gauge selects the flattest area as the reference surface. The reference surface is determined by calculating the average height value of multiple measurement points. Based on the laser positioning data, a reference reference surface is reconstructed by collecting thickness data from multiple measurement points. The goal is to obtain a standardized plane so that the thickness measurement data can be offset corrected relative to the reference surface to eliminate the error caused by the uneven surface of the sample. At the same time, the least squares method is used to fit the reference surface.

[0037] Step S13: performing multi-point thickness sampling using a laser thickness gauge according to the reference surface data to obtain preliminary thickness sampling matrix data; Specifically, on the reference surface, the laser thickness gauge will scan along different paths. In order to obtain global information on the sample thickness, multiple evenly distributed points are selected on the sample surface for thickness sampling. The laser measures each sampling point multiple times to ensure the reliability of the data. According to the position distribution on the sample surface, a series of thickness data are obtained. These data are organized into a matrix, and each element of the matrix represents the thickness value at the corresponding position.

[0038] Step S14: performing thickness distribution interpolation calculation according to the thickness preliminary sampling matrix data to obtain thickness distribution interpolation data; Specifically, in order to achieve a smoother thickness distribution graph, an interpolation algorithm is used to interpolate the data between the sampling points. The interpolation method is linear interpolation, quadratic interpolation or high-order interpolation, and the specific selection depends on the smoothness of the sample surface and the required interpolation accuracy.

[0039] Specifically, smoothness feature extraction and variation feature extraction are performed based on the thickness preliminary sampling matrix data to obtain preliminary sampling smoothness feature data and preliminary sampling variation feature data respectively; weighted calculation is performed based on the preliminary sampling smoothness feature data and preliminary sampling variation feature data to obtain preliminary sampling irregularity degree data (such as ),in To preliminarily sample irregularity data, is the weight coefficient of the initial sampling smoothness feature data, To preliminarily sample the smoothness feature data, is the weight coefficient of the initial sampling variation feature data, is the preliminary sampling variation characteristic data; when it is determined that the preliminary sampling irregularity data is greater than the preset first irregularity threshold data, the thickness preliminary sampling matrix data is linearly interpolated to obtain thickness distribution interpolation data; when it is determined that the preliminary sampling irregularity data is less than or equal to the preset first irregularity threshold data and greater than the preset second irregularity threshold data, the thickness preliminary sampling matrix data is linearly interpolated to obtain thickness distribution interpolation data; when it is determined that the preliminary sampling irregularity data is less than or equal to the preset second irregularity threshold data, the thickness preliminary sampling matrix data is linearly interpolated to obtain thickness distribution interpolation data, wherein the preset first irregularity threshold data is greater than the preset second irregularity threshold data.

[0040] Step S15: Generate a dynamic distribution map based on the thickness distribution interpolation data to obtain two-dimensional thickness distribution map data.

[0041] Specifically, based on the interpolated data, image processing technology is used to generate a two-dimensional thickness distribution map. First, the thickness is represented by converting the interpolated data into grayscale values ​​or color values. Thicker areas can be represented by dark colors, thinner areas can be represented by light colors, or the color mapping relationship can be adjusted according to actual needs. In order to more intuitively display the thickness distribution, parameters such as color scale and contrast can be dynamically adjusted when generating the map, so that the thickness changes in the image are more obvious. By setting a suitable threshold, certain areas with large thickness changes are highlighted.

[0042] Preferably, the spatial positioning and optical axis calibration are specifically: By using the laser thickness gauge’s integrated inertial sensors and external vision sensors, the spatial position data of the equipment is captured in real time; Specifically, the laser thickness gauge integrates an inertial measurement unit (IMU), which includes sensors such as accelerometers, gyroscopes, and magnetometers. Accelerometers can measure the linear acceleration of the device, gyroscopes can measure the angular velocity of the device, and magnetometers can be used to calibrate the direction. By fusing the data of these sensors, the spatial position data of the device is captured in real time, including the displacement (X, Y, Z) and rotation angle (angle around the X, Y, Z axis) of the device. The external visual sensor is a camera or a laser radar (LiDAR), which is used to obtain the position of the device in three-dimensional space through image recognition or laser scanning. The position coordinates of the device relative to the reference point are calculated by the position of the calibration point or known object (such as a reference point or ruler) in the visual image. These visual data are combined with inertial data to obtain the precise spatial position of the device. Combined with the data of the inertial sensor and the external visual sensor, the position and attitude information of the device are updated in real time. Through data fusion technology, the spatial position data of the device is obtained in real time to ensure high accuracy and reliability of measurement.

[0043] Perform posture calculation based on spatial position data to obtain device posture data; Specifically, attitude refers to the angle and direction of the device in space, expressed in Euler angles (roll, pitch, yaw) or quaternions. Based on the linear acceleration and angular velocity data collected by the inertial sensor (accelerometer and gyroscope), the angle change of the device is obtained by integration, and then the attitude of the device is calculated to obtain the attitude data of the device, that is, the rotation angle of the device relative to the initial position, including three rotation angles (pitch, roll, yaw).

[0044] Perform posture error analysis based on the equipment posture data to obtain posture error data; Specifically, there are sources of error in attitude calculation, such as sensor noise, sensor accuracy limitations, integration errors, etc. Error analysis requires determining the deviation of the device attitude by comparing the data of the inertial sensor with the data of external sensors (such as visual sensors or reference points). The attitude error is calculated by comparing the difference between the calculated attitude data and the actual position of the external visual sensor or known reference point. Error analysis can be done in the following ways: compare the angle and position of the device relative to the reference point provided by the external visual sensor with the attitude calculated by the inertial sensor to obtain the error. Use techniques such as Kalman filtering to combine the data of the inertial sensor and the visual sensor to minimize the error. The error data of the device attitude is obtained, which contains the deviation of the device attitude estimate at each measurement time, such as the error of the pitch angle, roll angle and yaw angle.

