Industrial furnace wall temperature field data visualization method and system

Abstract

The application discloses an industrial furnace wall temperature field data visualization method and system, calculates industrial furnace wall temperature field data through thermocouple temperature measurement data and heat transfer theory, fills the temperature field data into a preset volume texture, adopts first light projection to down-sample the volume texture, obtains expected color values of the first projection light, adopts second light projection to sample the volume texture based on a preset second light projection termination condition, corrects mixed and accumulated color values after the second light projection sampling termination according to the expected color values of the first projection light, and obtains volume rendering image results through color rendering according to the corrected mixed and accumulated color values, so that the technical problem that the prior art cannot quickly perform high-quality data visualization on industrial furnace wall temperature field data is solved, fast and high-quality data visualization on industrial furnace wall temperature field data is realized, and a good balance between visualization speed and quality is achieved.

Description

Technical Field

[0001] This invention mainly relates to the field of visualization technology, specifically a method and system for visualizing temperature field data of industrial furnace walls. Background Technology

[0002] Temperature field data, as a crucial indicator for determining the normal operation of industrial furnaces and kilns, provides key feedback information for real-time control during long-term, efficient operation. Real-time visualization of industrial furnace wall temperature field data greatly assists on-site personnel in monitoring the operational status of industrial furnaces and kilns, ensuring their safety and stability. Furthermore, it allows for a more intuitive display of the latest temperature field data, increasing its value and utilization. Therefore, visualizing industrial furnace wall temperature field data is of great significance.

[0003] Currently, the commonly used method for visualizing temperature field data is to map two-dimensional temperature field data measured by thermocouples onto the surface of a two-dimensional or three-dimensional geometric model. This method can only visualize the surface temperature field data and cannot obtain the real-time temperature distribution inside the object. Alternatively, some methods use extensive theoretical derivations and interpolation calculations to map and remap temperature field data to obtain three-dimensional thermograms, but this method has significant errors and low visualization quality. Volume rendering technology is also a commonly used visualization method. Based on three-dimensional volume data, it simultaneously displays all volume details on a two-dimensional display device and is now widely used in many fields such as medical diagnosis, natural geographic exploration, and 3D reconstruction. Popular volume rendering methods include ray casting, shear-deformation algorithms, and snowballing algorithms. Among these, ray casting has a cumbersome calculation process, slow rendering speed, and generally poor volume texture filling effects, making it difficult to balance speed and quality in volume rendering.

[0004] To address the problems of large real-time interaction latency, low quality, and slow speed in current visualization methods, and considering the non-strict symmetry of the furnace wall structure in many industrial furnaces, this invention proposes a data visualization method for the temperature field of industrial furnace walls. This invention comprehensively considers the furnace wall structure and the arrangement of thermocouples used for temperature measurement, combining heat transfer theory and a method that integrates multiple influencing factors. It proposes a temperature correction model for calculating the inner surface temperature of the furnace wall based on thermocouple temperature measurement data, simulating the three-dimensional geometric structure of temperature correction, calculating the magnitudes of various influencing factors and correction parameters, and realizing the calculation of the inner surface temperature of the industrial furnace wall, providing boundary conditions for constructing the temperature field of the industrial furnace wall. Based on volume rendering technology, this invention proposes a secondary ray casting algorithm. The first ray casting aims to obtain the expected color value of the projected ray. Under the basic conditions of ray casting, this invention proposes a Russian roulette algorithm based on ray importance, setting ray termination conditions for the secondary ray casting, thus accelerating the volume rendering process. Simultaneously, the second ray projection calculates the mixed and accumulated color value, performs threshold judgment and color correction on it to reach or approach the expected color value, and finally obtains a high-fidelity temperature field data visualization result. To better evaluate the speed and quality of the volume rendering process, this invention proposes a balance score calculation method for evaluating the speed and quality of the volume rendering process. With sufficient adjustment of system parameters, it effectively balances the speed requirements of volume rendering with the image quality requirements.

[0005] The invention patent with publication number CN 113984233 A discloses a method and system for visualizing the temperature field of a three-dimensional furnace wall. This patent collects temperature field data of a three-dimensional furnace wall through thermocouples, and then generates a three-dimensional furnace wall temperature field through coordinate transformation, interpolation calculation, mapping and remapping. However, this patent first maps to generate a two-dimensional heat map through a large number of distance weight calculations and interpolation operations, and then remaps to a three-dimensional heat map, which has a large calculation error.

