A method and system for active infrared detection of annular curved surface defects based on continuous rotation photo-thermal reconstruction

Abstract

The application discloses a kind of annular curved surface defect active infrared detection method and system based on continuous rotation photo-thermal reconstruction, belong to nondestructive testing and active infrared thermal imaging technical field.To solve the problem that annular curved surface defect detection is difficult to give consideration to whole circle coverage and positioning accuracy.The present application includes: the annular curved surface to be measured is continuously rotated, periodic modulation photo-thermal excitation is applied, and time sequence infrared thermal image sequence is synchronously collected;Radial temperature profile is extracted from each image and mapped to corresponding rotation circle number and circumferential angle, and multi-circle development data body is constructed;Through rotation period correction, eliminate circumferential misplacement;Along the direction of rotation circle, static thermal characteristics are extracted, and cross-circle period response model is established to extract dynamic thermal characteristics;According to static and / or dynamic thermal characteristics, the circumferential position of defect is determined, and the size of defect is evaluated in combination with geometric size parameters;Two-dimensional development characteristics are mapped to annular curved surface model, and three-dimensional thermal feature reconstruction result is generated.The present application realizes the quality detection and evaluation of annular curved surface metal component.

Description

Technical Field

[0001] This invention relates to an active infrared detection method and system for annular surface defects based on continuous rotational photothermal reconstruction, belonging to the field of nondestructive testing and active infrared thermal imaging technology. Background Technology

[0002] Rotating mechanical components with toroidal surface structures are widely used in rotating machinery, rail transportation, aerospace, and precision equipment. The surface quality of these toroidal surfaces directly affects the contact performance and operational reliability of the components. During the processing, assembly, or service of toroidal surface metal components, defects such as scratches, grinding burns, friction burns, and tempering burns are easily generated on the inner or outer toroidal surfaces. These defects alter the local surface roughness, absorptivity, thermal diffusion characteristics, and stress state of the toroidal surface, easily inducing fatigue crack propagation and leading to early failure of the toroidal surface. Therefore, non-destructive testing, accurate location, and quantitative characterization of defects in toroidal surfaces have significant engineering application value.

[0003] I. Main limitations of existing annular surface defect detection technology:

[0004] Currently, surface defect detection on toroidal surfaces mainly includes methods such as visual inspection, machine vision inspection, eddy current testing, magnetic particle testing, ultrasonic testing, and infrared thermal imaging. Among these, visual inspection and machine vision struggle to identify low-contrast burns and near-surface structural changes, and are easily affected by metallic reflections and surface geometric distortions. Eddy current, magnetic particle, and ultrasonic testing typically rely on probe orientation and coupling conditions, making it difficult to achieve rapid and stable full-circumference scanning on toroidal surfaces. The detection results are limited by material conditions and signal interpretation complexity, making it difficult to directly obtain spatial distribution information within the complete circumference of the toroidal surface. Unlike conventional flat components, defects on toroidal surfaces can be distributed at any location on the complete circumference. Infrared cameras and vision cameras, when acquiring single frames from the side, can only cover a local area. If segmented shooting or manual rotation and stitching are used, repetitive positioning errors and viewing angle variations are easily introduced, making it difficult to meet engineering requirements in terms of detection efficiency and positioning accuracy.

[0005] II. Specific challenges faced by applying active infrared thermal imaging technology to the detection of toroidal curved surfaces:

[0006] Active infrared thermal imaging technology induces a thermal response in the object under test through an external heat source and acquires the surface temperature field using infrared thermal imaging equipment. It offers advantages such as non-contact operation, full-field imaging, and sensitivity to changes in thermal properties. However, most existing active infrared detection methods are geared towards flat samples or localized areas, directly analyzing single thermal images or two-dimensional thermal response images. For toroidal curved surfaces, the following specific difficulties arise:

[0007] (1) When observing from the side, a single frame image can only cover a local annular curved surface area and cannot directly obtain the thermal response distribution of the complete circle;

[0008] (2) The geometric distortion of the curved surface and the uneven heating make it difficult for traditional two-dimensional thermal maps to accurately reflect the true spatial location of defects;

[0009] (3) Scratches, burns and other defects are not visible. Their thermal response differences are reflected not only in the average temperature rise, but also in the dynamic characteristics such as periodic response amplitude, phase lag and background drift. It is difficult to achieve stable identification of weak defects by relying solely on instantaneous temperature maps or simple temperature difference maps.

[0010] III. The problem of periodic error accumulating in multi-turn unfolding artifacts during continuous rotational scanning:

[0011] To obtain complete circumferential inspection results, a straightforward approach is to continuously rotate the toroidal surface and scan it using lateral infrared observation. However, in practical engineering, turntable control errors, mechanical clearances, load fluctuations, and image acquisition synchronization deviations cause slight discrepancies between the actual rotation period and the nominal period. If the circumferential unfolding is performed directly according to the nominal period, this deviation gradually accumulates over multiple scans, causing the same physical defect to be mapped to different circumferential positions in different rotations. This results in artifacts such as circumferential misalignment, ghosting, double lines, and blurred boundaries, severely affecting the reliability of subsequent defect identification, positioning accuracy, and dimensional assessment. In existing technologies, some solutions use encoder feedback for angular domain resampling to correct rotational speed fluctuations, but these methods mainly focus on vibration signal order analysis and do not address the periodic correction and circumferential unfolding reconstruction of infrared thermal response image sequences under active photothermal excitation. Some solutions employ penetrant testing or visual imaging, but their detection processes are complex or only applicable to surface-visible defects, making it difficult to solve the problem of non-destructive characterization and three-dimensional spatial positioning of near-surface thermal property anomalies such as burns.