[0045] Perform attitude calibration according to attitude error data to obtain attitude calibration data; Specifically, according to the calculated attitude error data, the attitude of the device is corrected through reverse compensation or calibration algorithm. The attitude error is corrected by using proportional integral differential (PID) control algorithm, least squares method, Kalman filter and other methods. If the attitude error is calculated, the attitude of the device can be appropriately adjusted by comparing the error data. The calibrated attitude data will more accurately reflect the true attitude of the device and reduce the error caused by sensor error or external interference.

[0046] Perform laser optical axis alignment on the posture calibration data to obtain optical axis alignment data; Specifically, laser optical axis alignment is to ensure that the laser beam of the laser measuring instrument is in the same direction as the physical structure of the device (such as the laser emitter). The laser optical axis is adjusted by comparing the actual posture of the device with the actual direction of the laser emission (which can be measured by the angle of the reflection point or the target point). The angular deviation between the laser emission axis and the device installation axis is calculated through the posture calibration data, and then the alignment adjustment is performed. The laser optical axis is kept consistent with the actual posture of the device by rotating the laser or adjusting its emission angle. The calibration data of the laser optical axis is calculated to ensure that the direction of the laser emission matches the actual movement direction of the device and minimize the error.

[0047] Perform error correction according to the optical axis alignment data to obtain error correction data; Specifically, after the laser optical axis is aligned, there are still errors caused by changes in the device posture, interference from external factors, etc. Further error correction is required to ensure that the laser thickness gauge maintains measurement accuracy in different postures. According to the optical axis alignment data, the accuracy of the laser beam is optimized by correcting the deviation, and the beam error caused by changes in the device posture is eliminated by correcting and compensating the laser measurement results. The error correction data obtained represents the error correction value of the laser measuring instrument in different device postures after calibration. These data are used for subsequent laser positioning.

[0048] Laser positioning data is generated according to the spatial position data, attitude calibration data, optical axis alignment data and error correction data to obtain laser positioning data.

[0049] Specifically, the actual position, attitude, and optical axis direction of the laser are comprehensively calculated with the measurement results by combining the spatial position data, attitude calibration data, optical axis alignment data, and error correction data through a mathematical model (such as weighting the spatial position data and attitude calibration data based on the error correction data to obtain the corrected data, and combining the corrected data with the optical axis alignment data) to generate laser positioning data. This ensures that the laser measurement results are highly accurate and can accurately reflect the actual spatial position of the device.

[0050] Preferably, the sample reference plane reconstruction is specifically: Perform a preliminary scan of the sample using a laser thickness gauge according to the laser positioning data to obtain sample scanning data; Specifically, the laser thickness gauge performs a preliminary scan along the sample surface based on the previous laser positioning data. The laser beam is used to measure multiple points on the sample surface to obtain the coordinates and thickness data of each measurement point. The position of each scanning point is defined by the three-dimensional coordinate system of the laser. The laser dynamically adjusts the scanning path according to the laser positioning data to ensure that all important areas of the sample are covered. Through scanning, a set of three-dimensional point data on the sample surface is obtained, and the scanned data is preprocessed, including noise filtering and missing data supplementation, to ensure the accuracy and completeness of the scanned data. A smoothing filter algorithm is used to remove high-frequency noise, or interpolation methods are used to fill in missing scanned data.

[0051] Reconstructing a local reference plane according to the sample scanning data to obtain first reference plane data; Specifically, the local reference plane refers to the reference plane of each local area fitted by selecting a small area on the sample surface. For each local area, the least squares method is used to perform plane fitting on the scan data to obtain the reference plane of the area. The selection of local areas is achieved by a meshing method or a point-based clustering method (such as DBSCAN). After performing plane fitting on each local area, a local reference plane data set is obtained. All local reference planes are merged to form multiple reference plane fragments of the sample to obtain the first reference plane data.

[0052] Perform minimum convex hull reconstruction according to the sample scanning data to obtain the second reference surface data; Specifically, the minimum convex hull is the smallest convex polyhedron that surrounds the sample scan data point set, which can better express the shape of the sample surface. By calculating the convex hull of all scanned data points, a convex surface surrounding these points is obtained, providing a global reference datum surface covering the entire sample surface. The minimum convex hull of these points is calculated using the convex hull algorithm. The specific implementation method is to find a set of points so that all scanned points are located in the convex hull formed by these points. The calculation method can be calculated by calculating the boundary surface of the point set, that is, finding a set of planes that completely surround all scanned points. After calculating the minimum convex hull, a polyhedron containing multiple facets is obtained. By converting these facets into a smooth datum surface, the second datum surface data is obtained. Use a triangular mesh or a plane fitting algorithm to smooth the facets of the convex hull to form a continuous datum surface suitable for further analysis.

[0053] Performing shape feature fusion on the first reference plane data and the second reference plane data to obtain reference plane data; Specifically, in order to obtain a more accurate reference surface, it is necessary to combine the local reference surface with the global reference surface. The local reference surface (obtained by least squares fitting) accurately describes the shape characteristics of a small range, while the global reference surface (obtained by minimum convex hull reconstruction) can describe the shape of the entire sample. By combining the advantages of both, a more accurate and robust reference surface data can be obtained. Weighted fusion is performed according to the locality of the data. For example, in the flat area of ​​the sample, more reliance can be placed on the local reference surface, while in the area with large curvature changes, more reliance can be placed on the global reference surface. Dynamic adjustment is performed according to the characteristics of the sample surface. For example, different weights can be selected according to the curvature of the local area: areas with smaller curvature (i.e., flat areas) are given more local reference surface weights, and areas with larger curvature (i.e., curved areas) are given more global reference surface weights. After fusion, an accurate reference surface data combining global and local features is obtained.

[0054] The preliminary scan of the sample is as follows: Scan the sample surface with a laser thickness gauge to obtain point cloud data of the sample surface; Specifically, the laser thickness gauge scans the sample surface multiple times with a laser beam, and each reflection of the laser beam generates a measurement point. Each measurement point contains three-dimensional coordinates and a thickness value. During scanning, the device scans the sample surface sequentially according to a preset path, generating a point cloud data set containing thousands to millions of points.