[0006] The invention patent with publication number CN 112231894 A discloses a method and system for visual analysis of temperature field inside a blast furnace. This patent designs a visualization system for temperature field inside a blast furnace through simulation and draws auxiliary lines and scatter plots to realize the annotation of the temperature field visualization map. However, the data volume of this patent is large, and its implementation process has high requirements for memory and processor, and the running speed is slow, which affects the improvement of visualization quality. Summary of the Invention

[0007] The present invention provides a method and system for visualizing industrial furnace wall temperature field data, which solves the technical problem that existing technologies cannot quickly perform high-quality data visualization of industrial furnace wall temperature field data.

[0008] To address the aforementioned technical problems, the present invention proposes a method for visualizing industrial furnace wall temperature field data, which includes:

[0009] Based on thermocouple temperature measurement data and heat transfer theory, temperature field data of industrial furnace walls are obtained.

[0010] Fill the industrial furnace wall temperature field data into the preset volume texture.

[0011] The volume texture is downsampled using the first ray projection to obtain the expected color value of the first projected ray.

[0012] Based on the preset second ray projection termination condition, the volume texture is sampled using the second ray projection.

[0013] The mixed and accumulated color value of the second projected ray after the second ray projection sampling process is terminated is corrected based on the expected color value of the first projected ray.

[0014] The volumetric rendering result is obtained by shading and rendering based on the mixed and accumulated color values ​​of the corrected second projected light rays.

[0015] Furthermore, based on thermocouple temperature measurement data and heat transfer theory, the temperature field data of the industrial furnace wall are obtained, including:

[0016] Thermocouples are arranged in a ring in the cooling wall of the industrial furnace wall structure.

[0017] Obtain the preset temperature correction point on the inner surface of the industrial furnace wall corresponding to the thermocouple.

[0018] Based on heat transfer theory, a temperature correction model is established to calculate the inner surface temperature of the furnace wall in industrial furnaces.

[0019] Based on the temperature correction model, the temperature data at the temperature correction points are iteratively corrected according to the sum of squared temperature differences at all thermocouple embedding points in each layer, thereby obtaining the temperature field data of the industrial furnace wall. The specific form of the temperature correction model is as follows:

[0020]

[0021]

[0022] Among them, T ij k and T ij k+1 Let be the temperature values ​​of the i-th temperature correction point after k and k+1 iterations, respectively. Let ΔT be the temperature correction amount for the i-th temperature correction point corresponding to the j-th thermocouple in the (k+1)-th iteration. ijLet be the temperature correction amount at the i-th temperature correction point, and U be the set of the 5 thermocouples closest to the i-th correction point, where U = [j, j_left, j_right, j_up, j_down]. iμ The temperature correction ΔT is the distance between the μ-th thermocouple (one of the five nearest thermocouples) and the i-th temperature correction point. ij The degree of influence parameter, r μ and e μ Let k be the distance and temperature difference between the μ-th thermocouple and the ith correction point, respectively, among the 5 nearest neighbor thermocouples. eμ The temperature difference is the correction amount for temperature ΔT ij The degree of influence parameter, F R () is the distance control function between the five nearest thermocouples and the i-th correction point.

[0023] Furthermore, filling the industrial furnace wall temperature field data into a preset volume texture includes:

[0024] Create the initial volume texture.

[0025] For each layer of the initial volume texture, temperature correction point data and thermocouple temperature measurement data are sequentially filled using the Bresenham rasterization algorithm to obtain the first volume texture.

[0026] The temperature data between the inner surface of the furnace wall and the cooling wall where the thermocouple is located is calculated by interpolation and then filled into the first volume texture to obtain a filled volume texture.

[0027] Furthermore, based on a preset second ray projection termination condition, sampling the volume texture using second ray projection includes:

[0028] Determine whether the cumulative Alpha channel value of the second projected ray in the second ray projection is less than a preset channel threshold. If so, terminate the sampling of the second projected ray according to the first termination condition.

[0029] If not, the sampling of the second projected light is terminated according to the second termination condition. At the same time, the sampling count, sampling point, and three-channel mixed-cumulative color record values ​​are recorded when the mixed-cumulative Alpha channel value of the second projected light reaches the preset channel threshold. The light importance is calculated based on the sampling count, the three-channel mixed-cumulative color record values, and the expected color values ​​of the three channels. If the light importance is determined to be less than the preset importance threshold, sampling continues until the sampling count reaches the preset first sampling count while keeping the color values ​​of the three channels unchanged. If the light importance is determined to be greater than the preset importance threshold, the color values ​​of the three channels are adjusted according to the Russian Roulette algorithm model based on the light importance.

