A structure surface temperature testing system based on photosensitive coating color difference identification

Abstract

The present application relates to the technical field of temperature test, in particular to a structure surface temperature test system based on color difference identification of photosensitive coating, which comprises a color difference feature extraction module, a heat conduction track analysis module and a temperature gradient reconstruction module.In the present application, the dynamic gradient analysis technology is used to capture the space-time evolution characteristics of the chroma field, improve the identification sensitivity of the non-uniform thermal field, maintain the stable temperature resolution under complex illumination, identify the propagation path of the transient thermal shock based on the cluster motion mode recognition of the heat conduction track, shorten the system response delay, adopt the nonlinear temperature-sensitive characteristic compensation mechanism to overcome the measurement deviation caused by the coating process difference, realize the higher precision temperature reconstruction of the curved surface structure, combine the spatial correction and the directional weighting to enhance the spatial analysis ability of the temperature gradient field, accurately identify the small heat influence area, and significantly reduce the temperature field reconstruction error of the large-scale structure through the multi-dimensional coupling analysis, so that the engineering applicability is better than that of the traditional method.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of temperature testing, and particularly relates to a structure surface temperature testing system based on color difference recognition of photosensitive paint. BACKGROUND

[0002] The technical field of temperature testing includes methods and devices for detecting and monitoring temperature changes in different environments or structures. The core of this technical field is to accurately obtain temperature information of target objects to assess their running state, material performance changes and safety control. Overall, temperature testing technology mainly covers contact and non-contact temperature measurement methods. The former includes the application of thermocouples, thermistors and other sensors, and the latter often uses infrared temperature measurement, thermal imaging detection and other principles to sense the temperature of the target. In engineering applications, structure surface temperature testing is not only an important means of health monitoring during operation, but also plays a key role in the construction stage. For example, in the process of composite material forming, early concrete curing or thermal spraying construction, accurate acquisition of the temperature field is crucial for quality control. Therefore, a temperature sensing system suitable for the whole process (including the construction period and the operation period) has significant engineering value.

[0003] Among them, the structure surface temperature testing system based on color difference recognition of photosensitive paint refers to a testing device and method that applies a specific photosensitive paint to the surface of the structure, and uses the color change rule of the paint at different temperatures to obtain surface color difference information through optical image acquisition equipment, thereby indirectly judging the temperature of the structure surface. The system includes the selection and preparation of photosensitive paint materials, the analysis of the influence of coating process on temperature-sensitive characteristics, the quantitative establishment of color difference change and temperature relationship, the spectral analysis and processing of image data, and the temperature calculation method based on image feature extraction. Its completion process usually uses standard light source to irradiate to obtain the color image of the structure surface, analyzes the color component difference of the target area in the image, refers to the known temperature response curve of the paint, and finally realizes the numerical recognition of the temperature of the structure surface.

[0004] The traditional photosensitive coating temperature measurement method is limited to the static color difference analysis framework, and it is difficult to effectively process the space-time correlation characteristics in the dynamic temperature field, and it is easy to lose key heat conduction information in the rapid temperature change scene. The existing chroma-temperature mapping model does not consider the nonlinear characteristics of the thermal response of the material, resulting in the missing of key temperature nodes in the curing monitoring of multilayer composite materials. The discrete contact temperature measurement method is limited by the fixed point distribution method, and cannot completely capture the temperature gradient distribution characteristics of the curved surface structure. The color difference processing method based on global average ignores the difference of local heat conduction path, and it is difficult to accurately distinguish the internal and surface thermal evolution law in the curing monitoring of large volume concrete structure. The existing system lacks a dynamic compensation mechanism for the aging effect of the coating, and the measurement accuracy continues to decline due to the attenuation of the material performance in the long-term monitoring process. In addition, the traditional method does not establish a correlation model between chroma change and thermal physical process, and cannot effectively separate the superimposed influence of conduction heat and radiation heat in the complex heat source coexistence scene. SUMMARY

[0005] The purpose of the present application is to solve the problems existing in the prior art and to provide a structure surface temperature test system based on photosensitive coating color difference identification.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme: a structure surface temperature test system based on photosensitive coating color difference identification, comprising:

[0007] The color difference feature extraction module is used to acquire the pixel chroma distribution data of the coating surface through the imaging device, calculate the adjacent pixel brightness gradient by using the sliding window difference method, generate the chroma gradient vector, mark the color difference response feature points meeting the threshold, and transfer them to the heat conduction trajectory analysis module;

[0008] The heat conduction trajectory analysis module is used to perform time series analysis on the chroma gradient vector, calculate the consistency of the color difference response feature point cluster motion trajectory by using the vector field clustering algorithm, generate the main heat conduction direction field, and calculate the pixel chroma distribution data change amount per unit time to generate the chroma change rate field, and transfer them to the temperature gradient reconstruction module;

[0009] The temperature gradient reconstruction module is used to construct a spatial correction coordinate system based on the main heat conduction direction field, perform directional weighted fusion on the chroma change rate field by using an adaptive weight distribution algorithm, and calculate the temperature gradient distribution tensor according to the chroma-temperature mapping relationship Generate the temperature gradient distribution tensor, and transfer the abnormal region coordinates of the temperature gradient distribution tensor to the coating parameter optimization module.

[0010] As a further scheme of the present application, in the chroma-temperature mapping relationship, represents the brightness value of the CIELab* color space, represents the temperature measurement value, is a quadratic term coefficient, is a linear term coefficient, is a reference brightness compensation value;

[0011] The color difference response feature points are specifically gradient amplitude, spatial coordinates, and response time stamp, the main heat conduction direction field includes direction vector matrix, cluster center trajectory, and heat flow density distribution The color difference change rate field specifically refers to rate gradient tensor, time differential parameter, and abnormal fluctuation area, and the temperature gradient distribution tensor includes anisotropy coefficient, heat conduction rate matrix, and temperature mutation threshold;

[0012] In the heat flow density distribution calculation, represents the thermal conductivity of the material, represents the temperature gradient vector.

[0013] As a further scheme of the application, the sliding window difference method specifically performs the following process: a 3*3 pixel detection window is set, the brightness gradient of the center pixel and adjacent pixels in the CIELab* color space is calculated When and the hue angle changes , the feature point is marked as an effective color difference response feature point;

[0014] The vector field clustering algorithm adopts dynamic time warping (DTW) to calculate the consistency coefficient of feature point cluster motion trajectory Wherein represents the coordinate vector of the i-th feature point in the k-th time frame, is a velocity standard deviation threshold, and when , it is determined as a homologous heat conduction trajectory;

[0015] The maximum detection speed of the system is obtained through a calibration experiment, and the experimental method is as follows: the heat source moving speed is increased by 0.1 mm / s on the surface of a standard test piece, and when the color difference response feature point recognition rate is less than 95%, the critical speed value is recorded.

[0016] As a further scheme of the application, the time differential parameter is specifically second color difference sampling interval, and the rate gradient tensor calculation adopts a central difference method: , the color difference change acceleration is calculated synchronously, When the absolute value of the acceleration exceeds , it is marked as an abnormal fluctuation area;

[0017] The relevance of the abnormal fluctuation area determination and the temperature mutation is verified by establishing a partial differential equation , wherein is the density of the material, and cp is the specific heat capacity, for thermal conductivity, for heat source term.

[0018] As a further scheme of the present application, the method for constructing the space correction coordinate system is as follows: taking the maximum vector direction of the main heat conduction direction field as the X' axis, and establishing a rotation coordinate system with the deviation angle of the direction field;

[0019] The adaptive weight distribution algorithm adopts an exponential decay function , wherein represents the difference between the current pixel chrominance change rate and the neighborhood average rate, is the rate standard deviation;

[0020] The neighborhood average rate is calculated by using a 3*3 pixel window mean filtering algorithm.

[0021] As a further scheme of the present application, the method for determining the temperature mutation threshold is as follows: the standard deviation of the historical temperature gradient distribution tensor is calculated, and is set as the abnormal temperature mutation threshold, and when the heat conduction rate matrix value of the detection area exceeds , an early warning is triggered.

[0022] As a further scheme of the present application, the generation logic of the coating thickness control instruction set comprises: according to the ratio of the abnormal area temperature gradient amplitude to the coating reference thermal resistance , the thickness adjustment amount is calculated, wherein is the material heat conduction compensation coefficient;

[0023] The material heat conduction compensation coefficient is obtained through a thermal resistance experiment calibration curve, and the experimental method is as follows: under a reference temperature gradient, the thermal resistance change rate of a coating with different thickness is measured, and the slope of the linear regression curve is taken as the K value.