[0055] Perform particle iterative filtering on the sample surface point cloud data to obtain scanning noise reduction data; Specifically, in the process of point cloud denoising, each point is regarded as a particle, and the weights of these particles are iteratively optimized to remove unstable or noise points. During initialization, a particle is assigned to each point, the position is each coordinate in the point cloud, and the initial weight is set to be equal. According to the smoothness assumption of the sample surface shape, the weight of the particle is iteratively optimized. That is, the similarity of each particle is evaluated, and the particle with a larger weight is more consistent with the real surface. The weight of the particle can be determined by calculating the deviation between the neighborhood point and the current point. Resampling is performed, particles with higher weights are retained, particles with lower weights are removed, and the estimated value of the particle is iteratively updated. Continue this process until the noise of the point cloud is effectively removed and the denoised point cloud data is obtained.

[0056] Perform KD tree region division according to the scanned denoising data to obtain scanned division data; Specifically, KD tree (K-Dimensional Tree) is a spatial segmentation data structure used to quickly find neighboring points in a point cloud. By constructing a KD tree, efficient region partitioning and nearest neighbor search can be performed on the point cloud. Specifically, the KD tree divides the sample surface point cloud data into several small regions, and the boundaries of these small regions are adaptively divided according to the distribution of the point cloud. For the denoised point cloud data set, select a suitable dimension (such as x, y, or z) for partitioning. The dimension with the largest variance is initially selected for partitioning so that each partition can effectively partition the point cloud data. Based on the selected dimension, the point cloud is divided into two parts according to the median of the dimension, and a subtree is recursively constructed for each subset until a certain stopping criterion is reached (for example, the number of points in each leaf node is less than a certain threshold). The constructed KD tree can be used to efficiently query the neighborhood of point cloud data, as well as perform region partitioning and nearest neighbor query. Through the KD tree, the point cloud is divided into multiple small regions, and the points in each region are relatively close in spatial position. These regions can be represented by the nodes of the KD tree.

[0057] voxelize and resample the scanned divided data to obtain sample resampled data; Specifically, voxelization is the process of converting continuous 3D point cloud data into discrete voxel grids, where each voxel represents a small area in space. The voxelization process can be achieved by defining a voxel size in the point cloud data, with each voxel containing several points. Voxelization helps reduce the amount of data and improve processing efficiency. Select the size of the voxel and determine the size of each voxel in the x, y, and z directions. For each area in the scanned partitioned data, its space is divided into several voxel grids of size V. The points in each voxel are marked as the representative point of the voxel (which can be the average position or centroid of all points in the voxel). By resampling the voxelized data, the points in each voxel are gathered into a representative point, reducing the number of points and ensuring that the representative point can effectively express the point cloud features in the voxel. The resampled point of each voxel is the average position or geometric center of the points in the voxel.

[0058] The spatial coordinates are aligned according to the sample scanning data and the sample resampling data to obtain the sample scanning data.

[0059] Specifically, based on the sample resampled data, it is further aligned with the scanned denoised data. To ensure that data from different sources can be compared and further analyzed in a unified coordinate system, the least squares method is used to achieve rigid registration of the two sets of data. A reference point set (for example, points in the sample resampled data) is selected as the alignment target. A rigid transformation (including translation and rotation) is used to adjust the scanned denoised data so that it is aligned with the sample resampled data as much as possible to obtain the sample scanned data.

[0060] Preferably, the local reference plane reconstruction is specifically: Extracting sample morphological features from the sample scanning data to obtain sample morphological feature data; Specifically, morphological features are extracted from the sample scanning data, mainly including the curvature, normal vector, surface smoothness, etc. of the sample surface. The geometric properties of the point cloud data are used to describe the local morphological features of the sample surface. These features help to identify flat areas, curved areas, and other defective areas on the surface. The curvature value is calculated through the neighborhood of each point. The normal vector of each point is estimated by the covariance matrix of the neighborhood points. The neighborhood points are fitted by the least squares method to obtain the normal vector. The obtained morphological feature data includes the curvature and normal vector information of each point, as well as other geometric features, which can help identify the characteristics of the sample surface area.

[0061] Performing preliminary area division on the sample scanning data according to the sample morphological characteristic data to obtain first sample division data; Specifically, the sample surface is preliminarily divided into regions according to the morphological feature data. A common method is to divide the regions by curvature threshold or normal vector difference. Regions with smaller curvature can be regarded as flat regions, while regions with larger curvature can be regarded as curved regions. Regions with larger normal vector differences indicate that there are protrusions or depressions on the surface. Each point in the scan data is classified according to its curvature and normal vector information and divided into different regions. The curvature threshold and the normal vector difference threshold can be set for division. If both thresholds are met, they are regarded as the same region.

[0062] Performing morphological clustering calculation according to the first sample partitioning data to obtain sample partitioning clustering data; Specifically, clustering calculation is performed on the first sample partitioning data to further refine the regional division. Common clustering methods include density-based clustering algorithms (such as DBSCAN) or K-means clustering. Clustering is performed according to the distribution density and morphological characteristics of the points in the region. For each region, the density of the points in the region is calculated, and the Euclidean distance metric can be used to calculate the distance between the points. If the distance between two points is less than a preset threshold, they are considered to belong to the same cluster. The clustering results are evaluated to ensure that each cluster represents a connected area. Through clustering, new partitioning data is obtained, and each cluster represents an area with similar morphological characteristics.

[0063] Constructing a local connection graph according to the first sample partitioning data and the sample partitioning clustering data to obtain sample partitioning graph data; Specifically, the data is clustered according to the sample division, and a local connection graph is constructed. The nodes of the graph represent regions or points in the sample, and the edges represent the connection relationship between different regions. For each pair of regions or clusters, the distance or morphological similarity between them is calculated. If the distance between two regions is less than a certain threshold, they are considered to be spatially connected. The adjacency matrix of the graph is used to represent these connection relationships, and the weight of the edge can be weighted according to the distance or morphological similarity between regions. A graph containing several nodes and edges is obtained, where the nodes represent regions or clusters and the edges represent connection relationships.