[0030] Furthermore, the first termination condition is as follows:

[0031]

[0032] The second termination condition is as follows:

[0033]

[0034] Where, N RC The maximum number of samples is given, σ is the threshold weight of the alpha channel, and 0 < σ < 1, α T α is the threshold for the alpha channel. ∑ (k) represents the accumulated Alpha channel value of the k-th sample, shoot(l,SkyBox) is the function to determine whether ray l intersects with the object, and l.isHit is the result of the shoot(l,SkyBox) function. Let P be the coordinate point. k Spatial coordinates, x B y B z B Represents the coordinates of the bounding box boundary where the geometric carrier is located, and η is the judgment P. k Whether it is redundancy within the bounding box.

[0035] Furthermore, the specific formula for adjusting the color values ​​of the three channels based on the Russian Roulette algorithm model based on the importance of light is as follows:

[0036]

[0037] Significance = [|R ∑ (k σ )-R P |+|G ∑ (k σ )-G P |+|B ∑ (k σ )-B P |+β·k σ ] -1 ,

[0038] Where C(k) is the RGB color value of the kth sample, P is the preset model probability based on the ray importance Russian roulette algorithm, and N T To preset the first sampling number, δ S The preset importance threshold is defined, where Significance represents importance information, and R... ∑ (k σ ), G ∑ (k σ ) and B ∑ (k σR represents the three-channel blended cumulative color recording value. P G P B P The expected color values ​​for each of the three channels, where β represents the number of samples recorded, k. σ The weight of the impact on importance information.

[0039] Furthermore, correcting the mixed-cumulative color value of the second projected light after the second light projection sampling process has terminated, based on the expected color value of the first projected light, includes:

[0040] Compare the red channel R of the current blended cumulative color value ∑ (k T The red channel R with the expected color value P The numerical difference, denoted as ΔR = |R ∑ (k T )-R p |, if ΔR>ε, and R ∑ (k T ) < R p -ε, then let the projected ray l start from the sampling point Continue forward and calculate the blended cumulative color value to obtain the sampling point. The cumulative color value is C ∑ (k T +1), continue forward until ΔR' = |R ∑ (k T +n r )-R p |≤ε, and record the corrected blending cumulative color value of the red channel at this time as C1=C ∑ (k T +n r If ΔR > ε, and is R ∑ (k T )>R p +ε, then from the sampling point Back to directly read the sampling points The cumulative color value is C ∑ (k T -1), and continue backtracking until ΔR'=|R ∑ (k T -n' r )-R p |≤ε, and record C1=C ∑ (k T -n' r If ΔR ≤ ε, then no correction is needed, and let C1 = C ∑ (k T ), where ε is the preset error threshold, k T n represents the current sampling count.r and n' r These represent the number of steps forward or backward when the correction is complete.

[0041] Obtain the corrected mixed cumulative color values ​​of the green and blue channels, and obtain the corrected mixed cumulative color value of the second projected ray based on the average of the corrected mixed cumulative color values ​​of the three channels.

[0042] The industrial furnace wall temperature field data visualization system provided by this invention includes:

[0043] The present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for visualizing industrial furnace wall temperature field data provided by the present invention.

[0044] This invention proposes a method and system for visualizing industrial furnace wall temperature field data. Based on thermocouple temperature measurement data and heat transfer theory, the method obtains industrial furnace wall temperature field data, fills this data into a preset volume texture, downsamples the volume texture using a first ray projection to obtain the expected color value of the first projected ray, samples the volume texture using a second ray projection based on a preset termination condition for the second ray projection, corrects the mixed and accumulated color value of the second projected ray after sampling termination according to the expected color value of the first projected ray, and then renders the volume drawing image based on the corrected mixed and accumulated color value of the second projected ray. This invention solves the technical problem of existing technologies being unable to quickly and efficiently visualize high-quality industrial furnace wall temperature field data, and achieves rapid and high-quality data visualization of industrial furnace wall temperature field data.

[0045] The beneficial effects of this invention specifically include:

[0046] (1) As a key boundary condition for the furnace wall temperature field, the temperature distribution of the inner surface of the furnace wall in industrial furnaces is crucial for solving the furnace wall temperature field. This invention fully studies the structural characteristics of the furnace wall and the arrangement of thermocouples in industrial furnaces. Combining multiple influencing factors and heat transfer theory, with distance and temperature difference as the main influencing factors, a temperature correction model for obtaining the temperature distribution of the inner surface of the furnace wall is constructed. The three-dimensional geometric structure of temperature correction is simulated, and then the magnitude of each influencing factor and correction parameter is calculated. Finally, the temperature field data of the entire furnace wall is obtained.