[0024] As a further scheme of the present application, the system further comprises:

[0025] A coating parameter optimization module is configured to analyze the abnormal area coordinates, match a photosensitive material thermal response characteristic database, execute a thickness control strategy logic judgment, and generate a coating thickness control instruction set.

[0026] The coating thickness control instruction set specifically comprises a thickness adjustment value, a material thermal resistance parameter, and a spraying path coordinate corresponding to a B-spline curve control point coordinate.

[0027] As a further scheme of the present application, the method for generating the spraying path coordinates comprises: projecting the abnormal area coordinates to a three-dimensional curved surface model reconstructed by point cloud data acquired through laser scanning, generating a continuous spraying path by using B-spline curve interpolation, and controlling the point spacing , the radius of curvature .

[0028] As a further scheme of the present application, the method for matching the material thermal resistance parameter is: selecting a high-resistance material with a thermal resistance value of when the anisotropy coefficient in the temperature gradient distribution tensor is , and selecting a conventional material with a thermal resistance value of when the anisotropy coefficient is .

[0029] The anisotropy coefficient is determined through a thermal conduction comparison test, and the test method is: setting heat sources in different directions on the surface of a standard test piece, and measuring the critical mutation point of the ratio of the thermal conduction rate in each direction.

[0030] Compared with the prior art, the present application has the following advantages and positive effects:

[0031] In the present application, the spatiotemporal evolution characteristics of the chrominance field are captured by dynamic gradient analysis technology, which significantly improves the identification sensitivity of the non-uniform thermal field and still maintains stable temperature resolution under complex lighting conditions. The cluster motion pattern recognition based on the thermal conduction trajectory effectively analyzes the propagation path of the transient thermal shock and shortens the system response delay. The nonlinear temperature-sensitive characteristic compensation mechanism is adopted to overcome the measurement deviation caused by coating process differences, and higher precision temperature reconstruction is achieved in curved surface structure detection. The combination of spatial correction and directional weighting technology enhances the spatial analysis capability of the temperature gradient field, and accurately identifies the micro heat affected zone. The processing logic significantly reduces the temperature field reconstruction error in large-scale structure monitoring through multi-dimensional coupling analysis of the chrominance field and the thermal conduction path, and has better engineering applicability than traditional methods. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 The present application is a structure surface temperature test system based on light-sensitive coating color difference recognition, and the overall flowchart is shown in the figure. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical scheme and advantages of the present application clearer, the technical scheme realized by software will be described in detail below in combination with the system architecture diagram and the embodiments. It should be understood that the specific embodiments described herein are only used to explain the technical scheme of the present application, and do not constitute a limitation on the scope of protection.

[0034] ​In the description of the present application, the system architecture relationship or data processing flow indicated by the terms "hierarchy", "module", "interface", "data flow", "client", "server" and the like are defined based on the corresponding architecture diagram or flowchart of the embodiments. This way of expression is only used to clearly explain the logical relationship of each element in the technical solution, and is not limited to the physical deployment form. The "multiple" contains two or more technical units, including but not limited to multiple data nodes, processing threads, service instances or functional components, and other scalable elements. The specific number is determined according to the actual business scenario.

[0035] Please refer to Figure 1 The present application provides a technical solution: a structure surface temperature test system based on color difference recognition of photosensitive paint includes:

[0036] The color difference feature extraction module acquires the pixel chroma distribution data of the coating surface through the imaging device. Specifically, for example, on a set of gas turbine blade high temperature test platform, the blade surface is pre-sprayed with LCR Hallcrest TLCR75C10W type photosensitive temperature change paint. A Basler ace ac A1920-155um type industrial camera is used, matched with a Computar M2514-MP2 type megapixel level prime lens (focal length 25mm, aperture F1.4), and the camera is connected with the image acquisition card through a gigabit Ethernet interface. The camera lens is vertically aligned with the blade to be tested area, and the working distance is set to 300mm to ensure that the imaging field of view covers the key parts of the blade, such as the blade leading edge or the pressure surface. The camera exposure time is set to 500 microseconds, the gain is set to 0dB, and the frame rate is set to 20 frames / second. During the blade heat shock experiment, such as simulating the ignition or shutdown working condition, the camera continuously acquires image sequences of the coating surface. After each frame of image is captured, the image data (such as 2048x1080 pixel resolution) is transmitted to the computer memory in the original Bayer format. Then, the original image data is corrected for white balance, and the correction parameters are obtained by calibrating the standard color plate under the test environment lighting. Then, the RGB (red, green, blue) three-channel pixel values after white balance correction are converted into CIELab color space (lightness), (red / green axis), (yellow / blue axis) three components. For example, the RGB value of a certain pixel point is (210, 120, 80), and through the standard RGB to CIELab conversion formula (based on D65 standard light source), the corresponding value is 75.3, value is 30.2, value is 40.5. The , , The values ​​together constitute the pixel chromaticity distribution data of the coating surface at that moment.

[0037] After obtaining the chromaticity distribution data, the sliding window difference method is used to calculate the brightness gradient of adjacent pixels. Set a 3×3 pixel detection window, which is located in the entire Slide pixel by pixel on the image. For each 3×3 area covered by the window, take the center pixel as the reference and calculate the distance between it and the surrounding 8 adjacent pixels. For example, the value of the center pixel P0 of the current detection window is The value is , its right adjacent pixel P1 The value is Then the brightness gradient component between P0 and P1 is Similarly, calculate the brightness gradient of P0 and the other 7 adjacent pixels. These gradient values ​​together represent the intensity and direction of the brightness change near the central pixel P0. For example, if the central pixel , the right pixel , then the absolute value of the brightness difference between the two is .

[0038] Then, a chromaticity gradient vector is generated. The chromaticity gradient vector here is not a traditional mathematical gradient vector. , but refers to the comprehensive consideration of the characteristics of brightness gradient and hue angle change within the detection window. Specifically, for the center pixel of the detection window, in addition to calculating the brightness gradient between it and the adjacent pixels In addition, the hue angle between it and the adjacent pixels needs to be calculated Changes For example, the hue angle of the central pixel P0 is , the hue angle of its adjacent pixel P1 is , then the hue angle changes This group consists of and The constructed features are used for subsequent feature point marking.

[0039] Next, mark the color difference response feature points that meet the threshold. For each central pixel of the 3×3 detection window, calculate the brightness gradient based on its brightness gradient with the 8 adjacent pixels. and hue angle changes Make a judgment. When the brightness gradient between the central pixel and any of its adjacent pixels And the corresponding hue angle change between these two pixels When , the central pixel is marked as a valid color difference response feature point. These two threshold parameters and The setting is based on the following:

[0040] Luminance gradient threshold :

[0041] Reference: In CIELab* color space, (Total color difference) usually takes 2.3 as a reference value for a perceptible difference (Just Noticeable Difference, JND). , but the principle is similar: the human eye can only stably perceive when the brightness change reaches a certain level. This threshold is designed to filter out visually discernible coating brightness changes caused by temperature changes, while excluding background noise or weak, non-significant color fluctuations.

[0042] Experimental verification process:

[0043] Experimental materials: The same photosensitive coating as used in actual applications (such as LCRHalcrestTLCR75C10W) was evenly sprayed on 10 standard nickel-based high-temperature alloy specimens (size 50mm×50mm×2mm), and the coating thickness was controlled at 50±5μm.

[0044] Experimental equipment: precision temperature controlled heating table (temperature control accuracy ±0.1 ), the aforementioned imaging system (Basler camera and Computar lens), and a standard D65 light source illumination box.

[0045] Experimental steps: 1. Place the specimen in the lighting box and use the imaging system to focus on the specimen surface. 2. Set the initial reference temperature, for example, 750 , record the pixel points on the coating surface at this time 3. Take 0.2 The step length is used to slowly increase or decrease the temperature of the heating stage. After each step length change, it is stabilized for 30 seconds. The image is collected and the new value of each pixel is calculated. value, get 4. At the same time, invite 5 observers who have been trained in color recognition to judge whether there is a perceptible brightness difference between the current coating color and the coating color at the reference temperature under uniform observation conditions. Record the temperature difference when the observer first perceives the difference and the corresponding average 5. Repeat steps 3 and 4 to cover the main color change temperature range of the paint.

[0046] Experimental data and results: Table 1: Summary of threshold calibration experimental data;

[0047] ;

[0048] As shown in Table 1, based on multiple experimental data, When the value reaches about 2.3, the observer can stably identify the brightness change. Considering the robustness of the system application, 2.3 is selected as judgment threshold.