[0064] Finely dividing the sample partition graph data according to the sample partition cluster data to obtain second sample partition data; Specifically, based on the local connection graph, the graph is further finely divided, which is achieved through a graph partitioning algorithm (such as spectral clustering or graph cut), and the graph is divided into several smaller subgraphs according to the connection strength of the graph. The weight of each edge is calculated, and the weight of the edge represents the strength of the connection or the similarity of the region. The graph is divided into several subgraphs using a graph partitioning algorithm, each of which represents a smaller area in the sample. The finely divided data contains multiple small areas, and the morphological characteristics of each area are more consistent and more clearly distinguished from other areas.

[0065] Performing feature detection on the second sample division data to obtain sample division feature data; Specifically, feature detection is performed on each region in the second sample division data, mainly detecting the flatness, curvature, and normal vector differences of the region. The geometric features such as the average curvature and the maximum curvature of each region are calculated. The normal vector change of each region is calculated to analyze the smoothness of the regional surface. According to the extracted features, the region is classified into a flat region, a curved region, etc. Through feature detection, the morphological feature data of each region is obtained, including curvature, normal vector change, flatness, etc.

[0066] Specifically, more importantly, local area feature extraction is performed on the second sample division data to obtain local area feature data; pyramid convolution is performed on the local area feature data to obtain local area multi-scale feature data; high-dimensional feature mapping is performed on the local area multi-scale feature data to obtain preliminary sample division feature data; local fuzzy parameter calculation is performed on the local area feature data to obtain local fuzzy data; Gaussian blur is performed on the local area feature data based on the local fuzzy data to obtain regional Gaussian blur data; local merging is performed on the regional Gaussian blur data and the preliminary sample division feature data to obtain local merged data; local feature descriptor generation is performed on the local merged data to obtain sample division feature data.

[0067] According to the second sample division data, the sample is divided into local regions and features are extracted. Each local region represents a specific detail area on the sample surface, which has different geometric shapes and texture features. According to the point cloud data of each local region, the local surface features are extracted based on the data collected by the scanning device, including surface normal, curvature, texture and other information. After obtaining the local region feature data, it is further processed by the pyramid convolution method. Pyramid convolution uses convolution kernels of different scales to perform convolution operations on the local regions of the sample at different resolutions to obtain multi-scale feature representations. It can effectively capture the details of the local region while maintaining the overall features of the larger scale region. The generated local region multi-scale feature data expresses the local region features at multiple levels at different scales. After obtaining the local region multi-scale feature data, these multi-scale features are converted into higher-dimensional spatial representations through high-dimensional feature mapping. Using high-dimensional mapping technology, the original multi-scale feature vector is projected into a higher-dimensional space to better capture the potential patterns and structures hidden in the local area, such as based on polynomial mapping or Fourier transform mapping, to map low-dimensional data to high-dimensional space, and convert local area feature data into data representation in high-dimensional space through mapping functions. After high-dimensional feature mapping, the local area feature data is fuzzified to enhance its smoothness. The fuzzy parameters of each local area are calculated. These parameters are determined according to the feature distribution in the local area, such as the uniformity of the texture, the smoothness of the edge, etc., and are dynamically adjusted according to the changes in the sample surface to generate local fuzzy data. After obtaining the local fuzzy data, these data are used to perform Gaussian blur processing on the local area features. Gaussian blur is a smoothing operation that can effectively remove noise in the data while maintaining the main feature information. Based on the fuzzy parameters obtained above, compared with conventional Gaussian blur, polynomial mapping or Gaussian radial basis function (RBF) mapping can capture nonlinear relationships in the data in high-dimensional space. For complex and implicit structures in the data (such as nonlinear geometric shapes and textures between samples), high-dimensional mapping can better express them. By applying Gaussian blur to the local area feature data, the local details in the image can be further smoothed and the interference of high-frequency noise can be reduced. The obtained regional Gaussian blur data will more accurately reflect the smooth characteristics and main structure of the local area. After obtaining the regional Gaussian blur data and the preliminary sample segmentation feature data, the local area is merged. The local merging process integrates the local area feature data with the Gaussian blur data. Through weighted merging or fusion strategies, multiple information is aggregated into a feature description. The merged data represents the fine structure and smooth characteristics of the local area, generating local merged data. Through the generation of local feature descriptors, the key features of each sample are extracted, and the local geometry, texture and other important features of the sample are accurately represented to obtain the sample segmentation feature data.

[0068] Performing weighted screening on the second sample partitioning data according to the sample partitioning characteristic data to obtain local area partitioning data; Specifically, a weight is assigned to each region based on the extracted feature data. Representative regions are selected through weighted screening, and those that are abnormal or do not meet the standards are removed. Weights are set for each region based on feature data (such as curvature and flatness). Regions with lower weights are eliminated, and regions with higher weights are retained. Through weighted screening, local region division data is obtained, and these regions represent different local features of the sample surface.

[0069] Performing local reference surface fitting on the local area division data to obtain local reference surface fitting data; Specifically, the reference surface is fitted for the data of each local area, and the least square method is used to fit the plane or quadratic surface. The plane or surface is fitted for the point set of each local area, assuming that the surface of the local area can be approximated by a plane or a surface. The least square method is used to optimize the fitting result and minimize the fitting error. The reference surface obtained by fitting each local area is the local reference surface of the area, and the result obtained is the local reference surface data.

[0070] Perform local plane error correction based on local reference surface fitting data to obtain reference surface smoothing correction data; Specifically, after the local reference surface is fitted, local plane errors may occur due to measurement errors, noise or equipment instability. These errors cause local protrusions or depressions on the reference surface, affecting the overall flatness. In order to solve this problem, the reference surface is smoothed and corrected. Data smoothing technology is used to eliminate these local errors. The fitted reference surface is locally smoothed using weighted least squares or moving average methods. Alternatively, local plane error correction is performed using polynomial interpolation. For example, quadratic or cubic polynomials are used to fit the error data of the local area to smooth and correct the surface.

[0071] The local reference plane fusion is performed according to the reference plane smoothing correction data to obtain the first reference plane data.