[0012] In summary, existing technologies still lack an active infrared detection method and system that can reconstruct the complete circumferential thermal response under lateral local infrared observation conditions for toroidal curved surface structures, and further complete defect identification, circumferential positioning, size assessment, and three-dimensional visualization characterization. Summary of the Invention

[0013] The purpose of this invention is to solve the problem that existing methods for detecting defects on toroidal surfaces cannot simultaneously achieve full coverage and positioning accuracy. This invention provides an active infrared detection method and system for toroidal surface defects based on continuous rotational photothermal reconstruction.

[0014] The present invention discloses an active infrared detection method for toroidal surface defects based on continuous rotational photothermal reconstruction, comprising:

[0015] The toroidal surface under test is continuously rotated by a rotating device, and periodically modulated photothermal excitation is applied to the surface of the toroidal surface, while a time-series infrared thermal image sequence of a local area of ​​the toroidal surface is acquired simultaneously.

[0016] Extract the radial temperature profile sequence along the direction of the annular surface cross section from each frame of the time-series infrared thermal image sequence;

[0017] Based on the time parameters of the rotational motion, the radial temperature profile of each frame is mapped to the corresponding rotation circle number and circumferential angle to construct a multi-circle unfolded data volume;

[0018] The rotation period of the multi-turn unfolded data volume is corrected to eliminate circumferential misalignment caused by the deviation between the actual rotation period and the nominal period.

[0019] Based on the corrected multi-cycle unfolded data volume, static thermal features are extracted along the rotation cycle direction, and cross-cycle periodic response models are established for each spatial location to extract dynamic thermal features.

[0020] The circumferential position of the toroidal surface defect is determined based on the static thermal characteristics and / or dynamic thermal characteristics, and the geometric dimensions of the defect are evaluated in conjunction with the toroidal surface geometric dimension parameters.

[0021] Based on the geometric structure of the toroidal surface metal component, a toroidal surface model is established, and the static thermal features and / or the dynamic thermal features are mapped to the toroidal surface model to generate a three-dimensional thermal feature reconstruction result.

[0022] Preferably, the periodically modulated photothermal excitation is sinusoidal modulated laser excitation, and its excitation power satisfies the following:

[0023] ;

[0024] in, For average excitation intensity, To excite the amplitude, For modulation frequency, For time.

[0025] Preferably, the specific method for extracting the radial temperature profile sequence along the annular surface cross-section direction from each frame of the time-series infrared thermal image sequence is as follows:

[0026] Infrared thermal imaging equipment synchronously acquires a sequence of local temperature response images of a toroidal surface, obtaining the first... Frame infrared temperature image ;

[0027] In each frame of the temporal infrared thermal image sequence, a narrow strip of Region of Interest (ROI) covering the cross-sectional direction of the toroidal surface is selected. The width of the ROI is perpendicular to the cross-sectional direction of the toroidal surface, and the height is parallel to the cross-sectional direction of the toroidal surface.

[0028] The average temperature of all pixels along the width direction of the ROI is obtained. Radial temperature profile corresponding to the frame:

[0029] ;

[0030] in, This represents the number of pixels in the width direction of the ROI. For the cross section of the annular surface or the radial direction coordinates.

[0031] Preferably, the specific method for mapping the radial temperature profile of each frame to the corresponding rotation circle number and circumferential angle based on the time parameters of the rotational motion to construct a multi-circle unfolded data volume is as follows:

[0032] Let the effective rotation period be , No. Frame rotation time is The reference start time is The cumulative rotation angle is:

[0033] ;

[0034] Further obtain the circumferential angle and rotation number :

[0035] ;

[0036] ;

[0037] Radial temperature profiles of each frame Map to the corresponding rotation number and discrete circumferential angle interval Construct a multi-loop unfolded data volume: ;

[0038] in, Represents the discretized circumferential direction of the first... The center value of each angle interval.

[0039] Preferably, the specific method for performing rotation period correction on the multi-ring unfolded data volume, and extracting static thermal features along the rotation ring direction based on the corrected multi-ring unfolded data volume is as follows:

[0040] A candidate period set is set up near the nominal rotation period, and candidate unfolded data volumes are constructed for each period.

[0041] Calculate the edge sharpness, odd-even circle difference, and cross-circle standard deviation indices;

[0042] The optimal candidate period is selected as the effective rotation period based on the comprehensive score. ;

[0043] For the corrected unfolded data volume Taking the median along the direction of rotation, we obtain the median temperature characteristic map:

[0044] ;

[0045] This feature is used to characterize steady-state or quasi-steady-state thermal anomalies caused by defects in toroidal surfaces.

[0046] Preferably, the specific method for establishing a cross-cycle periodic response model for each spatial location to extract dynamic thermal features is as follows:

[0047] For the corrected unfolded data volume Each spatial location Establish a cross-cycle temperature response model along the direction of rotation:

[0048] ;

[0049] in, As the temperature baseline, The linear drift coefficient is... For the centralized time variable, For rotation number, For the first The excitation phase corresponding to the loop, and These are two orthogonal thermal response components. For residual terms;

[0050] After independently fitting the parameters for each spatial location, the periodic response amplitude is calculated. and phase :

[0051] ;

[0052] ;

[0053] This yields amplitude diagrams, phase diagrams, and linear drift diagrams that reflect the differences in the dynamic response of defects, serving as dynamic thermal characteristics.

[0054] Preferably, the specific method for determining the circumferential position of the annular surface defect based on static thermal characteristics and / or dynamic thermal characteristics includes:

[0055] Median temperature feature map and / or periodic response amplitude characteristic map radial direction Integrating, we obtain the circumferential integral curve that varies with the circumferential angle:

[0056] ;

[0057] ;

[0058] in, For the radial integration range, Radial pixel resolution, This is the integral curve of the median temperature. The integral curve represents the amplitude.

[0059] The circumferential position of the toroidal surface defect is determined based on the peak value of the circumferential integral curve within the local angular range.

[0060] ,or ;

[0061] The circumferential angle corresponding to the center of the defect;

[0062] When a defect causes a temperature increase, the location of the local maximum value of the integral curve is taken as the defect location.