[0049] Hue angle change threshold :

[0050] Reference: Changes in hue angle directly reflect changes in the color's hue, such as from orange to yellow. A change in hue angle typically indicates a hue shift that is discernible to the naked eye, helping to distinguish true material color changes from light and dark changes caused solely by factors such as uneven lighting.

[0051] Experimental verification process:

[0052] Experimental materials and equipment Threshold calibration experiment.

[0053] Experimental steps:

[0054] same Calibrate experimental steps 1 and 2.

[0055] With 0.2 As the step size, slowly change the temperature of the heating stage, collect images and calculate the pixel 、 Value, and then calculate the hue angle and its variation .

[0056] Ask the observer to judge whether there is a perceptible difference in "color" rather than just "lightness" between the current coating color and the coating color at the reference temperature. Record the temperature difference when the observer first perceives the difference in hue and the corresponding average value.

[0057] Experimental data and results: Table 2: Summary of threshold calibration experimental data;

[0058] ;

[0059] As shown in Table 2, the experiment shows that the average The hue angle change can be stably perceived as a change in hue. As judgment threshold.

[0060] Example: Assume that in a 3×3 detection window, the chromaticity value of the central pixel P0 is The adjacent pixel P1 on the right has a chromaticity value of . Calculate the difference between P0 and P1 : Calculate the hue angle of P0 : (Use the inverse tangent function and convert the result into an angle.) Calculate the hue angle of P1 : Calculate the hue angle change between P0 and P1 : . Judgment: Due to and , so the center pixel P0 (or its boundary with P1) is marked as a valid color difference response feature point. The coordinates of these marked feature points (e.g., (x, y) pixel coordinates) and the corresponding marking time (frame number) are recorded and passed to the subsequent heat conduction trajectory analysis module. For example, if the coordinates of P0 are (1024, 530) and the current image is the 100th frame, the data (1024, 530, 100) is output as the feature point information.

[0061] See also Figure 1 The heat conduction trajectory analysis module receives the color difference response feature point data passed by the color difference feature extraction module. This data typically includes the feature point's 2D coordinates (x, y) in the image and the time frame number t in which it was identified. For example, analyzing 100 consecutive image frames (each frame is 0.05 seconds apart, for a total of 5 seconds of data), thousands of such feature points may be detected.

[0062] First, the color difference response feature point data is analyzed in time series. This process associates the feature points of the same physical thermal event at different time frames to form a motion trajectory. The set of feature points identified in the frame image ,in , the system will try to match it with the Feature point set in the frame image Matching is based on spatial proximity and motion continuity. For example, if In the Frame, then in Find the most likely corresponding point within a preset search radius of the frame (e.g. 5 pixels) Through this frame-by-frame matching, a time series of feature points, namely the trajectory, can be obtained. ,in It's a trajectory In the The coordinate vector of the time frame (if the trajectory exists in this frame).

[0063] Then, the vector field clustering algorithm is used to calculate the consistency of the motion trajectory of the color difference response feature point cluster. Here, a variant of the dynamic time warping (DTW) algorithm is used to measure the consistency of the two trajectories. and The formula for calculating the consistency coefficient is: Parameter details:

[0064] :Indicates the The trajectory and The consistency coefficient between the two trajectories. Its value range is [0,1]. The closer the value is to 1, the more similar or consistent the two trajectories are.

[0065] : Indicates the number of time frames that two trajectories have in common when compared. For example, if the trajectory Exists in frames 1 to 10, trajectory exists in frames 3 to 12, and they all need to be compared on these frames, then is the number of frames where they overlap, which is 8 frames here (from frame 3 to frame 10). More precisely, it is the number of point pairs on the DTW alignment path. But the formula here is more like a direct frame-by-frame comparison, so is the number of frames in the common time period of the two trajectories.

[0066] : The index of the summation symbol, representing the sequence number of the time frame, from 1 to .

[0067] :Indicates the The track in coordinate vectors on a common time frame. For example, , that is, the pixel coordinates of the feature point in the image.

[0068] :Indicates the The track in coordinate vectors on a common time frame, .

[0069] :Indicates that On the time frame, the trajectory Feature points and trajectories The square of the Euclidean distance between the feature points. This term calculates the difference between the spatial positions of two points.

[0070] : velocity standard deviation threshold, which actually plays a role of a spatial scale parameter (distance unit) in this formula to normalize the square of distance. Its value is .

[0071] : maximum detectable velocity of the system, which is obtained through calibration experiments.

[0072] : the acquisition process: 1. Experimental preparation: uniformly spray a standard test piece (for example, Inconel718 alloy, 100mm x 100mm x 3mm) with a photosensitive coating consistent with actual application. Set up a point heat source (for example, a semiconductor laser with a power of 20W, spot diameter 1mm) that can accurately control the moving speed. The imaging system (as described above) is directly opposite the test piece surface. 2. Experimental execution: the point heat source moves at an initial speed on the test piece surface along a straight line. The imaging system acquires images at a fixed frame rate (for example, 20fps). The chromatic aberration feature extraction module identifies the chromatic aberration response feature points generated by the heat source moving track. Record the number of successfully identified feature points, and compare it with the number of feature points that should be generated in theory (estimated based on the length of the heat source path and the feature point density), and calculate the recognition rate. 3. Velocity increment: increase the moving speed of the heat source by 0.1mm / s, repeat step 2. 4. Critical velocity determination: when the recognition rate of chromatic aberration response feature points is first lower than 95%, record the moving speed of the heat source at this time as . Then is set to this or slightly lower than it.

[0073] Experimental data example: Table 3: Calibration experiment data | heat source moving speed (mm / s);

[0074] ;

[0075] As shown in Table 3, when the moving speed of the heat source is 1.5mm / s, the recognition rate drops to 94.6%, which is lower than 95%. Therefore, the critical velocity value is 1.4mm / s. So set .

[0076] Value calculation and setting: . Here is in mm / s. However, the in the formula is the square of distance (units such as pixel² or mm²). Therefore, should be a distance unit. If refers to the maximum moving speed of the feature points allowed on the image, then can be understood as the distance between two time frame intervals a measure of the allowed spatial deviation between points on homologous trajectories. Assuming the pixel size of the imaging system is (mm / pixel) and the camera frame rate is (fps), then . The maximum distance a feature point can move in one frame interval is then . At this time, can be set to . For example, . If the camera frame rate is , then . . Then . If the camera field of view is 100 mm corresponding to 2000 pixels, then 1 pixel is about 0.05 mm. Then . Therefore, is set to pixels in practice. (pixel²). (pixel²).

[0077] Operational logic of the formula: for two trajectories and , at each pair of time frames in which they coexist, the squared spatial distance between their corresponding feature points is calculated. This squared distance is converted into a similarity value between 0 and 1 through a Gaussian function (where ): the smaller the distance, the closer the value to 1; the larger the distance, the closer the value to 0. controls the sensitivity of this conversion, a smaller makes the similarity more sensitive to distance changes. Finally, the similarity values at all coexisting time frames are averaged to obtain the overall consistency coefficient .

[0078] Example: suppose there are two trajectories and , which have feature points in consecutive 3 time frames ( ). pixels, pixel².

[0079] Time frame 1: pixels, pixels;

[0080] pixel²;

[0081] ;

[0082] Time frame 2: pixels, pixels;

[0083] pixel2;

[0084] ;

[0085] Time frame 3: pixels, pixels;

[0086] pixel2;

[0087] ;

[0088] .

[0089] The advantage of the formula is that, by the combination of exponential decay function and distance square, the weight of near distance points is much larger than that of far distance points, while The introduction of parameter allows to adjust the sensitivity of the decision according to the maximum detection speed of the system (reflecting the typical motion scale of feature points), so as to more robustly identify trajectory clusters with similar motion patterns.

[0090] When , it is determined as homologous heat conduction trajectory.

[0091] The setting of threshold value 0.85 is based on:

[0092] Reference basis: This threshold value is usually selected by experiments in the case of known multiple independent heat sources and single extended heat source, analyzing the distribution between homologous trajectories and distribution between heterologous trajectories, and selecting a value that can better distinguish the two.

[0093] Experimental verification process: 1. Scene one (homologous): use an area type heat source (such as a 5mm x 5mm square heater) or a slowly moving point heat source to generate a large number of closely related feature point trajectories. Calculate the value between each pair of trajectories. 2. Scene two (heterologous): use two or more spatially separated and independently controlled point heat sources, which generate feature point trajectories belonging to different sources. Calculate the value between trajectories from different heat sources. 3. Data analysis: collect a large number of values of scene one and scene two. For example, collect 1000 pairs of trajectories values.