[0072] Specifically, after the error correction is completed in each local area, the adjacent reference surface areas are fused by weighted averaging or interpolation methods. After all local areas have been corrected, these local reference surfaces are fused through a global model. The best global reference surface is found through a global optimization algorithm (such as the least squares method). This process takes into account the smooth transition between each local reference surface and minimizes the seam error between regions.

[0073] Preferably, the minimum convex hull reconstruction is specifically: Perform point cloud screening and simplification on sample scanning data to obtain sample screening data; Specifically, point cloud data contains a large number of redundant or irrelevant points, so it is necessary to improve processing efficiency through screening and simplification. The screening method is based on distance thresholds, density or geometric features (such as curvature, normal vector, etc.). Use distance-based screening methods to remove outliers. For example, if the distance from a point to its neighboring points is greater than the set threshold, the point is considered to be a noise point. Alternatively, calculate the neighborhood density of each point in the point cloud, and remove low-density areas by setting a density threshold to ensure that representative points on the sample surface are retained. Alternatively, by calculating the curvature of each point, points with excessive curvature (such as sharp or curved parts) can be retained, and points in flat areas can be removed to reduce computational complexity. After screening, sample screening data is obtained.

[0074] Perform point cloud space optimization based on sample screening data to obtain point cloud optimization data; Specifically, point cloud optimization aims to make point cloud data more representative of the sample surface by removing redundant points and adjusting the distribution of points. Common optimization methods include grid-based homogenization, coordinate deviation correction, etc. Using a gridding method, set a suitable grid size and divide the point cloud into several small units (grids). Select representative points in each unit, and select the center of mass or the nearest point as the representative. Adjust the deviation points in the point cloud, and use the least squares method to globally translate and rotate the point cloud so that the point cloud as a whole is closer to the actual shape of the sample. After the initial optimization, perform several iterations, each time adjusting the position and distribution of the points to reduce excessive deviations or uneven density between points in the point cloud, thereby obtaining point cloud optimized data.

[0075] Construct an initial convex hull set based on point cloud optimization data to obtain preliminary convex hull data; Specifically, the convex hull is the smallest convex polyhedron that encloses all points in the point cloud. By constructing the convex hull of the point cloud data, the outer boundary of the point cloud is obtained. The initial convex hull is constructed using the Wiener enclosing method, Quickhull algorithm, etc. First, a set of points is selected as the preliminary boundary, by calculating the boundary area of ​​the point cloud and selecting points on the boundary. Based on the preliminary boundary points, the minimum convex hull of these points is calculated using a convex hull algorithm (such as Quickhull). The convex hull shape of the point cloud is determined by solving the boundary points of the plane or polyhedron. The volume and number of faces of the preliminary convex hull are calculated. If the preliminary results do not meet expectations, the boundary points are re-screened and the convex hull is recalculated. The preliminary convex hull data is obtained, which contains the boundary points and faces of the convex hull, expressed as a set of convex hull shapes.

[0076] Perform local adjustments on the preliminary convex hull data to obtain convex hull boundary optimization data; Specifically, the preliminary convex hull often cannot perfectly fit the sample surface, and there are overfitting or non-smooth boundaries. Local adjustment aims to correct the boundary of the convex hull according to the actual morphology of the sample to make it more consistent with the surface features. The convex hull boundary is smoothed by weighted least squares method so that the boundary is no longer too sharp or irregular, and the smoothness of the boundary is maintained. Each face of the convex hull is refined, and the position of the convex hull face is adjusted using the morphological features of the point cloud (such as normal vector and curvature) to make the convex hull fit the actual surface better. For the corners and irregular parts of the convex hull, a local optimization method is used to adjust the position of the boundary points to reduce overfitting. Through the above adjustment process, the optimized convex hull boundary data is obtained, which includes the optimized and adjusted boundaries and faces.

[0077] Performing surface reconstruction on the convex hull boundary optimization data to obtain convex hull optimization data; Specifically, based on the optimized convex hull boundary, the surface reconstruction is performed to obtain a more accurate sample surface model. Use NURBS (non-uniform rational B-spline) or other surface fitting methods to generate a smooth surface. First, the surface mesh is constructed using a polygonal mesh algorithm through the optimized convex hull boundary. This mesh represents the general shape of the sample surface. The fitting method such as the least squares method is used to calculate the fitted plane or surface based on the point cloud data and the boundary mesh. The smoothness and accuracy of the surface are optimized by adjusting the weight and fitting error. The fitted surface is finely corrected to ensure that the surface can fit the sample surface well in each local area. The obtained surface is the optimized convex hull data, which represents the shape of the sample and fits the actual surface more accurately.

[0078] The preliminary convex hull data and the convex hull optimized data are fully and locally fused to obtain the second reference surface data.

[0079] Specifically, the preliminary convex hull and the optimized convex hull data are fused to ensure that the reference surface can adapt to the overall shape and local details at the same time in the global scope. Between the preliminary convex hull and the optimized data, weighted averaging, surface smoothing and other methods are applied so that the final fused surface can smoothly transition and accurately match the point cloud data. By adjusting the weights in the fusion process, the local details of the sample surface are enhanced, so that the reference surface is globally consistent, while retaining details in the local area. The fused data is the second reference surface data, which combines the data of the preliminary convex hull and the optimized convex hull to form a complete and smooth reference surface.

[0080] Preferably, step S2 specifically includes: Step S21: performing regional calibration according to the two-dimensional thickness distribution map data to obtain regional calibration data; Specifically, the purpose of regional calibration is to divide the two-dimensional thickness distribution map data into several regions, each region represents a different regional feature of the sample surface, which is convenient for analysis and calculation. Regional calibration is implemented based on threshold segmentation, region growing or clustering algorithm. Based on the range of thickness values, the points in the two-dimensional thickness distribution map data are divided into several regions according to different thickness values. For example, set an upper and lower limit of thickness, and all points whose thickness values ​​fall within this interval are classified into the same region. Alternatively, if the thickness distribution map is more complex, a density-based clustering method is used to divide points with similar thickness and adjacent positions into one region. When clustering, the distance between points and the thickness difference can be used as the basis for calculation. Alternatively, the boundaries of each region are determined by continuity checks, such as using edge detection methods to ensure that the boundaries of each region are clear.