[0063] When a defect causes a decrease in temperature, the location of the local minimum value of the integral curve is taken as the defect location.

[0064] This enables precise positioning of defects on annular surfaces in the circumferential direction.

[0065] Preferably, the specific method for evaluating the geometric dimensions of defects by combining the geometric dimensional parameters of the toroidal surface is as follows:

[0066] Based on the characteristic distribution of the defective region and the adjacent non-defective region, calculate the contrast ratio, signal-to-noise ratio, or statistical separability index.

[0067] Given the geometric dimensions of the toroidal surface, the circumferential angle difference and radial pixel difference occupied by the defect in the unfolded data volume are converted into the actual circumferential length and the actual radial length, respectively.

[0068] Let the angles corresponding to the two endpoints of the defect in the circumferential direction be . and The pixel positions corresponding to the two endpoints in the radial direction are and The toroidal surface directly corresponds to Radial pixel resolution is ,but:

[0069] ;

[0070] ;

[0071] ;

[0072] in, This represents the actual circumferential length of the defect. This represents the actual radial length of the defect. Estimate the length of the defect.

[0073] Preferably, the specific method for establishing a toroidal surface model based on the geometric structure of the toroidal surface metal component, mapping the static thermal features and / or the dynamic thermal features to the toroidal surface model, and generating a three-dimensional thermal feature reconstruction result is as follows:

[0074] A toroidal surface model is established based on the geometric dimensions of the toroidal surface. The two-dimensional circular unfolded feature map is mapped to the three-dimensional toroidal surface according to the circumferential angle and radial position to form a three-dimensional thermal feature reconstruction result.

[0075] The present invention discloses an active infrared detection system for toroidal surface defects based on continuous rotational photothermal reconstruction, comprising:

[0076] A rotating bearing module is used to support the inner or outer side of the annular curved metal component to be tested and to make its annular curved surface rotate continuously.

[0077] A photothermal excitation module is used to apply periodically modulated photothermal excitation to the annular curved surface; the photothermal excitation module includes a semiconductor laser and a collimation and beam expansion unit and a beam homogenization unit disposed on its output optical path, for forming a planar heat source covering the annular curved surface;

[0078] An infrared thermal imaging acquisition module is used to acquire a sequence of temporal infrared thermal images of a local area of ​​a toroidal surface during continuous rotation from the side; the infrared thermal imaging acquisition module includes response band coverage. Infrared thermal imager;

[0079] A synchronization control module is used to synchronize the rotation of the rotating bearing module, the excitation output of the photothermal excitation module, and the image acquisition of the infrared thermal imaging acquisition module. The synchronization control module includes a data acquisition card, which is used to output a slow sine modulation signal to control the output power of the semiconductor laser and to output a trigger signal to control the image acquisition timing of the infrared thermal imager.

[0080] The data processing module is used to perform circumferential unfolding and thermal feature reconstruction on the time-series infrared thermal image sequence, so as to output the circumferential position, geometric size and three-dimensional thermal feature reconstruction results of the annular surface defect.

[0081] Advantages of this invention:

[0082] 1. By continuously rotating and scanning, the local observation field of the lateral infrared camera is expanded to the complete circumference of the toroidal surface, breaking through the limitation that a single frame image can only cover a local area. The entire circle of the toroidal surface can be detected without manual repositioning and complex probe scanning. Furthermore, the two-dimensional unfolded features are mapped to the toroidal surface model to form a three-dimensional thermal feature reconstruction result, which intuitively displays the spatial location and coverage of the defect.

[0083] 2. An effective rotation period correction method is introduced. By scanning candidate periods near the nominal period and combining indicators such as edge sharpness, odd-even cycle consistency and cross-cycle stability, the optimal effective period is determined. This effectively eliminates multi-cycle unfolding misalignment, ghosting and boundary blurring caused by turntable control error and acquisition synchronization deviation, and improves the defect circumferential positioning accuracy and detection consistency.

[0084] 3. A cross-cycle linear thermal response model was established based on slow sinusoidal laser thermal excitation. Multidimensional thermal features such as median temperature, periodic response amplitude, phase, and linear drift were extracted. At the same time, the steady-state thermal anomaly and dynamic thermal response differences in the defect area were utilized to enhance the identification stability and anti-background drift ability of low-contrast scratches, burns, and weakly visible defects.

[0085] 4. By using radial integral analysis, defect peak location, and toroidal surface geometry mapping, the circular location, size estimation, and three-dimensional visualization output of toroidal surface defects were achieved. Attached Figure Description

[0086] Figure 1 This is a principle block diagram of the active infrared detection method for annular surface defects based on continuous rotation photothermal reconstruction as described in this invention.

[0087] Figure 2 This is a flowchart of the active infrared detection method for annular surface defects based on continuous rotational photothermal reconstruction as described in this invention;

[0088] Figure 3 This is a schematic diagram of the detection system environment according to a specific embodiment of the present invention;

[0089] Figure 4 The images shown are two-dimensional infrared thermal feature reconstruction results obtained by the method of the present invention, wherein Figure (a) is the median temperature map, Figure (b) is the amplitude map, Figure (c) is the phase map, and Figure (d) is the linear drift map.

[0090] Figure 5 This is a diagram showing the circumferential location and radial integration results of defects based on the circumferential unfolded feature map;

[0091] Figure 6 This is a diagram showing the reconstruction result of the three-dimensional annular thermal features of the annular surface defect. Detailed Implementation

[0092] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0093] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0094] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but this is not intended to limit the scope of the invention.

[0095] Example 1:

[0096] The following is combined with Figures 1-6 This embodiment describes an active infrared detection method for toroidal surface defects based on continuous rotational photothermal reconstruction, comprising the following steps:

[0097] Step 1: Rotate the toroidal surface under test continuously using a rotating device, apply periodically modulated photothermal excitation to the surface of the toroidal surface, and simultaneously acquire a time-series infrared thermal image sequence of a local area of ​​the toroidal surface.