[0047] (2) In view of the non-strict symmetry of the structure of each layer of the furnace wall in industrial furnaces, this invention proposes an asymmetric Bresenham rasterization algorithm. Following the steps applicable to rasterization of asymmetric data structures, a rasterization model for the asymmetric furnace wall structure of industrial furnaces is constructed, realizing the creation and initialization of the furnace wall texture, and filling the acquired temperature field data information into the furnace wall texture.

[0048] (3) To improve the speed of the ray projection rendering process, this invention proposes for the first time a Russian Roulette algorithm based on ray importance, supplemented by threshold judgment of Alpha channel value and boundary judgment of skybox and geometric carrier, and comprehensively sets ray termination conditions. The algorithm first calculates the importance information of each projected ray, and then adjusts the color accumulation process of the projected ray based on the importance information, ultimately achieving the effect of terminating the ray projection accumulation process in advance, which greatly speeds up the ray projection process.

[0049] (4) Based on the theory of ray casting, a volume rendering model with secondary ray casting is proposed. In the first ray casting, the Alpha channel value is calculated and modified first, and then the expected color value of the projected ray is calculated, which provides the basis for the calculation process of the secondary ray casting. At the same time, in order to improve the volume rendering speed, a ray termination condition is set. However, ray termination may cause color deviation, which will affect the volume rendering quality. Therefore, in the second ray casting, the mixed and accumulated color value of the projected ray is calculated and compared with the expected color value obtained from the first ray casting. Color correction is performed in stages, and finally a good balance between volume rendering speed and quality is achieved.

[0050] (5) To address the problem that traditional ray casting algorithms struggle to balance volume rendering speed and quality, this invention proposes a balanced score evaluation criterion for assessing volume rendering speed and quality. This criterion integrates the similarity of volume rendering results, volume rendering time, and some adjustable parameters of the volume rendering system, effectively evaluating the balance between speed and quality in the volume rendering process. It provides guidance for parameter adjustment during the volume rendering process, thus enabling the volume rendering process to achieve satisfactory results after sufficient parameter adjustment. Attached Figure Description

[0051] Figure 1 This is a flowchart illustrating the method for visualizing the temperature field data of the industrial furnace wall according to Embodiment 2 of the present invention.

[0052] Figure 2 This is a cross-sectional view of the furnace wall of the industrial furnace in Embodiment 2 of the present invention;

[0053] Figure 3 This is a schematic diagram showing the relative positions of the actual buried points of three adjacent thermocouples and the temperature correction points in the furnace wall according to Embodiment 2 of the present invention.

[0054] Figure 4 This is a diagram showing the relationship between the correction point and the thermocouple position in Embodiment 2 of the present invention;

[0055] Figure 5 This is a schematic diagram illustrating the solution of angular relationships in Embodiment 2 of the present invention;

[0056] Figure 6A flowchart for Bresenham in Embodiment 2 of the present invention is drawn;

[0057] Figure 7 This is a flowchart of the secondary ray projection algorithm in Embodiment 2 of the present invention;

[0058] Figure 8 The results are from the furnace wall temperature field data visualization system of Embodiment 3 of the present invention;

[0059] Figure 9 A volumetric frame rate comparison chart for different Alpha channel thresholds in Embodiment 3 of the present invention is shown.

[0060] Figure 10 A volumetric image similarity comparison diagram is drawn for different Alpha channel thresholds in Embodiment 3 of the present invention;

[0061] Figure 11 This is a structural block diagram of the industrial furnace wall temperature field data visualization system according to an embodiment of the present invention.

[0062] Figure label:

[0063] 10. Memory; 20. Processor. Detailed Implementation

[0064] To facilitate understanding of the present invention, the present invention will be described more fully and in detail below with reference to the accompanying drawings and preferred embodiments, but the scope of protection of the present invention is not limited to the following specific embodiments.

[0065] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.

[0066] Example 1

[0067] The method for visualizing industrial furnace wall temperature field data provided in Embodiment 1 of the present invention includes:

[0068] Step S101: Based on thermocouple temperature measurement data and heat transfer theory, obtain industrial furnace wall temperature field data.

[0069] Step S102: Fill the industrial furnace wall temperature field data into the preset volume texture.

[0070] Step S103: Downsample the volume texture using the first ray projection to obtain the expected color value of the first projected ray.