[0081] Step S22: Calculate the thickness variance of the regional calibration data to obtain regional variance data; Specifically, for each region, the variance of its thickness data is calculated. The variance can reflect the degree of thickness dispersion in the region. The larger the variance, the more drastic the thickness variation in the region; the smaller the variance, the more uniform the thickness in the region. For each region, the variance is calculated. The variance of each region indicates the degree of thickness fluctuation in the region, and the variance data of all regions constitute the regional variance data.

[0082] Step S23: Calculate thickness non-uniformity according to the regional variance data to obtain thickness non-uniformity data; Specifically, the overall thickness non-uniformity is calculated based on the variance data of each region, which helps to evaluate the stability and uniformity of the sample surface thickness. The thickness non-uniformity is obtained by calculating the weighted average or maximum value of the variance of all regions. Taking into account the different contributions of different regions to the overall non-uniformity, the regional variance is weighted averaged according to the area of ​​the region or the number of thickness points to obtain the thickness non-uniformity data. Alternatively, if you only focus on the most non-uniform area in the sample, you can also select the area with the largest variance as the thickness non-uniformity index of the sample to obtain the thickness non-uniformity data. The overall thickness non-uniformity is calculated by the above method, which is a key indicator for evaluating the uniformity of the sample.

[0083] Step S24: performing critical area detection on the thickness non-uniformity data to obtain critical area data; Specifically, critical areas refer to areas where thickness fluctuates violently or exceeds a preset tolerance range, which affects the quality and performance of the sample. Therefore, the detection of critical areas is helpful for analysis. A critical threshold is set according to actual needs. If the thickness variance or non-uniformity of a certain area exceeds the threshold, the area is considered to be a critical area. For each area, if its variance or thickness non-uniformity exceeds the critical threshold, it is marked as a critical area. Collect the marks of all detected critical areas to obtain critical area data.

[0084] Step S25: Integrate the thickness non-uniformity data and the critical area data to obtain thickness uniformity data.

[0085] Specifically, the thickness non-uniformity data and critical area data are integrated to obtain the thickness uniformity evaluation results. Not only the overall thickness non-uniformity will be considered, but also the distribution of critical areas and their impact. Combining the thickness non-uniformity data and critical area data, the uniformity of the sample can be comprehensively evaluated through a combination of multi-dimensional indicators. For example, the overall non-uniformity is calculated and the critical area is marked. The thickness uniformity data uses the following comprehensive evaluation method: ,in, is the thickness uniformity evaluation value / thickness uniformity data, is the thickness non-uniformity data, is the number of critical regions, is the total number of areas of the sample. This formula takes into account the combined effects of thickness non-uniformity and the number of critical areas on uniformity evaluation. Through the integration step, the thickness uniformity data is obtained , which can be used for further sample quality analysis and optimization.

[0086] Preferably, step S3 specifically includes: Step S31: formulating sample screening rules according to thickness uniformity data to obtain sample screening rule data; Specifically, sample screening rules are developed by analyzing the thickness uniformity data. These rules are based on the sample uniformity data to evaluate which samples meet the requirements of the tensile strength test. The screening rules are based on factors such as thickness uniformity, maximum variance, and the size of the critical area. Key indicators such as regional variance are extracted from the thickness uniformity data. and overall heterogeneity . Set standard threshold and If the uniformity data of a sample exceeds these thresholds, the sample is considered to be unqualified for the tensile strength test. For critical area data, the maximum number of critical areas allowed can be set. , samples exceeding this number will also be excluded. The sample screening rules are expressed by the following conditions: , , , and obtain screening rule data. According to these screening conditions, all samples are divided into samples that meet the rules and samples that do not meet the rules, and the data of samples that meet the rules are saved as sample screening rule data.

[0087] Step S32: scoring the samples according to the sample screening rule data to obtain sample scoring data; Specifically, a score is assigned to each sample based on the thickness uniformity data and the screening rule data. The score is based on the degree of uniformity of the sample. Samples with high scores have better thickness uniformity and are suitable for tensile strength testing. Weights are assigned to different uniformity indicators. For example, thickness non-uniformity and regional variance The weight coefficient can be set according to actual requirements and , the number of critical regions can also affect the score. The score of each sample It can be obtained by the following weighted calculation formula: ,in, and It is a sample The uniformity data, , and is the weight coefficient, It is a sample By scoring all samples, we can obtain the sample scoring data S={S1,S2,…,Sn}, where For samples 's rating.

[0088] Step S33: classifying and sorting the sample scoring data to obtain sample classification and sorting data; Specifically, the samples are classified and sorted according to the scoring data. The samples can be divided into several categories, for example, high-quality samples, general samples and unqualified samples, and high-quality samples are given priority for tensile strength testing. For example, the scoring range is set as: high-quality samples , general samples: , unqualified samples: , For high rating, For all samples, the score is Sort from high to low, giving priority to samples with higher scores. After sorting, the sample classification and sorting data is obtained. ,in( ) represents the sample By classifying and sorting all samples, we can get the sample classification and sorting data.

[0089] Step S34: performing Monte Carlo simulation performance prediction based on the sample classification data to obtain sample performance distribution simulation data; Specifically, the Monte Carlo simulation method is used to predict the tensile strength of the sample. Monte Carlo simulation simulates the performance distribution of the sample by random sampling, considering the effect of thickness uniformity on tensile strength. According to the thickness uniformity data of the sample, the relationship between tensile strength and thickness uniformity is defined. The tensile strength F is the product of uniformity U and variance Function, calculated as: ,in, , and is the regression coefficient, obtained by fitting the experimental data. Using the Monte Carlo method, a certain number of samples are randomly selected and the uniformity data is used to and Calculate the tensile strength of each sample , and generate the tensile strength distribution of the sample through multiple sampling simulations. Get performance distribution simulation data: Through the simulation process, get the tensile strength distribution data of the sample, reflecting the performance fluctuation of the sample under different uniformity conditions.

[0090] Step S35: Mark key samples according to the sample performance distribution simulation data to obtain tensile strength test sample selection data.