[0098] In this embodiment, the annular curved surface metal component specimen is placed horizontally on an electric rotary table and rotated continuously at an approximately constant angular velocity. An infrared camera observes the annular curved surface region from the side to ensure that the annular curved surface surface can enter the camera's field of view. A laser is located above the specimen and, after collimation, beam expansion, and homogenization, forms a planar heat source to apply periodic thermal excitation to the annular curved surface surface. Since the camera's side-view field of view can only cover a local annular curved surface region at a time, continuous rotation scanning allows different circumferential positions to pass through the camera's observation area sequentially, thereby achieving complete 360° annular curved surface thermal response acquisition.

[0099] Signal generation and synchronization control were handled by the NI USB-6229 data acquisition card. A slow sinusoidal modulation signal was output from the PC to control the output power of the semiconductor laser, causing it to generate periodic thermal excitation. Simultaneously, another channel of the data acquisition card was used to trigger the infrared camera, achieving synchronization between laser excitation, turntable rotation, and infrared image acquisition. The infrared camera used in the experiment was a Guide-IPT640M with an image resolution of 640×512 pixels and a response band of 8–14 μm. The acquired infrared thermal image sequences were transmitted to the computer for processing via an RJ45 interface.

[0100] The maximum rotational speed of the turntable used in the experiment was 10° / s, corresponding to a minimum time of 36 seconds to complete one revolution. To avoid overheating or damage to the motor due to prolonged operation of the turntable near its maximum speed, the nominal rotational period of the turntable was set to [missing information]. The laser modulation frequency was set to 0.0025Hz. For each frequency group, the acquisition process covered approximately two complete excitation cycles and included multiple turntable rotation cycles to ensure that the same circumferential position could be repeatedly sampled under different excitation phases. The infrared camera acquired data at a frame rate of 30fps, enabling the acquisition of sufficiently dense local annular surface temperature profiles during each rotation. A schematic diagram of the implementation environment is shown below. Figure 3 As shown.

[0101] A detection system is set up and the annular curved metal component to be tested is installed. The inner or outer side of the annular curved metal component to be tested is mounted on a rotating platform, so that the annular curved area sequentially enters the lateral observation field of the infrared thermal imaging equipment during continuous rotation; a modulated laser photothermal excitation module, an infrared thermal image acquisition module, a synchronization control module, and a data processing module are arranged.

[0102] In this embodiment, a modulated laser is applied and an infrared image sequence is acquired. The rotating platform is controlled to rotate continuously at a set nominal period, and a slow sinusoidal modulation is applied to the laser heat source to ensure that the laser power satisfies the following:

[0103] ;

[0104] in, For average excitation intensity, To excite the amplitude, For modulation frequency, For time. Infrared thermal imaging equipment synchronously acquires a sequence of local temperature response images of a toroidal surface, obtaining the first... Frame infrared temperature image .

[0105] Step 2: Extract the radial temperature profile sequence along the direction of the annular curved surface cross section from each frame of the time-series infrared thermal image sequence;

[0106] In each frame of infrared image, a narrow strip-shaped region of interest (ROI) covering the cross-section of the toroidal surface is selected, and the temperature along the width of the ROI is averaged to obtain the [number of frames]. Radial temperature profile corresponding to the frame:

[0107] ;

[0108] in, This represents the number of pixels in the width direction of the ROI. For the cross section of the annular surface or the radial direction coordinates.

[0109] Step 3: Based on the time parameters of the rotational motion, map the radial temperature profile of each frame to the corresponding rotation circle number and circumferential angle to construct a multi-circle unfolded data volume;

[0110] Let the effective rotation period be , No. Frame rotation time is The reference start time is The cumulative rotation angle is:

[0111] ;

[0112] Further obtain the circumferential angle and rotation number :

[0113] ;

[0114] ;

[0115] Based on this, the radial temperature profile of each frame is... Mapping to the corresponding rotation circle number and circumferential angle, constructing the unfolded data volume. .

[0116] Step 4: Perform rotation period correction on the multi-turn unfolded data volume to eliminate circumferential misalignment caused by the deviation between the actual rotation period and the nominal period;

[0117] A candidate period set is set up around the nominal rotation period, and candidate unfolded data volumes are constructed for each. The edge sharpness, odd-even cycle difference, and cross-cycle standard deviation are calculated. The optimal candidate period is selected as the effective rotation period based on the comprehensive score. This reduces multi-loop unfolding misalignment, ghosting, and boundary blurring.

[0118] Step 5: Based on the corrected multi-cycle unfolded data volume, extract static thermal features along the rotation direction, and establish cross-cycle periodic response models for each spatial location to extract dynamic thermal features.

[0119] Construct a median temperature feature map. Then, analyze the corrected expanded data volume. Taking the median along the direction of rotation, we obtain the median temperature characteristic map:

[0120] ;

[0121] This feature is used to characterize steady-state or quasi-steady-state thermal anomalies caused by defects in toroidal surfaces.

[0122] A slow sinusoidal linear thermal response model was established and phase-locked loop features were extracted. For each position in the expanded data volume... Establish a cross-cycle temperature response model:

[0123] ;

[0124] in, As the temperature baseline, The linear drift coefficient is... For centralized time, For the excitation phase, and These are two orthogonal thermal response components. After parameter fitting, the amplitude and phase of the periodic response are calculated:

[0125] ;

[0126] ;

[0127] This yields the amplitude diagram, phase diagram, and linear drift diagram.