[0071] Step S104: Based on the preset second ray projection termination condition, the volume texture is sampled using second ray projection.

[0072] Step S105: Correct the mixed and accumulated color value of the second projected light after the second light projection sampling process is terminated, based on the expected color value of the first projected light.

[0073] Step S106: Based on the mixed and accumulated color values ​​of the corrected second projected light, the volume rendering image result is obtained by shading and rendering.

[0074] The industrial furnace wall temperature field data visualization method provided in this invention obtains industrial furnace wall temperature field data based on thermocouple temperature measurement data and heat transfer theory. This data is then filled into a preset volume texture. A first ray projection is used to downsample the volume texture to obtain the expected color value of the first projected ray. Based on a preset second ray projection termination condition, the volume texture is sampled using second ray projection. The mixed and accumulated color value of the second projected ray after the second ray projection sampling process is terminated is corrected according to the expected color value of the first projected ray. Finally, based on the corrected mixed and accumulated color value of the second projected ray, a volume rendering image is obtained. This method solves the technical problem of existing technologies being unable to quickly and efficiently visualize high-quality industrial furnace wall temperature field data, and achieves rapid and high-quality data visualization of industrial furnace wall temperature field data.

[0075] Example 2

[0076] Embodiment 2 of the present invention proposes a method for visualizing industrial furnace wall temperature field data, which realizes real-time display and visualization of industrial furnace wall temperature field data. Figure 1 This is a diagram illustrating the implementation steps of the method proposed in this embodiment of the invention, including the following steps:

[0077] (1) An internal surface temperature correction model was constructed based on thermocouple temperature measurement data and heat transfer theory.

[0078] (2) Create an initial volume texture and fill the volume texture based on the asymmetric Bresenham circular rasterization algorithm.

[0079] (3) Calculate and modify the Alpha channel value. The first ray projection downsamples the volume texture to obtain the expected color value of the projected ray.

[0080] (4) The second ray projection sampled the volume texture, set the conditions that the ray projection should meet, proposed a Russian roulette algorithm based on the importance of the ray, set the ray termination condition, and performed a mixing and accumulation operation of the color value and the Alpha channel value. It was determined whether the mixed and accumulated color value was equal to or close to the expected color value, and then it was decided whether to perform correction.

[0081] (5) By evaluating the speed and quality of volume rendering, the balance score of the volume rendering process is obtained, and it is determined whether to perform parameter updates and process iterations.

[0082] (6) Display the rendering results.

[0083] The specific implementation plan is as follows:

[0084] (1) Construct an internal surface temperature correction model based on thermocouple temperature measurement data and heat transfer theory;

[0085] Acquiring furnace wall temperature field data is the foundation for constructing an industrial furnace wall data visualization system. Based on existing furnace wall thermocouple temperature measurement data, this invention constructs a steady-state heat transfer model of the furnace wall based on the inverse problem theory of heat conduction, derives the temperature calculation formula, and calculates the temperature distribution on the inner surface of the furnace wall through an iterative correction strategy, ultimately obtaining the industrial furnace wall temperature field data information to be visualized.

[0086] like Figure 2 The image shows a cross-sectional view of the wall of an industrial furnace. The furnace wall structure, from the outside in, consists of the furnace shell, filling layer, cooling wall, refractory bricks, and inner surface. Thermocouples are actually arranged within the cooling wall. A temperature correction model for the inner surface of the industrial furnace wall is constructed using thermocouple temperature data and heat transfer theory to obtain temperature field data for the entire furnace wall. Figure 2 It is known that, assuming there are temperature correction points on the inner surface corresponding to thermocouples, the temperature of these correction points can be preset by experts based on experience. Then, the heat transfer between furnace walls can be calculated using solid heat transfer formulas, thus deriving the temperature at the thermocouple location within the cooling wall. However, directly calculating this heat transfer process using a heat transfer model yields relatively large errors. Therefore, based on the structural characteristics of the furnace wall and the actual distribution of thermocouple embedding points, a temperature correction model is proposed to analyze the relationship between influencing factors and the temperature correction amount, thereby calculating a more accurate inner surface temperature distribution.