[0091] Specifically, based on the tensile strength distribution obtained by Monte Carlo simulation, key samples that show strong performance or meet test requirements are marked. A performance threshold is set. If the tensile strength prediction value of a sample is higher than the threshold , then the sample is considered to be a key sample. Greater than The samples are marked as key samples. The sample selection data for tensile strength test is obtained. After screening and marking, the sample data that meets the tensile strength test requirements is finally obtained.

[0092] Preferably, step S4 is specifically: Step S41: performing a tensile strength test operation through a tensile strength test device according to the tensile strength test sample selection data to obtain stress time series data; Specifically, after selecting a sample that meets the requirements, it will be clamped onto the tensile strength testing equipment. The tensile strength testing equipment will perform tensile loading according to the shape and material properties of the sample, and record the changes in stress and time of the sample during the stretching process in real time. The sample data selected from the previous steps will contain information such as its size, shape, and material, which will be used to configure the tensile testing equipment. In the tensile testing equipment, the sample is gradually stretched, and the equipment records the displacement changes and applied force of the sample in real time, and calculates stress and strain based on the force and displacement data. The testing equipment will record the changes in stress and time in real time to form stress-time series data.

[0093] Step S42: performing curve fitting on the stress time series data to obtain stress-strain curve data; Specifically, stress-time series data is converted into stress-strain data, and a stress-strain curve is obtained by fitting. Curve fitting is to convert time series data into a function of strain, and smooth and fit it to obtain a more accurate stress-strain relationship. From the stress time series to stress-strain data, the displacement data recorded by the test equipment is used to calculate the strain at each moment to obtain stress-strain data. In order to obtain a smooth stress-strain curve, the stress-strain data is fitted using the least squares method. For example, a quadratic or cubic polynomial can be used to fit the data to obtain a continuous stress-strain curve. By fitting, a continuous stress-strain curve is obtained.

[0094] Step S43: Calculate the extreme strength value according to the stress-strain curve data to obtain the extreme strength value data; Specifically, the strength extreme value refers to the maximum stress point in the stress-strain curve, that is, the maximum load that the material can withstand when it is tensile. This point corresponds to the critical point before the sample breaks. Using the fitted stress-strain curve, a numerical method (such as the derivative method) is used to calculate the derivative of the stress-strain curve and find its maximum value point. The curve is , first calculate its first-order derivative ,in , , is the constant term of the curve derivative. By solving the equation with the first-order derivative equal to zero, we can get the extreme point of the curve. The stress value corresponding to the extreme point is the extreme strength value, which is expressed as or (Yield strength). For example, in the above fitting curve, the stress at the extreme point can be obtained by substituting the strain value corresponding to the extreme value into the polynomial fitting formula to obtain the strength extreme value data. The strength extreme value is a key indicator for each sample, indicating the maximum tensile strength of the sample.

[0095] Step S44: performing tensile strength evaluation according to the stress-strain curve data and the strength extreme value data to obtain tensile strength test data.

[0096] Specifically, the overall tensile strength performance of the sample is evaluated by combining the stress-strain curve and the extreme strength data. The evaluation is mainly based on the maximum stress value of the sample during the stretching process, as well as the elastic and plastic characteristics of the sample. The tensile strength takes the maximum stress value in the stress-strain curve as the tensile strength of the material. If there are multiple stress-strain curves, calculate the maximum stress value of each curve, and take the maximum value as the tensile strength evaluation result. When considering multiple performance indicators, a weighted method is used to comprehensively combine stress extreme values, elastic modulus, fracture strain and other data to give a tensile strength evaluation. For example, the comprehensive evaluation formula is: in, is the elastic modulus of the sample, is the maximum strain of the sample, , , is the weight coefficient. By evaluating the tensile strength of all samples, a set of tensile strength test data can be obtained, including the maximum stress value of each sample and other evaluation indicators.

[0097] Preferably, the present application also provides a wire insulation layer measurement sample system, which is used to perform the wire insulation layer measurement sample method as described above, and the wire insulation layer measurement sample system includes: A two-dimensional thickness distribution map generation module is used to scan and measure the thickness of the insulation layer of the numbered sample using a laser thickness gauge to generate two-dimensional thickness distribution map data; A thickness uniformity analysis module is used to perform thickness uniformity analysis based on two-dimensional thickness distribution map data to obtain thickness uniformity data; A tensile strength test sample selection module is used to select tensile strength test samples according to thickness uniformity data to obtain tensile strength test sample selection data; A tensile strength test module is used to perform a tensile strength test according to the tensile strength test sample selection data to obtain tensile strength test data; The wire insulation layer performance evaluation module is used to evaluate the wire insulation layer performance according to the two-dimensional thickness distribution map data, thickness uniformity data and tensile strength test data, and obtain the wire insulation layer performance evaluation data to perform auxiliary work for wire insulation layer sample measurement.

[0098] Therefore, from any point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the attached application documents rather than the above description, and it is intended that all changes falling within the meaning and scope of equivalent elements of the application documents are included in the present invention.

[0099] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be 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 the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for measuring a wire insulation layer sample, characterized in that: The following steps are involved: Step S1: Scan and measure the thickness of the insulation layer of the numbered sample using a laser thickness gauge to generate two-dimensional thickness distribution map data; Step S2: performing thickness uniformity analysis according to the two-dimensional thickness distribution map data to obtain thickness uniformity data; Step S3: Selecting tensile strength test samples according to the thickness uniformity data to obtain tensile strength test sample selection data; Step S4: performing a tensile strength test according to the tensile strength test sample selection data to obtain tensile strength test data; Step S5: Conducting wire insulation layer performance evaluation based on the two-dimensional thickness distribution map data, thickness uniformity data and tensile strength test data to obtain wire insulation layer performance evaluation data for conducting wire insulation layer sample measurement auxiliary operations.