[0128] Step 6: Determine the circumferential position of the annular surface defect based on the static thermal characteristics and / or dynamic thermal characteristics, and evaluate the geometric dimensions of the defect in conjunction with the annular surface geometric dimension parameters;

[0129] Construct the radial integral curve and locate the defect. Integrate the median temperature map and amplitude map radially to obtain the circumferential integral curve:

[0130] ;

[0131] ;

[0132] in, This is the integral curve of the median temperature. This is the amplitude integral curve. The defect circumferential position is determined based on the peak value of the integral curve within a local angular range. ,or ;

[0133] Evaluate defect separability and estimate size. Calculate contrast, signal-to-noise ratio, or statistical separability indices based on the characteristic distribution of the defective region and adjacent non-defective regions; under known toroidal surface geometry, convert circumferential angle difference and radial pixel difference into actual length. If the two endpoints of the defect are... and The diameter of the toroidal surface is Radial pixel size is ,but:

[0134] ;

[0135] ;

[0136] ;

[0137] Among them, L * Estimate the length of the defect.

[0138] Step 7: Establish a toroidal surface model based on the geometric structure of the toroidal surface, and map the static thermal features and / or the dynamic thermal features to the toroidal surface model to generate a three-dimensional thermal feature reconstruction result.

[0139] Perform 3D toroidal thermal feature reconstruction and output the results. Establish a toroidal surface model based on the toroidal surface geometry, and unfold the 2D circular feature map according to the circumferential angles. and radial position The data is mapped onto a three-dimensional toroidal surface to form a three-dimensional thermal feature reconstruction result, and outputs a two-dimensional feature map, circumferential integral curve, defect circumferential position, defect size, and three-dimensional visualization result.

[0140] Figures 4 to 6 Active infrared detection results for toroidal surface defects based on continuous rotational photothermal reconstruction are presented. Figure 4 As can be seen, after circumferential expansion and slow sinusoidal thermal response feature extraction, the toroidal surface defect appears as obvious strip-shaped or locally high-response regions in the median temperature map and amplitude map, while the phase map and linear drift map provide auxiliary dynamic thermal response information. Figure 5 By further combining the two-dimensional circular unfolding features with the circumferential response curve, it can be seen that the defect location has a good correspondence with the peak area in the integral curve, indicating that the method can effectively locate the defect circumferential position. Figure 6 By mapping the two-dimensional unfolded features onto the three-dimensional toroidal surface, the spatial distribution, local orientation, and response intensity of defects in the circumferential direction of the toroidal surface are intuitively displayed, verifying the effectiveness of the method in the three-dimensional visualization and characterization of defects on toroidal surfaces.

[0141] Example 2:

[0142] The following is combined with Figure 1 This embodiment describes an active infrared detection system for toroidal surface defects based on continuous rotational photothermal reconstruction, comprising:

[0143] The rotating bearing module is used to support the inner or outer side of the annular curved metal component to be tested and to make its annular curved surface rotate continuously.

[0144] In this embodiment, the annular curved metal component specimen is placed horizontally on an electric rotary table. The maximum rotational speed of the rotary table is 10° / s, and the nominal rotational period is set as follows: .

[0145] A photothermal excitation module is used to apply periodically modulated photothermal excitation to the annular curved surface; the photothermal excitation module includes a semiconductor laser and a collimation and beam expansion unit and a beam homogenization unit disposed on its output optical path, for forming a planar heat source covering the annular curved surface.

[0146] Periodic thermal excitation is applied to the toroidal surface. The laser power follows a sinusoidal modulation form, with the modulation frequency set to 0.0025 Hz.

[0147] An infrared thermal image acquisition module is used to acquire a sequence of temporal infrared thermal images of a local area of ​​a toroidal surface during continuous rotation from the side. The module includes an infrared thermal imager; in this embodiment, a Guide-IPT640M infrared thermal imager is used, with an image resolution of 640×512 pixels and a response band covering 8μm~14μm. The infrared camera observes the toroidal surface area from the side, with a single frame image covering a local area of ​​the toroidal surface. Continuous rotation allows different circumferential positions of the toroidal surface to sequentially enter the observation field of view. The acquisition frame rate is 30fps, and the acquired infrared thermal image sequence is transmitted to a computer for processing via an RJ45 interface.

[0148] A synchronization control module is used to synchronize the rotation of the rotating bearing module, the excitation output of the photothermal excitation module, and the image acquisition of the infrared thermal imaging acquisition module. The synchronization control module includes a data acquisition card, which in this embodiment uses an NI USB-6229 data acquisition card. It is used to output a slow sine modulation signal to control the output power of the semiconductor laser and to output a trigger signal to control the image acquisition timing of the infrared thermal imager, thereby achieving synchronization between laser excitation, turntable rotation, and infrared image acquisition.

[0149] The data processing module is used to perform circumferential unfolding and thermal feature reconstruction on the time-series infrared thermal image sequence, so as to output the circumferential position, geometric size and three-dimensional thermal feature reconstruction results of the annular surface defect.

[0150] Specifically, the data processing module performs the following processing on the acquired infrared image sequence: For each frame of infrared image, a narrow strip ROI covering the cross-sectional direction of the annular surface is selected, and the temperature along the width of the ROI is averaged to obtain the radial temperature profile corresponding to each frame; the cumulative rotation angle, circumferential angle, and rotation circle number are calculated based on the rotation time parameter, and the radial temperature profile of each frame is mapped to the corresponding rotation circle number and circumferential angle to construct the unfolded data volume; a candidate period set is set near the nominal rotation period, and candidate unfolded data volumes are constructed respectively, and edge sharpness, odd-even circle difference, and cross-circle standard deviation indices are calculated. The optimal effective rotation period is selected based on the comprehensive score to reduce multi-circle unfolding misalignment, ghosting, and boundary blurring; the median of the corrected unfolded data volume is taken along the rotation circle direction. The process involves: obtaining a median temperature feature map; establishing a cross-cycle temperature response model for each position in the unfolded data volume; calculating the periodic response amplitude and phase after parameter fitting to obtain an amplitude map, a phase map, and a linear drift map; integrating the median temperature map and amplitude map radially to obtain a circumferential integral curve; determining the defect circumferential position based on the peak value of the integral curve within a local angle range; evaluating the separability of the defect based on the feature distribution between the defect area and adjacent non-defect areas; converting the circumferential angle difference and radial pixel difference into actual lengths; establishing a toroidal surface model based on the toroidal surface geometry; mapping the two-dimensional circumferential unfolded feature map to the three-dimensional toroidal surface to form a three-dimensional thermal feature reconstruction result; and outputting a two-dimensional feature map, a circumferential integral curve, the defect circumferential position, the defect size, and a three-dimensional visualization result.