[0094] For scene one (homologous), The value is expected to be high, for example, most of the distribution is between 0.8 and 0.99.

[0095] For scenario two (heterogeneous), The value is expected to be low, for example, most of the distribution is between 0.4 and 0.7.4. Threshold selection: select a threshold so that the two classes of trajectories can be maximally distinguished. For example, if the value of more than 95% is greater than 0.85, and the value of more than 95% is less than 0.85, then 0.85 is a reasonable threshold.

[0096] Setting the value to 0.85 indicates that the two trajectories are required to have a high degree of spatial proximity on average over their common time frame to be considered homologous.

[0097] In the above example, . Because , the trajectories and are not determined to be homologous heat conduction trajectories in this example.

[0098] By calculating for all trajectory pairs and applying a threshold judgment, the trajectories can be divided into different clusters. Each cluster represents an independent heat conduction event or region. Subsequently, based on these clustering results, a main heat conduction direction field is generated. For each cluster, the average displacement vector (i.e. velocity vector) of all trajectories within the cluster at each time frame is calculated. These average velocity vector fields are smoothed in space and time (e.g. Gaussian filtering or mean filtering), forming a continuous main heat conduction direction field. This field is represented by a two-dimensional vector at each spatial location (or grid cell), indicating the main direction and relative rate of heat propagation at that location.

[0099] The color change rate field is generated by synchronously calculating the change in pixel chrominance distribution data per unit time. This process calculates the change in value of each pixel (or its neighborhood) over time. The time differentiation parameter is specifically seconds of chrominance sampling interval.

[0100] The setting of

[0101] Reference: The selection of this sampling interval needs to balance the system's response ability to fast heat events and the burden of data processing. The response time (time required for color change to stabilize) of photosensitive paint is usually tens to hundreds of milliseconds. For typical industrial monitoring scenarios, the temperature change rate can be several degrees Celsius per second to several hundred degrees Celsius per second.

[0102] Experimental verification process: 1. Select a typical photosensitive coating whose response curve under specific temperature excitation is known ( 2. Apply step temperature excitation or fast linear temperature rise excitation, use a high frame rate camera (e.g. 100-200fps, i.e. = 0.01s or 0.005s) record 3. Downsample the benchmark data to simulate different (such as 0.02s, 0.05s, 0.1s, 0.2s). 4. Compare and use different Calculated Compared with the calculation using high-resolution benchmark data 5. Choose one , which can fully capture the expected fastest Changing features (e.g., peak rate, changing trend) without causing data redundancy and a surge in computational complexity due to oversampling.

[0103] For example, for a coating with a response time of about 50ms, if you want to capture at least 5 data points of its change process, the sampling interval needs to be However, if you are interested in more macroscopic temperature trends (such as events lasting several seconds), a sampling interval of 0.1 seconds (10fps) or 0.05 seconds (20fps) is usually sufficient. seconds, that is, the camera frame rate is 20 frames / second.

[0104] The rate gradient tensor is calculated using the central difference method: .

[0105] Formula parameter description:

[0106] :Indicates the time When the pixel brightness The first derivative with respect to time, The rate of change.

[0107] : In time The first sampling moment after Plus a sampling interval )’s pixel brightness value.

[0108] : In time The first sampling moment before Subtract one sampling interval )’s pixel brightness value.

[0109] : The time span used in the central difference method.

[0110] Operation logic: This formula takes the current moment The two sampling points before and after The difference between the two sampling points is divided by the time interval between the two sampling points. , to estimate The instantaneous rate of change at a given moment. This is a numerical differentiation method with second-order accuracy.

[0111] Example: Assume Seconds, at a certain pixel, Second;

[0112] ;

[0113] ;

[0114] ;

[0115] this That is, the pixel at 1.00 seconds The rate at which the value changes; constitutes an element of the chromaChangeRate field.

[0116] Synchronously calculate the chromaticity change acceleration The calculation formula is usually: ;

[0117] Formula parameter description:

[0118] :Indicates the time When the pixel brightness The second derivative with respect to time, The change of acceleration.

[0119] : At the current moment The pixel brightness value.

[0120] , , :Same as above.

[0121] Operational logic: This formula compares A moment and two moments before and after The relationship between the value of the acceleration is used to estimate the acceleration. The second-order central difference of the value series is performed.

[0122] Example: Continue using seconds, at the same pixel, Second;

[0123] (Add the current time value); (To show non-zero acceleration, we adjust The value of ), .

[0124] When the absolute value of acceleration exceeds The time is marked as abnormal fluctuation area.

[0125] Acceleration threshold The setting basis is:

[0126] Reference: This threshold is used to identify The value of the area where the value changes sharply and rapidly may indicate a very rapid temperature shock or potential measurement data anomaly (such as light source flickering, image artifacts caused by severe vibration). This value should be much larger than the value caused by normal heat conduction process. acceleration.

[0127] Experimental verification process: 1. Scenario 1 (normal / slow thermal change): Apply slow and uniform heating or cooling to the coated specimen and record Data and calculation . Usually these values ​​are small. 2. Scenario 2 (rapid thermal shock): Use pulsed laser or rapid contact with high / low temperature objects to apply rapid thermal shock to the specimen locally and record Data and calculation . These values ​​will increase significantly. 3. Scenario 3 (interference): Introduce known interference during the imaging process, such as periodic flickering of the light source, slight vibration of the camera, etc., and record Data and calculation 4. Data analysis: Under normal circumstances, Maybe 0-500 within the range.

[0128] Under known interference, May reach 500-2000 .

[0129] Under real rapid thermal shocks of particular concern (e.g., critical heat flux where phase change or damage of the material may occur), Probably more than 3,000 5. Threshold selection: Choose a threshold high enough to filter out most operational noise and normal variations while still capturing truly severe thermal events. It is an empirical value determined based on such experiments, used to mark areas where the rate of change itself is also changing rapidly.

[0130] In the above example, the calculated acceleration value is Because , this pixel is not marked as an abnormal fluctuation region at this moment. If another pixel calculates the acceleration as , this point will be marked.

[0131] The relevance of abnormal fluctuation region determination and temperature jump is verified by establishing a partial differential equation .

[0132] Formula parameter description:

[0133] : Material density (kg / m³). For example, for the substrate Inconel718 under the coating, .

[0134] : Material specific heat capacity (J / (kg·K)). For example, Inconel718 near room temperature .

[0135] : Temperature (K or ).

[0136] : Time (s).

[0137] : The rate of change of temperature with time.

[0138] : Divergence operator.

[0139] : Material thermal conductivity (W / (m·K)). For example, Inconel718 near room temperature . For anisotropic materials, is a tensor.

[0140] : Temperature gradient vector, representing the rate and direction of temperature change in space.

[0141] : Internal heat source power per unit volume (W / m³). If there is no internal heat source, Q=0.

[0142] Operation logic and purpose: This is the heat conduction equation, which describes the spatiotemporal distribution of the temperature field in the unsteady heat conduction process. The left represents the rate of change of energy within the object over time (heat storage term), the first term on the right represents the net incoming heat flow caused by temperature difference (divergence form of Fourier's heat conduction law), and the second term represents the internal heat source.

[0143] Verification procedure: When a region is labeled as "abnormal fluctuation" by the chroma change acceleration analysis, it means that the value of this region experienced a drastic acceleration change in a short time, which might correspond to a sharp change in temperature . To confirm that this "abnormal fluctuation" is indeed caused by a real, physically reasonable "temperature jump", rather than measurement noise or non-ideal response of the paint itself, one can substitute the temperature information (obtained by -T relationship conversion, see subsequent module) of this region into this heat conduction equation. 1. Obtain the property parameters: consult or measure the values of the material under study (usually the substrate under the coating). These values can be functions of temperature. 2. Estimate the boundary conditions and heat sources: estimate the boundary conditions (e.g. heat flux density, convective heat transfer coefficient) and possible internal heat sources around this abnormal fluctuation region according to the experimental setup or actual working conditions. 3. Numerical solution: use numerical methods such as finite difference method (FDM), finite element method (FEM) or finite volume method (FVM) to simulate the temperature field evolution in this region. Use the temperature field obtained by inversion as the initial condition or reference. 4. Result comparison: compare the temperature change rate and temperature acceleration (if needed) obtained by numerical simulation with the phenomenon indirectly observed by change. If a region is labeled as abnormal fluctuation (corresponding to high ), and its corresponding value is also large, and this can be explained by the heat conduction equation under reasonable property parameters and boundary conditions, then this abnormal fluctuation is confirmed as a real temperature jump. Conversely, if the equation cannot explain such a drastic temperature change (e.g. requires unrealistic heat flux density or thermal conductivity), then the abnormal fluctuation may be due to other non-thermal factors.

