Method for inverting thermal radiation spatial data based on microfacet theory

Abstract

This invention discloses a method for inverting spatial thermal radiation data based on micro-surface element theory, comprising the following steps: S1, acquiring BRDF data from multiple angles, and simultaneously acquiring measured directional emissivity data of the target rough surface; obtaining initial data; S2, fitting the initial data from S1 to the initial physical emissivity model based on micro-surface element theory; S3, applying the measured directional emissivity data of the target rough surface to the initial physical emissivity model for fitting. If the model does not meet the standard, the model parameters are automatically adjusted using optimization algorithms such as backpropagation, and iterative fitting is performed until the error meets a preset threshold; if the standard is met, the current model parameter set and structure are locked. This invention offers high precision and strong physicality; by using a small amount of measured emissivity data to correct and fine-tune the model, high-precision target emissivity information can be quickly obtained under different roughness conditions.

Description

Technical Field

[0001] This invention belongs to the field of thermal radiation physics, specifically relating to a method for inverting spatial data of thermal radiation based on micro-surface element theory. Background Technology

[0002] The bidirectional reflectance distribution function describes the emissivity in the outgoing direction after a given incident irradiance. It is an essential tool for describing the reflective properties of materials and rendering them correctly.

[0003] like Figure 1 As shown, microsurface theory posits that a macroscopically rough surface is composed of many optically smooth micro-surfaces in different directions, each with a different emissivity in the same direction. Electromagnetic waves emitted from a randomly rough surface may be blocked by adjacent micro-surfaces. Similarly, electromagnetic waves reflected from these micro-surfaces may also be blocked. This means that both the emission and reflection of electromagnetic waves from each micro-surface can be blocked. The process of an electromagnetic wave emitting from a micro-surface can be categorized as leaving the surface directly, leaving the surface after one reflection, leaving the surface after two reflections, and so on. Since each element has a different slope, their emissivity in the same direction is also different, and the infrared emissivity at the observation angle is equal to the sum of these emissivity components.

[0004] The thermal radiation characteristics of a material are important parameters for evaluating its energy exchange behavior in high-temperature environments, and are typically characterized by indicators such as emissivity, radiative intensity, and spatial radiation distribution. Traditional thermal radiation measurements are mostly based on directly observing the radiative intensity in a certain direction or within a finite solid angle, thereby calculating the material's directional emissivity. However, this method has many shortcomings in practical applications, such as: lack of spatial information: conventional measurement devices usually only support single-angle or small-range angular scanning, making it difficult to cover all radiation directions within the material's hemispherical space, resulting in the inability to accurately obtain the complete spatial radiation distribution; and data error accumulation: the limited field of view and optical path structure of the detector introduce angular averaging effects, especially in materials with high zenith angles or strong anisotropy, where measurement deviations are significantly amplified, thus affecting the reliability of the inversion.

[0005] Existing methods for estimating thermal radiation characteristics based on models or empirical formulas also have significant limitations. For example, for materials with micro / nano structures, periodic structures, or significant roughness, their thermal radiation behavior is highly dependent on the incident-outgoing direction relationship, and traditional macroscopic models cannot accurately reflect the scattering and radiation mechanisms of microstructures. In addition, under high temperature, dynamic changes, or complex external field environments, the radiation characteristics of materials will change significantly with temperature, angle, and wavelength, and existing methods cannot achieve real-time and accurate spatial data inversion. Summary of the Invention

[0006] The purpose of this invention is to provide a method for inverting spatial thermal radiation data based on micro-surface element theory, which uses micro-surface element theory to realize the inversion of radiation characteristics in the entire space.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] The method for inverting spatial thermal radiation data based on micro-surface element theory includes the following steps:

[0009] S1. Collect BRDF data from multiple angles, and simultaneously collect measured data on the directional emissivity of the rough surface of the target; denoise the collected BRDF data from multiple angles to obtain initial data; lay the data foundation for subsequent fitting and parameter extraction;

[0010] S2. Perform data fitting on the initial data in S1 to extract key surface scattering characteristic parameters: the correlation parameters of the normal distribution function De(a) and the geometric attenuation function G(ω); based on the above parameters, combine the initial physical emissivity model of the micro-surface element theory;

[0011] S3. The measured directional emissivity data of the rough surface of the target are collected and applied to the initial physical emissivity model for fitting. The fitting error is judged based on the fitting residual. If it does not meet the standard, the model parameters are automatically adjusted through optimization algorithms such as backpropagation, and iterative fitting is performed until the error meets the preset threshold. If it meets the standard, the current model parameter set and structure are locked.

[0012] S4. Repeat steps S1-S3 to build models for targets with different roughnesses; form a model library, input the target roughness into the model, and you can obtain the target's thermal radiation spatial data.