[0151] The spatial arrangement of the components of the system is as follows: Figure 3 The diagram shows the installation positions and relative orientations of the rotating platform, laser, infrared thermal imager, and data processing computer.

[0152] The working principle of this invention is as follows:

[0153] This invention mounts the inner or outer side of the toroidal curved metal component under test on a rotating platform and rotates it continuously. A modulated laser applies periodic photothermal excitation to the toroidal surface, while an infrared thermal imaging device acquires a sequence of local thermal response images of the toroidal surface from the side. Each frame of the local infrared thermal image is converted into a radial temperature profile along the cross-sectional direction of the toroidal surface by temperature window extraction. Combining the image acquisition time, turntable rotation time, and effective rotation period, the local radial temperature profiles of different frames are mapped to corresponding circumferential angles and rotation numbers, constructing a complete toroidal surface unfolded data volume. Furthermore, the optimal rotation period is determined through candidate rotation period scanning. The rotation period is optimized to improve the consistency of multi-turn unfolding and the clarity of defect boundaries. A slow sinusoidal linear thermal response model is established on the unfolded data volume to extract multi-dimensional infrared thermal feature maps such as median temperature, periodic response amplitude, phase, and linear drift. Then, through radial integral curves, defect peak localization, and toroidal surface geometric dimension mapping, the identification, circumferential localization, range characterization, and size estimation of toroidal surface defects are realized. Finally, the two-dimensional circumferential unfolding feature map is mapped to the toroidal surface model of the toroidal surface metal component to form a three-dimensional toroidal thermal feature reconstruction result, realizing an intuitive characterization of the spatial location, circumferential coverage, and thermal response intensity of toroidal surface defects.

[0154] An infrared camera acquires a set of time-series images during continuous rotating scanning. Let the... Frame infrared temperature image is ,in , Total number of frames and These represent the horizontal and vertical pixel coordinates within the image plane, respectively. This indicates the temperature at that pixel, in units of... In the experiment, the inner side of the toroidal curved metal component was placed on a turntable and rotated continuously at an approximately constant angular velocity. The laser heat source was modulated using a slow sinusoidal method. Let the excitation power be... Then we have:

[0155] ;

[0156] in, For average excitation intensity, The excitation amplitude is expressed in W. Let be time, in seconds (s). The excitation phase, measured in rad, is for the sinusoidal modulation frequency. The excitation phase can be written as:

[0157] ;

[0158] in, This is the initial phase. During data acquisition, each frame of the infrared image... Each corresponds to a collection time. Rotation time of one turntable and an excitation phase .

[0159] Since the infrared camera observes the toroidal surface from the side, only a local area of ​​the toroidal surface can be seen in a single frame image. Therefore, to achieve a full scan of the toroidal surface, a narrow strip-shaped region of interest (ROI) needs to be selected in the image first. Let the image range of the ROI be... ,in , , , All coordinates are pixel coordinates. The number of pixels in the width direction of the ROI is defined as follows: In this article, the ROI width is usually taken as a smaller value. This is used to reduce the impact of lateral field-of-view variations. The longitudinal direction of the ROI covers the cross-sectional direction of the toroidal surface and is denoted as radial or cross-sectional coordinates in the subsequent unfolded diagram. For the first In a frame image, the temperature within the ROI can be expressed as: ,in , , Let be the number of pixels in the height direction of the ROI. To suppress lateral noise and transform the side-view local window into a one-dimensional toroidal surface cross-section temperature profile, the average is calculated over the width direction of the ROI, yielding:

[0160] ;

[0161] in Indicates the first The radial temperature profile of the frame at the local annular surface window, in units of Thus, the original two-dimensional infrared image sequence Converted into a local toroidal surface temperature profile sequence To reconstruct a complete 360° circumferential unfolded diagram from the local side-view observations, each frame needs to be mapped to the corresponding circumferential angle and rotation number based on the turntable movement. Let the nominal turntable period be... The unit is seconds (s). Theoretically, if the turntable period is strictly equal to... Then the first The cumulative rotation angle of the frame relative to the start time can be written as:

[0162] ;

[0163] in, To accumulate rotation angles, This serves as the zero point for rotation time reference, measured in seconds. However, in actual experiments, there may be a slight deviation between the nominal turntable period and the true effective rotation period. If used directly... When the same physical defect is expanded in a circle, it will be mapped to different circumferential positions in different rotations, resulting in artifacts such as circumferential misalignment, double lines, ghosting, and blurring after multiple rotations are superimposed.

[0164] Therefore, an effective turntable cycle is introduced. And use it to recalculate the cumulative angle corresponding to each frame:

[0165] ;

[0166] From a cumulative perspective, we can further obtain the first... The circumferential angle and rotation number corresponding to the frame:

[0167] ;

[0168] in, The circumferential angle of the current frame within one revolution; For rotation number; This indicates rounding down to the nearest integer.

[0169] Sure Then, the local temperature profile of each frame This can be mapped to the corresponding rotation number. and circumferential angle To construct the unfolded graph of the rules, the circumferential range of 0° to 360° is discretized into... Each angle interval has an angular resolution of [number] . , No. The center angle of each angular interval is denoted as:

[0170] ;

[0171] No. The angle interval where the frame falls is numbered as follows For the Circle, First The circumferential angle interval and the first A radial position, satisfying all and The set of frames is denoted as The unfolding temperature within this unit is defined as:

[0172] ;

[0173] in The number of frames that fall into this cell. The unit is Thus, the original infrared time series was converted into a three-dimensional data volume. , .in This represents the effective number of rotations. This 3D data volume describes the different rotations. Different circumferential angles and different radial positions The temperature response of the toroidal surface at that location is the basis for the construction of all subsequent feature maps.