[0144] The benefit of this formula is that it provides a physical law-based verification mechanism that can link the observed drastic color change to the thermal physical properties of the material itself, thereby enhancing the reliability of the temperature jump event determination and distinguishing between real rapid thermal phenomena and potential measurement artifacts.

[0145] Example: Suppose in a certain abnormal fluctuation region (e.g. 1 mm x 1 mm), the temperature is inverted from to in s by change. Then the average . The substrate is Inconel718: , (high temperature value), (high temperature value). This means that to achieve such a heating rate, the net input power per volume (from conduction or internal heat sources) needs to reach . Next, analyze . If the surface of this region is subjected to a strong external heat flow impact (e.g. laser heating), this heat flow will be converted into term (if it is volume absorption) or affect through boundary conditions. By estimating the maximum heat flux density that can actually be achieved (e.g. ), and considering the heat penetration depth, it can be determined whether matches in energy level. If not, the authenticity of this "abnormal fluctuation" is questionable.

[0146] The heat conduction trajectory analysis module finally generates the main heat conduction direction field and the calculated color change rate field data, which are passed to the temperature gradient reconstruction module for further processing.

[0147] Please refer to Figure 1 , the temperature gradient reconstruction module, receives the main heat conduction direction field data and the color change rate field data passed by the heat conduction trajectory analysis module. For example, the main heat conduction direction field is represented in the form of a vector matrix, where is the pixel resolution or the number of grid units divided in the imaging area, and each vector indicates the main direction and relative intensity of heat propagation at position . The color change rate field is a scalar matrix, and each element represents the change rate of value at , such as the aforementioned calculated .

[0148] First, based on the main heat conduction direction field, a spatial correction coordinate system is constructed. The specific construction method is: analyze all direction vectors in the input main heat conduction direction field, and count the modulus and frequency of each direction vector. Select the direction vector with the largest modulus (or the most significant after weighted averaging) as the dominant direction. For example, in the analysis of a blade leading edge heat shock, it is found that most of the strong heat flow traces point in the blade spanwise direction, and the average direction forms an angle with the horizontal axis of the image. Then take this average direction ( ) as the direction of the new coordinate system X' axis. The original image coordinate system is XY, and the new spatial correction coordinate system is X'Y'. The conversion from XY to X'Y' is realized through a rotation matrix, and the rotation angle The condition "deviation angle of the direction field from the rotation coordinate system" here means that if there are multiple significant conduction directions and the included angle between them is less than , the average direction can be taken as the X' axis; if there are multiple dominant directions with a large included angle, regional processing or a more complex non-Cartesian coordinate system may be needed.

[0149] Here, it is assumed that there is a clear dominant direction with an included angle of with the original image coordinate system X axis. Then the formula for transforming any original coordinate to the new coordinate system is:

[0150] For example, the coordinates of a point P in the original image are (100, 50) pixels.

[0151] ;

[0152] pixels.

[0153] pixels.

[0154] The coordinates of point P in the new coordinate system are (107.16, 31.88) pixels.

[0155] Subsequently, an adaptive weight distribution algorithm is used to directionally weight and fuse the chroma rate of change field. The purpose of this step is to adjust the contribution of the chroma rate of change of each pixel according to the main direction of heat propagation, so as to more accurately reflect the temperature change along the heat flow direction. The weight calculation formula is . Parameter details:

[0156] : the weight assigned to the chroma rate of change of the current pixel, with a value range of (0, 1].

[0157] : the absolute value of the difference between the chroma rate of change of the current pixel and the average chroma rate of change of its neighborhood. The average rate of the neighborhood is calculated using a 3x3 pixel window mean filtering algorithm. For example, for pixel , the chroma rate of change is . The chroma rates of change of the 9 pixels in its 3x3 neighborhood (including itself) are , respectively. Then the average rate of the neighborhood is . .

[0158] : rate standard deviation, set to . Here ​Associated with the system maximum detection speed (the maximum moving speed of feature points on the image) in the heat conduction trajectory analysis module, but here applied to the chrominance change rate field. Need to convert (unit: mm / s or pixel / s) to the scale related to the chrominance change rate (unit: / s). A more reasonable explanation is, is the typical standard deviation of the chrominance change rate according to a large amount of experimental data, or an adjustment parameter set based on the value dynamic range and the expected maximum change rate. Assuming that through calibration experiments, the standard deviation of the value change rate observed in a typical rapid thermal event is about , then can be set. Or, if the original definition of is followed and an attempt is made to associate: if represents the movement of a feature point on the image per second pixels, and the movement of each pixel corresponds to change units (which depends on the temperature gradient and the slope of the -T curve), then can be . This association is more indirect. Here we assume that is independently calibrated according to the statistical properties of the chrominance change rate. Setting basis for :

[0159] Reference basis: This parameter controls the sensitivity of the weight function to the rate deviation. A smaller makes the weight more sensitive to changes in .

[0160] Experimental verification process: 1. Collect multiple sets of chrominance change rate field data under different thermal conditions. 2. Calculate the of each pixel and its 3x3 neighborhood average , obtaining the distribution of . 3. Statistically calculate the standard deviation of these values, denoted as . 4. Set as a multiple of , for example , where is an adjustment factor (for example, between 0.5 and 2), and is optimized by the quality of the subsequent fusion results. Or, according to the aforementioned , if it can be reasonably converted to the scale of the chrominance change rate, follow .

[0161] Assume that through experimental analysis, for the thermal event of interest, The rate of change is mainly distributed in The standard deviation of the average neighborhood difference is about Then set .

[0162] Example: A pixel The (computed from the previous module) is The rate of change of chroma (unit: ( ) / s) of its 3x3 neighborhood is:

[0163] ;

[0164] ;

[0165] ;

[0166] Set , The weighted rate of change of chroma for this pixel is This weighted fusion is not directly multiplying by , but using as the weight distribution when performing information fusion in the main heat conduction direction later. For example, when calculating the gradient along the main heat conduction direction X' axis, if the rate of change of chroma of a point is significantly different from its neighborhood difference (the is small), its contribution to the gradient calculation will be reduced. A more specific fusion method may be, in the corrected coordinate system X'Y', for each point, a weighted average rate is calculated along the X' direction (main heat conduction direction) using the values and corresponding weights of its neighboring points, or these weights are used when calculating the temperature gradient.

[0167] Next, generate the temperature gradient distribution tensor according to the chroma-temperature mapping relationship .

[0168] Formula parameter description:

[0169] : Luminance value in CIELab* color space.

[0170] : Temperature measurement value (unit: or K).

[0171] : Quadratic coefficient (unit: or ).

[0172] : Coefficient of linear term (unit: or ).

[0173] : Reference brightness compensation value (unit: ) representing the value of at , or the starting value of the calibration temperature zone.

[0174] Determination of the coefficient (calibration experiment):

[0175] Experimental materials: The same photosensitive coating as in actual application, uniformly sprayed on standard test pieces with known thermal physical properties (for example, similar in size and material to actual workpieces).

[0176] Experimental equipment: Precise programmable temperature-controlled heating table or oven (temperature control accuracy ), thermocouple (contact temperature measurement, accuracy ), the aforementioned imaging system, standard D65 light source.

[0177] Experimental steps: 1. Place the coated test piece on the heating table, and place the thermocouple close to the surface of the test piece (or embedded near the surface of the test piece) to obtain the true temperature. 2. Select multiple (for example, 10-20) stable temperature points within the expected working temperature range (for example to ). 3. At each temperature point, after the temperature stabilizes (for example, stable for 5 minutes), use the imaging system to collect the image of the coating surface, and calculate the average value of the selected area. At the same time, record the thermocouple reading . 4. Collect all pairs of data corresponding to the temperature points.

[0178] Data processing and coefficient calculation: Use the least squares method to fit the collected data pairs (where , is the number of calibration points) to a quadratic polynomial , thereby obtaining the optimal coefficient value.

[0179] Experimental data and examples: Table 4: - Example of calibration experiment data for the relationship between

[0180] ;

[0181] As shown in Table 4, 9 sets of ​Data. By fitting a quadratic polynomial to these data (for example, using the polyfit function in MATLAB or the trendline function in Excel), the coefficients can be obtained. Assume that the fitting result is: Then the chromaticity-temperature mapping relationship of the paint is: .