[0013] Preferably, the specific method for acquiring BRDF data from multiple angles in step (1) is as follows:

[0014] An incident zenith angle of 30° was selected, a reflected zenith angle of 0-85°, an incident azimuth angle of 0°, and a reflected azimuth angle of 0-360°. The near-center region was spaced 1° apart, and the far-center region was spaced 10° apart. The near-center region was defined as ±5° of the specular reflection direction from the incident zenith angle, and the others were considered far-center regions. BRDF data of the material surface was acquired using an optical measurement system.

[0015] The method for collecting the directional emissivity is as follows: using an infrared radiometer or a Fourier transform infrared spectrometer, the measured emissivity data of the same rough surface in different directions are collected.

[0016] Preferably, the specific method of step (2) is as follows:

[0017] Based on the Cook-Torrance micro-element theory, using the initial BRDF data obtained in step S1, a nonlinear least-squares fit is performed on the surface reflection model; the fitted model is:

[0018] (1)

[0019] Where, k d and k s Here, denoted by , D(m; α) represents the diffuse reflection coefficient and the specular reflection coefficient, respectively; D(m; α) is the micro-element normal distribution function; F is the Fresnel reflection function; G² is the bidirectional geometric occlusion function; α is the surface roughness; and n is the refractive index of the material. These are the unit vectors for the incident and exit directions, respectively. Let θ be the angle between the emission direction and the normal of the micro-element. i , θ r These are the incident and exit angles, respectively;

[0020] The fitting process outputs a set of optimal physical parameters P. BRDF ={k d , k s , α, n, κ,}, and the defined normal distribution function D and geometric occlusion function G; these parameters and functions fully describe the microscopic geometry and basic optical properties of the surface, and are the physical basis for subsequent emissivity modeling; the optimal set of physical parameters is the set of physical parameters when the predicted and measured RMSE is less than 0.02;

[0021] Based on the above parameters and combined with the initial physical emissivity model of the micro-surface element theory, the model includes a direct emission term model and a multiple reflection term model; the direct emission term model is as follows:

[0022] (2)

[0023] in, Let F(θ) be the directional emissivity of the micro-element in its local coordinate system. local (n) is a Fresnel function; The emission angle is the result of multiple reflections; it is calculated from the material optical constants extracted above. G1 is a one-way occlusion function derived from G2, which represents the visibility probability of a micro-element in the observation direction. The ratio of projected areas; the integral is performed over the normal directions m of all possible micro-surface elements, and the sphere surface integral d m It can be represented as:

[0024] (3)

[0025] , These are the zenith angle and azimuth angle of the micro-surface element, respectively.

[0026] The model for multiple reflection terms is:

[0027] (4)

[0028] Among them, f r It is a micro-element BRDF model; The angle at which the radiation exits after being blocked.

[0029] The total directional emission rate is:

[0030] (5)

[0031] Preferably, the specific method in S3 is as follows:

[0032] The predicted curve obtained from the initial physical emissivity model constructed in S2 is compared with the measured directional emissivity data in S1 in the same direction, and the root mean square error (RMSE) and coefficient of determination (R²) are calculated. If the fitting error does not reach the preset threshold, a backpropagation parameter tuning mechanism is adopted to backpropagate the prediction error to the physical parameters of the initial physical emissivity model. Gradient descent or genetic algorithm is used to fine-tune the parameters to reduce the prediction deviation until the fitting error between the model prediction and the measured data meets the requirements, and the current optimal set of physical parameters and model structure are locked. The preset threshold RMSE < 0.02.

[0033] Preferably, the spatial data of thermal radiation in S4 includes directional emissivity parameters and functions, 2D / 3D distribution maps, and fitting error reports.

[0034] This invention overcomes the limitations of traditional single-angle observation by introducing micro-surface element theory, incorporating complex surface roughness and shading effects into the model construction process. The core lies in using a small amount of measured emissivity data to fine-tune the model, thereby rapidly obtaining high-precision target emissivity information under different roughness conditions. This invention solves the parameter uncertainty problem caused by complex geometric structures in thermal radiation inversion, and also has the following beneficial effects:

[0035] High accuracy: By introducing multiple reflection terms and closed-loop optimization based on measured data, the inversion accuracy of directional emissivity of rough surfaces is significantly improved, especially at high angles close to tangential observations.

[0036] Strong physicality: The model is based on rigorous micro-surface element thermal radiation physics, the parameters have clear physical meanings, and the extrapolation of the prediction results is better than that of purely empirical models.

[0037] Adaptive and automated: Integrating the entire process from data preprocessing, model fitting, validation to optimization, it lowers the professional threshold and improves modeling efficiency.