[0174] After obtaining the multi-loop unfolding data, for each spatial location Each of the rotating rings has a temperature value. To obtain a robust overall temperature distribution, the median temperature is taken along the direction of rotation and defined as:

[0175] ;

[0176] in, The unit is , indicating the position of the toroidal surface The typical temperature level throughout the experiment is robust to occasional noise, abnormal frames, and local disturbances, and can be used to display the overall temperature distribution, heating non-uniformity, and some obvious defect areas of the toroidal surface.

[0177] Because laser thermal excitation is a slow sinusoidal modulation, normal and defective regions will exhibit different periodic thermal responses due to differences in absorptivity, thermal diffusion characteristics, interfacial thermal resistance, or near-surface tissue conditions. For a fixed spatial location... Each lap corresponds to a time. and an excitation phase To simultaneously describe the local temperature baseline, long-term thermal drift, and periodic thermal response, a slow sinusoidal linear temperature response model is independently established for each spatial location:

[0178] ;

[0179] in This is a local temperature baseline, in units of ; This is the linear temperature drift coefficient, with units of . ; and These are two orthogonal thermal response components relative to the excitation phase, with units of 1. ; The excitation phase is expressed in rad. To reduce noise, the time variable is centered to decrease the correlation between the constant term and the linear drift term. ,in:

[0180] ;

[0181] For a certain fixed position Cross-circle observations can be written in matrix form. ,in:

[0182] ;

[0183] And the design matrix is:

[0184] ;

[0185] The parameters are estimated using the least squares method:

[0186] ;

[0187] You can get , , , Linear drift This reflects the slow heating or cooling trend at this location during the data acquisition process, excluding periodic excitation, and is expressed in units of... Sine response part It can be written equivalently as:

[0188] ;

[0189] Therefore, each spatial location The periodic thermal response amplitude and phase are defined as follows:

[0190] ;

[0191] ;

[0192] Therefore, the unit of the periodic thermal response amplitude is... Phase unit is rad, amplitude This characterizes the intensity of the periodic temperature response at this location to slow sinusoidal thermal excitation. When the defect region has a higher absorptivity, different thermal diffusion characteristics, or greater interfacial thermal resistance, its periodic temperature fluctuations may be enhanced. Therefore, abnormal areas such as burns and scratches often exhibit locally high responses in terms of amplitude. Phase This represents the phase shift of the local temperature response relative to the excitation signal, and its variation is related to local thermal diffusion and thermal hysteresis. Generally speaking, the slower the thermal diffusion process, the more pronounced the phase hysteresis of the temperature response relative to the excitation. For the low-frequency sinusoidal excitation and finite number of rotations conditions actually used, temperature, amplitude, and linear drift usually provide more intuitive defect visibility, while phase can serve as an auxiliary feature to characterize the differences in local thermal diffusion.

[0193] While the invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the invention. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that different dependent claims and features described herein can be combined in ways different from those described in the original claims. It is also understood that features described in conjunction with individual embodiments can be used in other described embodiments.

Claims

1. An active infrared detection method for toroidal surface defects based on continuous rotational photothermal reconstruction, characterized in that, It includes: The toroidal surface under test is continuously rotated by a rotating device, and periodically modulated photothermal excitation is applied to the surface of the toroidal surface, while a time-series infrared thermal image sequence of a local area of ​​the toroidal surface is acquired simultaneously. Extract the radial temperature profile sequence along the direction of the annular surface cross section from each frame of the time-series infrared thermal image sequence; Based on the time parameters of the rotational motion, the radial temperature profile of each frame is mapped to the corresponding rotation circle number and circumferential angle to construct a multi-circle unfolded data volume; The rotation period of the multi-turn unfolded data volume is corrected to eliminate circumferential misalignment caused by the deviation between the actual rotation period and the nominal period. Based on the corrected multi-cycle unfolded data volume, static thermal features are extracted along the rotation cycle direction, and cross-cycle periodic response models are established for each spatial location to extract dynamic thermal features. The circumferential position of the toroidal surface defect is determined based on the static thermal characteristics and / or dynamic thermal characteristics, and the geometric dimensions of the defect are evaluated in conjunction with the toroidal surface geometric dimension parameters. Based on the geometric structure of the toroidal surface metal component, a toroidal surface model is established, and the static thermal features and / or the dynamic thermal features are mapped to the toroidal surface model to generate a three-dimensional thermal feature reconstruction result.

2. The active infrared detection method for annular surface defects based on continuous rotational photothermal reconstruction according to claim 1, characterized in that, The periodically modulated photothermal excitation is sinusoidal modulated laser excitation, and its excitation power satisfies the following: ； in, For average excitation intensity, To excite the amplitude, For modulation frequency, For time.

3. The active infrared detection method for annular surface defects based on continuous rotational photothermal reconstruction according to claim 1, characterized in that, The specific method for extracting the radial temperature profile sequence along the annular curved surface cross-section direction from each frame of the time-series infrared thermal image sequence is as follows: Infrared thermal imaging equipment synchronously acquires a sequence of local temperature response images of a toroidal surface, obtaining the first... Frame infrared temperature image ; In each frame of the temporal infrared thermal image sequence, a narrow strip of Region of Interest (ROI) covering the cross-sectional direction of the toroidal surface is selected. The width of the ROI is perpendicular to the cross-sectional direction of the toroidal surface, and the height is parallel to the cross-sectional direction of the toroidal surface. The average temperature of all pixels along the width direction of the ROI is obtained. Radial temperature profile corresponding to the frame: ； in, This represents the number of pixels in the width direction of the ROI. For the cross section of the annular surface or the radial direction coordinates.