[0182] The benefit of the formula is that the quadratic polynomial can better describe the nonlinear response characteristics of many photosensitive coatings over a wide temperature range than a linear relationship, thereby improving the accuracy of temperature measurement.

[0183] Generate temperature gradient distribution tensor: 1. Temperature field Acquisition: For each pixel in the chromaticity change rate field after direction weighted fusion ,That The value is known (either from the original image or after time averaging). Substitute the value into the inverse solution -T relationship to obtain the temperature of the point .because It is a The quadratic equation , which is solved as follows: . Usually based on the characteristics of the coating (such as temperature increase, Lower) Select the appropriate root.

[0184] Example: A pixel , using the above coefficients: , here appears , indicating the coefficients of the fit or the given The value has no real solution under this model, or the coefficients I chose do not fit the example data points exactly. This emphasizes the importance of accurate calibration. Let's rescale the coefficients so that hour It has good performance near , and ensures that the discriminant is positive. Suppose that another set of coefficients is obtained through accurate calibration and fitting, for example: (Note: Photosensitive coatings Follow The behavior of the changes varies, and this is just to construct a solvable example. Follow Rising and falling, the parabola opening downward corresponds to )when ,but . Discriminant . It is still a negative number. This shows the sensitivity of the quadratic equation solution to the coefficients. In practical applications, the selection and fitting quality of the calibration curve are very important. To continue the example, we assume a and inverse function between has been obtained directly by table lookup or piecewise linear interpolation, or a polynomial that performs well within the calibration range is used. For example, suppose the inverse function is calculated directly as .

[0185] temperature gradient calculation: after obtaining the entire temperature field , the temperature gradient is calculated in the spatially corrected coordinate system X'Y'. . These partial derivatives can be calculated using central difference method: where and are the pixel or grid spacing in the corrected coordinate system. The temperature gradient distribution tensor actually means that at each point there is a temperature gradient vector .

[0186] The main heat conduction direction field also includes the direction vector matrix, the cluster center trajectory, the heat flux density distribution .

[0187] Parameter description:

[0188] : heat flux vector (W / m²), representing the rate and direction of heat transfer per unit area.

[0189] : material thermal conductivity (W / (m·K)). This is the thermal conductivity of the substrate material below the coating, or the equivalent thermal conductivity considering the influence of the coating. This value is usually a function of temperature and needs to be consulted from the material manual or measured experimentally. For example, for nickel-based superalloy at , .

[0190] : temperature gradient vector calculated as described above.

[0191] Operation logic: Fourier's law of heat conduction, which describes that in the process of steady-state or transient heat conduction, the heat flux density is proportional to the temperature gradient.

[0192] Example: the calculated temperature gradient at a point is (assuming that the pixel size has been converted to physical size). The material thermal conductivity . . . Then the heat flux vector at this point is . Its module length is .

[0193] The advantage of the formula is that it links the directly measurable temperature gradient to a key engineering thermophysical parameter, heat flux density, which enables quantitative assessment of the thermal loading status of each point on the structure surface.

[0194] Finally, the coordinates of the abnormal area of the temperature gradient distribution tensor are transmitted to the coating parameter optimization module. The determination method of the temperature mutation threshold is: the standard deviation of the historical temperature gradient distribution tensor is counted . For example, under similar working conditions, the gradient of a large amount of collected temperature field data is calculated, and then the standard deviation of these gradient values (such as the gradient module ) is counted. Assuming that the historical data shows that the mean value of is , and the standard deviation is , set . When the current temperature gradient module of a certain area is (or directly ), it is considered that the temperature gradient of the area is abnormal. More directly, if refers to the fluctuation of temperature itself rather than the gradient, then "the standard deviation of the historical temperature gradient distribution tensor " should be understood as the standard deviation of the historical temperature data itself, or more likely the standard deviation of the temperature change rate . Assuming that it refers to the temperature change rate . Through the relationship between the change rate and , the value of can be obtained. The standard deviation of the historical value is . For example, . Set the abnormal temperature mutation threshold to .

[0195] When the heat conduction rate matrix (i.e. field) of the detection area exceeds , an early warning is triggered.

[0196] The setting of the early warning threshold :

[0197] Reference basis: This threshold is usually set based on the heat shock resistance of the material, the safety operating window of the specific process, or the critical temperature rise rate at which the structure may be damaged.

[0198] Experimental verification process: 1. Perform thermal cycle or thermal shock experiments on materials / components identical to those used in actual applications. 2. Gradually increase the temperature rise / cooling rate ​3. At each rate, check for micro-cracks, yielding, coating peeling, or other forms of damage. 4. Record the critical temperature rate at which unacceptable damage or performance degradation begins to occur. 5. Pre-warning threshold Should be set below the critical rate with a certain safety margin (e.g. 70-80% of the critical rate).

[0199] Example of experimental data: for a certain turbine blade material, it is found that micro-cracks begin to appear on the leading edge of the blade when the temperature rate reaches and lasts for a certain period of time. To ensure safety, the pre-warning threshold is set to .

[0200] For example, a certain region calculates . Due to , the region triggers a warning and records its coordinates (e.g. the center pixel coordinates and the region range). The list of coordinates of these abnormal regions is passed to the coating parameter optimization module.

[0201] See Figure 1 , Coating Parameter Optimization Module, receives the coordinates of the abnormal regions passed by the Temperature Gradient Reconstruction Module. For example, the input data can be a list of elements, each containing the identification of an abnormal region, the center point coordinates, the approximate range of the region, or the set of contour points. As .

[0202] First, analyze the coordinates of the abnormal regions. This process is to convert the input coordinate information into the data structure required for internal processing of the module. For example, if the input is the pixel coordinates of the region, it needs to be converted into physical size coordinates (if the subsequent spraying path, etc. requires physical units). If the camera calibration parameters are known (e.g. the conversion factor of mm / pixel), then .

[0203] Then, match the light-sensitive material thermal response characteristic database. This database stores the curve parameters (such as the aforementioned ) of different light-sensitive materials, thermal conductivity, specific heat, density, response time and stability at different temperatures, adhesion to different substrates, and corresponding thermal resistance characteristics, etc. For each abnormal region parsed, according to the temperature range (inverted from ), temperature change rate, etc. information, query the detailed characteristics of the most suitable or currently used coating material in the database. This step is mainly to confirm the known parameters of the current coating, providing a basis for subsequent thickness regulation.

[0204] Next, the thickness control strategy logic is executed. The core of this logic is to determine the thickness of the abnormal area according to its temperature characteristics (such as the temperature gradient amplitude). or temperature change rate ) and the performance parameters of the coating itself (such as the benchmark thermal resistance ) to compare and determine if and how to adjust the coating thickness.

[0205] The coating thickness control instruction set specifically includes the thickness adjustment value, material thermal resistance parameters, and the spray path coordinates corresponding to the B-spline curve control point coordinates. The calculation logic is: .

[0206] Formula parameter description:

[0207] : Coating thickness adjustment amount (mm). A positive value indicates an increase in thickness, a negative value indicates a decrease in thickness (if the process allows, otherwise it usually refers to the difference between the target thickness when re-spraying and the current thickness).

[0208] :Material thermal conductivity compensation coefficient (mm·s / or mm·m²·K / (W·s) if The unit of this coefficient should be the same as and The units match so that The unit of is length. The original unit is .

[0209] : A value representing the temperature characteristics of the abnormal area. It can be the temperature gradient amplitude (unit ), or the rate of temperature change (unit ), or simply a temperature difference (unit ). From the context of "the temperature gradient amplitude in the abnormal area "Look, here Should refer to the temperature gradient magnitude or a temperature difference associated with the gradient. is the rate of temperature change, then Unit is appropriate, the result is (mm).

[0210] :Coating base thermal resistance (m²·K / W). This is a standard unit of thermal resistance. Not a standard thermal resistance unit, but some dimensionless or related The units of the parameters are then matched, and The units of the parameters are then matched, and is a simplified parameter, whose units are compensated by , so that the units of can be directly derived from (mm). For example, if is ( ), and is a reference rate value (e.g., the maximum acceptable rate ), then the units of are mm. More formally, is the thermal resistance. The thermal resistance of a coating , where is the thickness, is the thermal conductivity of the coating.

[0211] Material thermal conduction compensation factor acquisition:

[0212] Reference: The value reflects the sensitivity of the coating thickness change to the suppression of temperature change (or temperature gradient).