2. The method according to claim 1, characterized in that Step S1 is specifically as follows: Perform spatial positioning and optical axis calibration on the laser thickness gauge to generate laser positioning data; Reconstruct the sample reference surface through a laser thickness gauge according to the laser positioning data to obtain reference surface data; According to the reference surface data, multi-point thickness sampling is performed by a laser thickness gauge to obtain preliminary thickness sampling matrix data; Perform thickness distribution interpolation calculation according to the thickness preliminary sampling matrix data to obtain thickness distribution interpolation data; A dynamic distribution map is generated according to the thickness distribution interpolation data to obtain two-dimensional thickness distribution map data.

3. The method according to claim 2, characterized in that The spatial positioning and optical axis calibration are specifically as follows: By using the laser thickness gauge’s integrated inertial sensors and external vision sensors, the spatial position data of the equipment is captured in real time; Perform posture calculation based on spatial position data to obtain device posture data; Perform posture error analysis based on the equipment posture data to obtain posture error data; Perform attitude calibration according to attitude error data to obtain attitude calibration data; Perform laser optical axis alignment on the posture calibration data to obtain optical axis alignment data; Perform error correction according to the optical axis alignment data to obtain error correction data; Laser positioning data is generated according to the spatial position data, attitude calibration data, optical axis alignment data and error correction data to obtain laser positioning data.

4. The method according to claim 2, characterized in that: The sample reference plane reconstruction is specifically as follows: Perform a preliminary scan of the sample using a laser thickness gauge according to the laser positioning data to obtain sample scanning data; Reconstructing a local reference plane according to the sample scanning data to obtain first reference plane data; Perform minimum convex hull reconstruction according to the sample scanning data to obtain the second reference surface data; Performing shape feature fusion on the first reference plane data and the second reference plane data to obtain reference plane data; The preliminary scan of the sample is as follows: Scan the sample surface with a laser thickness gauge to obtain point cloud data of the sample surface; Perform particle iterative filtering on the sample surface point cloud data to obtain scanning noise reduction data; Perform KD tree region division according to the scanned denoising data to obtain scanned division data; voxelize and resample the scanned divided data to obtain sample resampled data; The spatial coordinates are aligned according to the sample scanning data and the sample resampling data to obtain the sample scanning data.

5. The method according to claim 4, characterized in that The local reference surface reconstruction is specifically as follows: Extracting sample morphological features from the sample scanning data to obtain sample morphological feature data; Performing preliminary area division on the sample scanning data according to the sample morphological characteristic data to obtain first sample division data; Performing morphological clustering calculation according to the first sample partitioning data to obtain sample partitioning clustering data; Constructing a local connection graph according to the first sample partitioning data and the sample partitioning clustering data to obtain sample partitioning graph data; Finely dividing the sample partition graph data according to the sample partition cluster data to obtain second sample partition data; Performing feature detection on the second sample division data to obtain sample division feature data; Performing weighted screening on the second sample partitioning data according to the sample partitioning characteristic data to obtain local area partitioning data; Performing local reference surface fitting on the local area division data to obtain local reference surface fitting data; Perform local plane error correction based on local reference surface fitting data to obtain reference surface smoothing correction data; The local reference plane fusion is performed according to the reference plane smoothing correction data to obtain the first reference plane data.

6. The method according to claim 4, characterized in that The minimum convex hull reconstruction is specifically: Perform point cloud screening and simplification on sample scanning data to obtain sample screening data; Perform point cloud space optimization based on sample screening data to obtain point cloud optimization data; Construct an initial convex hull set based on point cloud optimization data to obtain preliminary convex hull data; Perform local adjustments on the preliminary convex hull data to obtain convex hull boundary optimization data; Performing surface reconstruction on the convex hull boundary optimization data to obtain convex hull optimization data; The preliminary convex hull data and the convex hull optimized data are fully and locally fused to obtain the second reference surface data.

7. The method according to claim 1, characterized in that Step S2 is specifically as follows: Performing regional calibration according to the two-dimensional thickness distribution map data to obtain regional calibration data; Calculate the thickness variance of the regional calibration data to obtain regional variance data; Calculate thickness non-uniformity according to regional variance data to obtain thickness non-uniformity data; Perform critical area detection on thickness non-uniformity data to obtain critical area data; The thickness non-uniformity data and the critical area data are integrated to obtain the thickness uniformity data.

8. The method according to claim 1, characterized in that Step S3 is specifically as follows: Formulate sample screening rules based on thickness uniformity data to obtain sample screening rule data; Scoring the samples according to the sample screening rule data to obtain sample scoring data; Classifying and sorting the sample scoring data to obtain sample classification and sorting data; Monte Carlo simulation performance prediction is performed based on sample classification data to obtain sample performance distribution simulation data; Key samples are marked according to the sample performance distribution simulation data to obtain the tensile strength test sample selection data.

9. The method according to claim 1, characterized in that: Step S4 is specifically as follows: According to the tensile strength test sample selection data, the tensile strength test operation is carried out through the tensile strength test equipment to obtain stress time series data; Perform curve fitting on stress time series data to obtain stress-strain curve data; Calculate the extreme strength value according to the stress-strain curve data to obtain the extreme strength value data; The tensile strength is evaluated based on the stress-strain curve data and the extreme strength data to obtain the tensile strength test data.

10. A wire insulation layer measurement sample system, characterized in that: Used to perform the wire insulation layer measurement sample method according to claim 1, the wire insulation layer measurement sample system comprises: A two-dimensional thickness distribution map generation module is used to scan and measure the thickness of the insulation layer of the numbered sample using a laser thickness gauge to generate two-dimensional thickness distribution map data; A thickness uniformity analysis module is used to perform thickness uniformity analysis based on two-dimensional thickness distribution map data to obtain thickness uniformity data; A tensile strength test sample selection module is used to select tensile strength test samples according to thickness uniformity data to obtain tensile strength test sample selection data; A tensile strength test module is used to perform a tensile strength test according to the tensile strength test sample selection data to obtain tensile strength test data; The wire insulation layer performance evaluation module is used to evaluate the wire insulation layer performance according to the two-dimensional thickness distribution map data, thickness uniformity data and tensile strength test data, and obtain the wire insulation layer performance evaluation data to perform auxiliary work for wire insulation layer sample measurement.

Images