[0038] Visual and intuitive: 2D / 3D visualization capabilities make complex spatial distribution characteristics clear at a glance, greatly assisting scientific analysis and engineering decision-making. Attached Figure Description

[0039] Figure 1 Images of microscopic and macroscopic surfaces;

[0040] Figure 2 Flowchart of this invention;

[0041] Figure 3 The effect of random rough surface morphology on infrared emissivity;

[0042] Figure 4 BRDF measurement results and fitting results of this invention;

[0043] Figure 5 Emissivity measurement results and 2D fitting results of this invention;

[0044] Figure 6 3D hemispherical emissivity fitting results. Detailed Implementation

[0045] The present invention will be further described below with reference to embodiments, but the scope of protection of the present invention is not limited thereto.

[0046] like Figure 2 As shown, the method for inverting spatial thermal radiation data based on micro-surface element theory includes the following steps:

[0047] S1. Collect BRDF data from multiple angles, selecting an incident zenith angle of 30°, a reflected zenith angle of 0-85°, an incident azimuth angle of 0°, and a reflected azimuth angle of 0-360°. The near-center region is spaced 1° apart, and the far-center region is spaced 10° apart. The near-center region is defined as ±5° of the specular reflection direction from the incident zenith angle, and the others are far-center regions. Obtain BRDF data of the material surface using an optical measurement system.

[0048] Simultaneously, measured data of the directional emissivity of the target rough surface are collected using an infrared radiometer or a Fourier transform infrared spectrometer, and the influence of surface roughness and morphology on infrared emissivity is studied as follows: Figure 3 As shown;

[0049] The collected BRDF data from multiple angles were denoised to obtain initial data, laying the data foundation for subsequent fitting and parameter extraction.

[0050] S2. Based on the Cook-Torrance micro-element theory, the initial data in S1 is fitted to extract key surface scattering characteristic parameters: the correlation parameters between the normal distribution function De(a) and the geometric attenuation function G(ω); specifically, the fitting model is as follows:

[0051] (1)

[0052] Where, k d and k s Here, denoted by , D(m; α) represents the diffuse reflection coefficient and the specular reflection coefficient, respectively; D(m; α) is the micro-element normal distribution function; F is the Fresnel reflection function; G² is the bidirectional geometric occlusion function; α is the surface roughness; and n is the refractive index of the material. These are the unit vectors for the incident and exit directions, respectively. Let θ be the angle between the emission direction and the normal of the micro-element. i , θ r These are the incident and exit angles, respectively;

[0053] The fitting process outputs a set of optimal physical parameters P. BRDF ={k d , k s , α, n}, and the defined normal distribution function D and geometric occlusion function G; these parameters and functions fully describe the microscopic geometry and basic optical properties of the surface, and are the physical basis for subsequent emissivity modeling;

[0054] Based on the above parameters, and combined with the initial physical emissivity model of the micro-surface element theory; the model includes a direct emission term model and a multiple reflection term model; the direct emission term model is as follows:

[0055] (2)

[0056] in, The directional emissivity of the micro-element in its local coordinate system is calculated from the material optical constants extracted above. G1 is a one-way occlusion function derived from G2, which represents the visibility probability of a micro-element in the observation direction. The ratio of projected areas; the integral is performed over the normal directions m of all possible micro-surface elements, and the sphere surface integral d m It can be represented as:

[0057] (3)

[0058] , These are the zenith angle and azimuth angle of the micro-surface element, respectively.

[0059] The model for multiple reflection terms is:

[0060] (4)

[0061] Among them, f r It is a micro-element BRDF model; The angle at which the radiation exits after being blocked.

[0062] The total directional emission rate is:

[0063] (5)

[0064] S3. Apply the measured directional emissivity data of the target rough surface to the initial physical emissivity model in the same direction for comparison and fitting, and calculate the root mean square error (RMSE) and coefficient of determination (R²). 2 If the fitting error does not reach the preset threshold, a backpropagation parameter tuning mechanism is adopted to backpropagate the prediction error to the physical parameters of the initial physical emissivity model. Gradient descent or genetic algorithm is used to fine-tune the parameters to reduce the prediction deviation until the fitting error between the model prediction and the measured data meets the requirements, and the current optimal set of physical parameters and model structure are locked; the preset threshold RMSE < 0.02;

[0065] S4. Repeat steps S1-S3 to build models for targets with different roughnesses; form a model library, input the roughness of the target to be measured into the model, and the thermal radiation spatial data of the target to be measured can be obtained. The thermal radiation spatial data includes directional emissivity parameters and functions, 2D / 3D distribution maps, and fitting error reports; the BRDF measurement results and fitting results are as follows: Figure 4 As shown, the emissivity measurement results and 2D fitting results are as follows: Figure 5 As shown, the hemispherical emissivity fitting results are as follows: Figure 6 As shown, according to Figure 4 , Figure 5 , Figure 6 The results show that the thermal radiation spatial data inversion results of this invention are basically consistent with the actual measurement results, and the inversion accuracy is high.