[0213] Experimental verification process (thermal resistance experiment calibration curve acquisition): 1. Experimental materials: select the target photosensitive coating, prepare a series of different thickness of coating samples (for example, the thickness from 20 μm to 100 μm, the step is 10 μm) on the standard substrate. 2. Experimental equipment: heat flow meter (used to measure the heat flux through the sample), precision temperature control heat source and cold plate, thermocouple. 3. Experimental steps: a. For each different thickness of the sample, apply a known, constant heat flow or maintain a constant temperature difference (the temperature difference between the upper and lower surfaces of the coating). b. Measure the temperature response under steady-state or specific transient conditions. For example, under the reference temperature gradient (or temperature change rate), measure the thermal resistance of different thickness of the coating. c. More in line with the units of in the formula , it may be measured under a specific "temperature driving force" (e.g., the difference between the substrate temperature and the ambient temperature), how different coating thickness affects the temperature change rate of the substrate surface or the final temperature gradient . d. Assuming that it is measured under different thickness , to maintain the surface temperature at some target value (relative to some reference temperature There is a temperature difference ) some "effort" required or some "effect" produced. e. The description "measure the rate of change of thermal resistance of the coating of different thicknesses, take the slope of the linear regression curve as K": could be a plot of vs or similar indirect relationship.

[0214] One possible calibration: apply a standard rapid thermal shock to the back of the substrate, measure the rate of change of the maximum temperature of the surface of the substrate under the coating on the front as a function of coating thickness . Plot vs . If this relationship is approximately linear, the slope can be taken as a part of

[0215] Given that has units of , has units of (assuming that is the equivalent temperature difference caused by the temperature difference or gradient), if it is a dimensionless factor or a quantity related to time, then can be made to have units of mm.

[0216] More specific experiment: test coupons of different coating thicknesses under a standard thermal shock (e.g. heat source power , time of action ). Record the maximum temperature rise produced on the surface of each coupon. If it is desired to reduce the temperature rise by increasing the thickness, and represents a benchmark, acceptable temperature rise (or quantity related to temperature rise), then may be .

[0217] Given , we accept this value.

[0218] Example: suppose that an analysis of the temperature gradient in some anomalous region yields an equivalent temperature driving force (e.g. this is a quantity that incorporates the effects of both the magnitude and duration of the gradient). The benchmark thermal resistance of the coating is . The units of may not be the standard thermal resistance units . If is , is , to get​ (mm), then the unit needs to be s (second). This indicates may represent a reference response time or thermal penetration time. Assume s (this is a reference thermal characteristic time constant). This means that the coating thickness needs to be increased by 1.0 mm. This value is too large, as the thermal barrier coating thickness is usually in the order of microns to hundreds of microns. We revisit the unit and . If is the temperature change rate, the unit . For example (an early warning value from the foregoing). Then . If , then the unit is mm. At this time must be dimensionless. Assume (a dimensionless normalization factor). . Still too large.

[0219] Determination of anisotropy coefficient (thermal conduction contrast test): 1. Experimental material: the same substrate as the actual structure, the surface is sprayed with the coating material to be evaluated. 2. Experimental equipment: two or more independently controllable point heat sources (such as lasers), imaging system. 3. Experimental steps: a. Apply heat source in one direction (e.g. X-axis) on the surface of the test piece, measure the heat propagation rate along the X-axis and Y-axis (or the long-short axis ratio of the isotherm ellipse formed). b. Change the direction of heat source application (e.g. Y-axis) and repeat the measurement. c. Anisotropy coefficient is defined as the ratio of the thermal conductivity of the material in different directions. For example, , where and are the effective thermal conductivities in two orthogonal directions. Or more simply, by observing the shape of the heat spot expansion, if the heat spot expands significantly faster in a certain direction than in other directions, it is considered that there is anisotropy. d. "Critical mutation point of measured thermal conduction rate ratio" may refer to the critical point at which the material changes from isotropic to anisotropic, or the critical point at which the degree of anisotropy changes significantly, when changing certain parameters (such as temperature, stress). Here we assume is a directly measured ratio.

[0220] Example: In the test, it is found that the effective thermal conductivity of heat along the main conduction direction (e.g. rolling direction) is , while the thermal conductivity perpendicular to this direction is . Then the anisotropy coefficient .

[0221] Material selection logic:

[0222] When , select high thermal resistance material with thermal resistance value .

[0223] When , select regular material with thermal resistance value .

[0224] Threshold value setting basis:

[0225] Reference basis: 1.2 is generally considered an empirical value to distinguish between slight anisotropy and significant anisotropy. When the ratio exceeds 1.2, it means that the thermal conductivity of the material in one direction is at least 20% higher than in another direction.

[0226] Experimental verification: Through thermal performance testing and structural reliability analysis on a series of materials with different degrees of anisotropy, a value is determined, beyond which the influence of anisotropy on structural thermal management or thermal stress distribution becomes non-negligible, and a more uniform thermal conductivity coating material (i.e., better isotropy or higher thermal resistance in a specific direction) needs to be used to compensate or adjust.

[0227] Thermal resistance threshold and setting basis:

[0228] Reference basis: These values represent the thermal resistance range of typical high thermal resistance and regular thermal resistance coating materials.

[0229] Experimental verification: According to application requirements (e.g., maximum allowed operating temperature, desired thermal protection level, etc.), through heat transfer calculation and experiments, determine the required thermal resistance range. For example, for components that require strong thermal insulation, select materials with a range. For components that allow certain heat transfer or require rapid heat dissipation, select materials with a range. These values usually correspond to the performance parameters of actual selectable materials in the material library.

[0230] In the example, . Because , the system will filter out high thermal resistance light-sensitive coatings with thermal resistance value from the database as candidates.

[0231] Generation of spraying path coordinates: 1. Project abnormal area coordinates to the three-dimensional surface model reconstructed by laser scanning to obtain point cloud data.

[0232] Acquiring 3D model: Pre-scan the actual workpiece (e.g. turbine blade) using a high-precision 3D laser scanner (e.g. Creaform HandySCAN BLACK Elite, accuracy up to 0.025mm) to obtain dense point cloud data of its surface. The point cloud data contains the 3D coordinates of millions of points .

[0233] Model reconstruction: Process the point cloud data using professional point cloud processing software (e.g. Geomagic DesignX or PolyWorks Modeler), including denoising, registration (if multiple scans), meshing, and surface reconstruction, to finally generate an accurate 3D CAD model of the workpiece (e.g. saved in.STL,.STEP, or.IGES format).

[0234] Coordinate projection: Identify the pixel coordinates of the abnormal area in the two-dimensional image Need to project back to the surface of the three-dimensional model through the camera calibration parameters (internal and external parameter matrices) and the pose of the workpiece in the camera coordinate system, to get the corresponding three-dimensional space coordinates . This process involves the conversion from 2D image points to 3D space points. For example, if the abnormal area is a rectangle, its four corner point pixel coordinates are converted to four corresponding points on the three-dimensional model.

[0235] Generate continuous spraying path using B-spline curve interpolation.

[0236] For the abnormal area contour points or internal filling point set projected onto the three-dimensional surface, use B-spline curve / surface algorithm for interpolation or fitting to generate smooth spraying tool center point (TCP) path.

[0237] Control point spacing : The B-spline curve is defined by a series of control points. The distance between control points of adjacent path segments (or the spacing between raster scan lines) on the generated spraying path is not greater than 2mm.

[0238] The setting of this parameter depends on: depends on the spraying width of the spraying equipment, the uniformity requirement of the coating, and the size of the abnormal area. Smaller spacing can improve the uniformity and accuracy of spraying coverage, but will increase the spraying time and data volume.

[0239] Experimental verification: By spraying different control point spacing paths and detecting the formed coating thickness uniformity and coverage integrity, select a spacing value that can meet the quality requirements and has high efficiency. 2mm is usually suitable for repair or local reinforcement spraying that requires high precision.

[0240] Curvature radius : The generated B-spline spraying path has a local curvature radius of not less than 5mm at any position.

[0241] The parameter setting is based on: the kinematics of the spraying robot or actuator, and the requirements of the spraying process itself (e.g. avoiding paint accumulation or too thin at sharp turns). Too small a radius of curvature can cause the robot to move unevenly or exceed its acceleration / jerk limits.

[0242] Experimental verification: Test the motion stability and spraying quality of the spraying equipment under different path curvatures. Record the curvature radius at which spraying defects (such as splashing, sagging, uneven thickness) or abnormal robot motion begin to appear, then set a minimum curvature radius greater than the critical value with a safety margin. 5mm is a common requirement for small complex curved surface spraying paths.