4. The active infrared detection method for annular surface defects based on continuous rotational photothermal reconstruction according to claim 1, characterized in that, Based on the time parameters of the rotational motion, the radial temperature profile of each frame is mapped to the corresponding rotation circle number and circumferential angle. The specific method for constructing a multi-circle unfolded data volume is as follows: Let the effective rotation period be , No. Frame rotation time is The reference start time is The cumulative rotation angle is: ； Further obtain the circumferential angle and rotation number : ； ； Radial temperature profiles of each frame Map to the corresponding rotation number and discrete circumferential angle interval Construct a multi-loop unfolded data volume: ; in, Represents the discretized circumferential direction of the first... The center value of each angle interval.

5. The active infrared detection method for annular surface defects based on continuous rotational photothermal reconstruction according to claim 1, characterized in that, The specific method for performing rotation period correction on the multi-cycle unfolded data volume, and extracting static thermal features along the rotation direction based on the corrected multi-cycle unfolded data volume is as follows: A candidate period set is set up near the nominal rotation period, and candidate unfolded data volumes are constructed for each period. Calculate the edge sharpness, odd-even circle difference, and cross-circle standard deviation indices; The optimal candidate period is selected as the effective rotation period based on the comprehensive score. ; For the corrected unfolded data volume Taking the median along the direction of rotation, we obtain the median temperature characteristic map: ； This feature is used to characterize steady-state or quasi-steady-state thermal anomalies caused by defects in toroidal surfaces.

6. The active infrared detection method for annular surface defects based on continuous rotational photothermal reconstruction according to claim 1, characterized in that, The specific method for establishing cross-cycle periodic response models for each spatial location to extract dynamic thermal features is as follows: For the corrected unfolded data volume Each spatial location Establish a cross-cycle temperature response model along the direction of rotation: ； in, As the temperature baseline, The linear drift coefficient is... For the centralized time variable, For rotation number, For the first The excitation phase corresponding to the loop, and These are two orthogonal thermal response components. For residual terms; After independently fitting the parameters for each spatial location, the periodic response amplitude is calculated. and phase : ； ； This yields amplitude diagrams, phase diagrams, and linear drift diagrams that reflect the differences in the dynamic response of defects, serving as dynamic thermal characteristics.

7. The active infrared detection method for annular surface defects based on continuous rotational photothermal reconstruction according to claim 1, characterized in that, The specific method for determining the circumferential position of the annular surface defect based on static thermal characteristics and / or dynamic thermal characteristics includes: Median temperature feature map and / or periodic response amplitude characteristic map radial direction Integrating, we obtain the circumferential integral curve that varies with the circumferential angle: ； ； in, For the radial integration range, Radial pixel resolution, This is the integral curve of the median temperature. The integral curve represents the amplitude. The circumferential position of the toroidal surface defect is determined based on the peak value of the circumferential integral curve within the local angular range. ,or ; The circumferential angle corresponding to the center of the defect; When a defect causes a temperature increase, the location of the local maximum value of the integral curve is taken as the defect location. When a defect causes a decrease in temperature, the location of the local minimum value of the integral curve is taken as the defect location. This enables precise positioning of defects on annular surfaces in the circumferential direction.

8. The active infrared detection method for annular surface defects based on continuous rotational photothermal reconstruction according to claim 1, characterized in that, The specific method for evaluating the geometric dimensions of defects by combining the geometric dimension parameters of the toroidal surface is as follows: Based on the characteristic distribution of the defective region and the adjacent non-defective region, calculate the contrast ratio, signal-to-noise ratio, or statistical separability index. Given the geometric dimensions of the toroidal surface, the circumferential angle difference and radial pixel difference occupied by the defect in the unfolded data volume are converted into the actual circumferential length and the actual radial length, respectively. Let the angles corresponding to the two endpoints of the defect in the circumferential direction be . and The pixel positions corresponding to the two endpoints in the radial direction are and The toroidal surface directly corresponds to Radial pixel resolution is ,but: ； ； ； in, This represents the actual circumferential length of the defect. This represents the actual radial length of the defect. Estimate the length of the defect.

9. The active infrared detection method for annular surface defects based on continuous rotational photothermal reconstruction according to claim 1, characterized in that, The specific method for establishing a toroidal surface model based on the geometric structure of the toroidal surface metal component, mapping the static thermal features and / or the dynamic thermal features to the toroidal surface model, and generating a three-dimensional thermal feature reconstruction result is as follows: A toroidal surface model is established based on the geometric dimensions of the toroidal surface. The two-dimensional circular unfolded feature map is mapped to the three-dimensional toroidal surface according to the circumferential angle and radial position to form a three-dimensional thermal feature reconstruction result.

10. An active infrared detection system for annular surface defects based on continuous rotational photothermal reconstruction, characterized in that, include: A rotating bearing module is used to support the inner or outer side of the annular curved metal component to be tested and to make its annular curved surface rotate continuously. A photothermal excitation module is used to apply periodically modulated photothermal excitation to the annular curved surface; the photothermal excitation module includes a semiconductor laser and a collimation and beam expansion unit and a beam homogenization unit disposed on its output optical path, for forming a planar heat source covering the annular curved surface; An infrared thermal imaging acquisition module is used to acquire a sequence of temporal infrared thermal images of a local area of ​​a toroidal surface during continuous rotation from the side; the infrared thermal imaging acquisition module includes response band coverage. Infrared thermal imager; A synchronization control module is used to synchronize the rotation of the rotating bearing module, the excitation output of the photothermal excitation module, and the image acquisition of the infrared thermal imaging acquisition module. The synchronization control module includes a data acquisition card, which is used to output a slow sine modulation signal to control the output power of the semiconductor laser and to output a trigger signal to control the image acquisition timing of the infrared thermal imager. The data processing module is used to perform circumferential unfolding and thermal feature reconstruction on the time-series infrared thermal image sequence, so as to output the circumferential position, geometric size and three-dimensional thermal feature reconstruction results of the annular surface defect.

Images