Claims

1. A method for inverting spatial thermal radiation data based on micro-surface element theory, characterized in that, Includes the following steps: S1. Collect BRDF data from multiple angles, and simultaneously collect measured data of directional emissivity of the rough surface of the target; perform noise reduction processing on the collected BRDF data from multiple angles to obtain the initial data; S2. Fit the initial data in S1 to extract the relevant parameters of the normal distribution function De(a) and the geometric decay function G(ω); based on the above parameters, combine the initial physical emissivity model of the micro-surface element theory; S3. The measured directional emissivity data of the rough surface of the target are collected and applied to the initial physical emissivity model for fitting. The fitting error is judged based on the fitting residual. If it does not meet the standard, the model parameters are automatically adjusted through optimization algorithms such as backpropagation, and iterative fitting is performed until the error meets the preset threshold. If the criteria are met, lock the current model parameter set and structure; S4. Repeat steps S1-S3 to build models for targets with different roughnesses; form a model library, input the target roughness into the model, and you can obtain the target's thermal radiation spatial data.

2. The thermal radiation spatial data inversion method based on micro-surface element theory according to claim 1, characterized in that, The specific method for collecting BRDF data from multiple angles in step (1) is as follows: The incident zenith angle was set to 30°, the reflected zenith angle to be 0-85°, the incident azimuth angle to be 0°, and the reflected azimuth angle to be 0-360°. The near-center region was spaced at 1° intervals, and the far-center region was spaced at 10° intervals. BRDF data of the material surface was obtained through an optical measurement system. The near-center region was defined as ±5° of the specular reflection direction from the incident zenith angle, and the others were defined as the far-center region. The method for collecting the directional emissivity is as follows: using an infrared radiometer or a Fourier transform infrared spectrometer, the measured emissivity data of the same rough surface in different directions are collected.

3. The thermal radiation spatial data inversion method based on micro-surface element theory according to claim 1, characterized in that, The specific method for step (2) is as follows: Based on the Cook-Torrance micro-element theory, using the initial BRDF data obtained in step S1, a nonlinear least-squares fit is performed on the surface reflection model; the fitted model is: (1) Where, k d and k s Here, denoted by , D(m; α) represents the diffuse reflection coefficient and the specular reflection coefficient, respectively; D(m; α) is the micro-element normal distribution function; F is the Fresnel reflection function; G² is the bidirectional geometric occlusion function; α is the surface roughness; and n is the refractive index of the material. These are the unit vectors for the incident and exit directions, respectively. Let θ be the angle between the emission direction and the normal of the micro-element. i Let θ be the incident angle. r The launch angle; The fitting process outputs a set of optimal physical parameters P. BRDF ={k d , k s , α, n}, and a defined normal distribution function D and geometric occlusion function G; the optimal set of physical parameters is the set of physical parameters when the prediction and measurement RMSE is less than 0.02; Based on the above parameters and combined with the initial physical emissivity model of the micro-surface element theory, the model includes a direct emission term model and a multiple reflection term model; the direct emission term model is as follows: (2) in, Let F(θ) be the directional emissivity of the micro-element in its local coordinate system. local (n) is a Fresnel function; The emission angle is the result of multiple reflections; it is calculated from the material optical constants extracted above. G1 is a one-way occlusion function derived from G2; The ratio of projected areas; the integral is performed over the normal directions m of all possible micro-surface elements, and the sphere surface integral d m It can be represented as: (3) , These are the zenith angle and azimuth angle of the micro-surface element, respectively. The model for multiple reflection terms is: (4) Among them, f r For micro-facet BRDF models, The angle of radiation after being blocked The total directional emission rate is: (5)。 4. The thermal radiation spatial data inversion method based on micro-surface element theory according to claim 1, characterized in that, The specific method in S3 is as follows: The predicted curve obtained from the initial physical emissivity model constructed in S2 is compared with the measured directional emissivity data in S1 in the same direction, and the root mean square error and coefficient of determination are calculated. If the fitting error does not reach the preset threshold, a backpropagation parameter tuning mechanism is adopted to backpropagate the prediction error to the physical parameters of the initial physical emissivity model. Gradient descent or genetic algorithm is used to fine-tune the parameters to reduce the prediction deviation until the fitting error between the model prediction and the measured data meets the requirements, and the current optimal set of physical parameters and model structure are locked. The preset threshold RMSE < 0.

02.

5. The thermal radiation spatial data inversion method based on micro-surface element theory according to claim 1, characterized in that, The thermal radiation spatial data in S4 includes directional emissivity parameters and functions, 2D / 3D distribution maps, and fitting error reports.

Images