[0243] Example: For a circular abnormal area with a diameter of 10mm projected onto the surface of a blade, the B-spline path planning algorithm generates a series of spiral or reciprocating path segments to ensure complete coverage of the area. For example, the first control point of the path has a three-dimensional coordinate of mm, and the second control point is mm. This list of points constitutes part of the spraying instructions.

[0244] Finally, the generated coating thickness control instruction set (including target area, calculated thickness adjustment amount , selected material thermal resistance parameter (or material type), and detailed B-spline spraying path three-dimensional coordinate sequence) is output to guide the subsequent coating repair or preparation equipment (such as an automated spraying robot) to perform operations. For example, the instructions may be: "Area Identifier: Area_XYZ, Target Thickness Increment: +0.05mm, Selected Material: HT-Coating-TypeA (Thermal Resistance ), Spraying Path Point Sequence: ".

[0245] The above examples demonstrate the preferred embodiments of the invention, and any equivalent adjustments to the technical solutions based on software engineering methods are within the scope of protection, including but not limited to: implementing algorithm logic in different programming languages, service-oriented reconstruction of functional modules, adjusting data interaction protocols, optimizing resource scheduling strategies, etc. Any implementation derived from reasonable modifications to the data processing flow, service call link, or system architecture level without deviating from the technical core of the invention should be considered within the protection scope defined by the claims of the invention.

Claims

1. A structural surface temperature measurement system based on color difference recognition of photosensitive coatings, characterized in that: The system comprises: The color difference feature extraction module is used to obtain the color distribution data of the pixels on the coating surface through the imaging device, calculate the brightness gradient of adjacent pixels using the sliding window difference method, generate the color gradient vector, mark the color difference response feature points that meet the threshold, and pass it to the heat conduction trajectory analysis module; A heat conduction trajectory analysis module is used to perform time-series analysis on the chromaticity gradient vector, calculate the consistency of the motion trajectory of the color difference response feature point cluster using a vector field clustering algorithm, generate a main heat conduction direction field, and simultaneously calculate the change in pixel chromaticity distribution data per unit time to generate a chromaticity change rate field, which is then passed to the temperature gradient reconstruction module; The temperature gradient reconstruction module is used to construct a spatial correction coordinate system based on the main heat conduction direction field, and use an adaptive weight distribution algorithm to perform direction weighted fusion on the chromaticity change rate field. generating a temperature gradient distribution tensor, extracting the coordinates of abnormal regions of the temperature gradient distribution tensor and transferring them to a coating parameter optimization module; In the chromaticity-temperature mapping relationship, Indicates the lightness value in the CIELab* color space, Indicates the temperature measurement value, is the coefficient of the quadratic term, is the coefficient of the first-order term, is the base brightness compensation value.

2. The structure surface temperature measurement system based on photosensitive coating color difference recognition according to claim 1 is characterized in that: The color difference response feature points are specifically gradient amplitude, spatial coordinates and response timestamps, and the main heat conduction direction field includes direction vector matrix, cluster center trajectory and heat flux density distribution. The chromaticity change rate field specifically refers to the rate gradient tensor, time differential parameter and abnormal fluctuation area, and the temperature gradient distribution tensor includes anisotropy coefficient, heat conduction rate matrix and temperature mutation threshold; In the heat flux distribution calculation, represents the thermal conductivity of the material, represents the temperature gradient vector.

3. The structure surface temperature measurement system based on photosensitive coating color difference recognition according to claim 2 is characterized in that: The specific implementation process of the sliding window difference method is as follows: set a 3×3 pixel detection window, calculate the brightness gradient between the center pixel of the window and the adjacent pixels in the CIELab* color space ,when And the hue angle changes When , it is marked as a valid color difference response feature point; The vector field clustering algorithm uses dynamic time warping (DTW) to calculate the consistency coefficient of the feature point cluster motion trajectory ,in Represents the coordinate vector of the i-th feature point in the k-th time frame, is the speed standard deviation threshold, when When it is determined to be a homologous heat conduction trajectory, represents the trajectory consistency coefficient between the i-th feature point and the j-th feature point, Indicates the total number of time frames, Represents the coordinate vector of the j-th feature point in the k-th time frame, represents the speed standard deviation threshold, Indicates the maximum detection speed of the system; The maximum detection speed of the system The critical speed value is obtained through calibration experiments. The experimental method is: the heat source moving speed is increased in steps of 0.1 mm / s on the surface of the standard specimen, and the critical speed value is recorded when the recognition rate of the color difference response feature point is lower than 95%.

4. The structure surface temperature measurement system based on photosensitive coating color difference recognition according to claim 3 is characterized in that: The time differential parameter is specifically: The chroma sampling interval is 1 second, and the rate gradient tensor is calculated using the central difference method: , synchronously calculate the chromaticity change acceleration , when the absolute value of acceleration exceeds When is marked as abnormal fluctuation area, represents the rate of change of the brightness component over time, Indicates the chroma sampling time interval, represents the brightness value at time t+Δt, represents the brightness value at time t-Δt, The second derivative representing the change in brightness; The correlation between the abnormal fluctuation area determination and temperature mutation is established by establishing a partial differential equation Verify, where is the material density, is the specific heat capacity, is the thermal conductivity, is the heat source term.

5. The structure surface temperature testing system based on photosensitive coating color difference recognition according to claim 4 is characterized in that: The method for constructing the spatial correction coordinate system is as follows: taking the maximum vector direction of the main heat conduction direction field as the X' axis, establishing the deviation angle with the direction field The rotating coordinate system of The adaptive weight allocation algorithm adopts an exponential decay function ,in, represents the adaptive weight value, Indicates the difference between the current pixel chromaticity change rate and the neighborhood average rate, represents the standard deviation of the rate, Indicates the maximum detection speed of the system; The neighborhood average rate is calculated using a mean filtering algorithm with a 3×3 pixel window.

6. The structure surface temperature measurement system based on photosensitive coating color difference recognition according to claim 5 is characterized in that: The method for determining the temperature mutation threshold is: statistically analyzing the standard deviation of the historical temperature gradient distribution tensor ,set up is the abnormal temperature mutation threshold. When the thermal conductivity matrix value of the detection area exceeds When the warning is triggered, represents the standard deviation of the historical temperature gradient distribution tensor, Indicates the deviation of the current temperature gradient from the historical average.

7. The structure surface temperature measurement system based on photosensitive coating color difference recognition according to claim 1 is characterized in that: The coating parameter optimization module is used to analyze the coordinates of the abnormal area, match the photosensitive material thermal response characteristic database, execute the thickness control strategy logic judgment, and generate a coating thickness control instruction set; The coating thickness control instruction set specifically includes a thickness adjustment value, a material thermal resistance parameter, and spraying path coordinates corresponding to the coordinates of the B-spline curve control points.

8. The structure surface temperature measurement system based on photosensitive coating color difference recognition according to claim 7 is characterized in that: The generation logic of the coating thickness control instruction set includes: According to the temperature gradient amplitude of the abnormal area Thermal resistance of coating benchmark Calculate the thickness adjustment ,in, Indicates the coating thickness adjustment amount, represents the temperature gradient amplitude in the abnormal area, Indicates the benchmark thermal resistance value of the coating, Indicates the thermal conductivity compensation coefficient of the material; The thermal conductivity compensation coefficient of the material is obtained through a thermal resistance experimental calibration curve. The experimental method is: measuring the thermal resistance change rate of coatings of different thicknesses under a reference temperature gradient, and taking the slope of the linear regression curve as the K value.

9. The structure surface temperature measurement system based on photosensitive coating color difference recognition according to claim 7, characterized in that: The method for generating the spraying path coordinates includes: projecting the abnormal area coordinates onto a three-dimensional surface model reconstructed by acquiring point cloud data through laser scanning, using B-spline curve interpolation to generate a continuous spraying path, and controlling the point spacing. , radius of curvature , where the control point spacing represents the straight-line distance between adjacent spray path control points, and the curvature radius represents the minimum curvature radius of the curved section of the spray path.

10. The structure surface temperature measurement system based on photosensitive coating color difference recognition according to claim 9, characterized in that: The matching method of the material thermal resistance parameters is: according to the anisotropy coefficient in the temperature gradient distribution tensor ,when Select thermal resistance value when high-resistance material, when Select thermal resistance value when Conventional materials, among which It represents the anisotropy coefficient in the temperature gradient distribution tensor, which is defined as the ratio of the maximum heat conduction rate to the minimum heat conduction rate; The anisotropy coefficient is determined by a heat conduction comparison test, wherein the test method is as follows: heat sources in different directions are set on the surface of a standard specimen, and a critical mutation point of the ratio of the anisotropic heat conduction rate is measured.

Images