[0087] First, the temperature difference is obtained by comparing the results from the heat transfer model with the actual thermocouple measurements. Then, the temperature difference is used as the termination condition for iteration. If the condition is not met, it is corrected using a modified model. After multiple iterations that satisfy the condition, the required accuracy of the internal surface temperature distribution is considered achieved. Figure 2 The actual locations of thermocouple embedding points in a certain layer of the cooling wall are marked as j (j = 1, 2, ..., N). Based on actual operating conditions, discrete points i (i ≤ ε1, 2, ..., N) on the inner surface of the furnace wall are taken as temperature correction points, so that each correction point corresponds to an actual thermocouple embedding point, and N is the number of actual thermocouples selected for this layer. Let the temperature of the i-th temperature correction point after the k-th correction be T. ij k Temperature correction amount is The relation is:

[0088]

[0089] The temperature value T at the corresponding thermocouple embedding point is then calculated in the forward direction. j k After k iterations of calculation, the sum of squares of temperature differences (ΔTss) at all thermocouple embedding points in this layer k The formula for calculating ) is as follows:

[0090]

[0091] In the formula, T j Let be the actual temperature value of the j-th thermocouple. Determine ΔTss. k Is it less than the iteration threshold ξ? If ΔTss k If ξ < ξ, then the iteration terminates.

[0092] Based on the furnace wall structural characteristics and heat transfer analysis results, the temperature correction amount at the correction point is related to the material's thermal conductivity λ, as well as factors such as the distance R between the actual thermocouple embedding point and the correction point, and the temperature difference E. The temperature correction amount ΔT for the i-th correction point is now established through iterative correction. ij Relationship with these factors:

[0093] ΔT ij =F(λ,R,E),i,j=1,2,...,N (3)

[0094]

[0095] Where r μ ,e μ (μin[j,j_left,j_right,j_up,j_down]) represents the distance and temperature difference between the 5 nearest thermocouples to the i-th correction point and that correction point, where R = F. R (), E = F E () represents the control functions for distance and temperature. Figure 3 This diagram illustrates the relative positions of the actual embedded points of thermocouples in three adjacent layers of the furnace wall to the temperature correction points. The thermocouples in the same layer are arranged approximately evenly and equidistantly on the same circumference, while thermocouples with the same serial number in different layers are almost located on the same vertical line. Therefore, the five thermocouples closest to the i-th correction point are considered as ΔT. ij Sources of influencing factors.

[0096] Plot the relative positions of a given correction point to its five nearest actual thermocouple embedding points, as shown below. Figure 4As shown, these five actual thermocouples are the thermocouple corresponding to the correction point (i=j), and the four thermocouples above, below, to the left, and to the right of it. This is equivalent to considering two mutually perpendicular circles, with the intersection of the two circles being the actual thermocouple corresponding to the i-th correction point. In the figure, r and e represent the distance and temperature difference between the thermocouple and the i-th correction point, respectively. Let the straight line between the i-th correction point and the j-th thermocouple be the standard line of the correction model, denoted as l. ij (i=j), because these 5 thermocouples will be connected to l later. ij The included angle is used to calculate the distance-temperature correction ΔT. ij Based on the degree of influence, it can be seen that within the same horizontal circular plane (the horizontal circular plane containing i,j,j_left,j_right), the absolute values ​​of the angles between the correction point and the positions of these three actual thermocouples are 0°, ∠1, and ∠2, respectively; within the vertical circular plane (the vertical circular plane containing i,j,j_up,j_down), the absolute values ​​of the angles between the correction point and the other two thermocouples are calculated from the distances between the thermocouples, and are ∠3 and ∠4, respectively. The calculation formulas can be found in [reference needed]. Figure 5 The formula is as follows:

[0097]

[0098] Where d(i,j) is the distance formula between two points. From equation (4), expanding equation (3) at the origin using Taylor's formula, we get:

[0099]

[0100] In equation (6), λ is the thermal conductivity of the material, therefore set up:

[0101]

[0102] Where, k eμ Regardless of the relative position between the two points, it can be set as a constant parameter to be determined. Let This describes the relationship between the distance between the thermocouple and the correction point and the temperature correction, and is based on the heat conduction equation for isotropic materials, assuming w iμ for:

[0103]

[0104] In the formula, d iμ =r oi / r oμ This represents the ratio of the polar radii of two points, where point o is the center of the circle containing the point o, and r is the radius of the circle containing the point o. oi and r oμ Let θ be the distance from the two points to the center of the circle. iμThe angle is obtained from equation (5) and expressed in radians. The temperature threshold ξ is determined by the physical structure and material properties of the blast furnace wall. In summary, the modified model can be written as:

[0105]

[0106] In the formula, U=[j,j_left,j_right,j_up,j_down], w iμ To show the influence of the distance between the thermocouple and the correction point on the temperature correction, r μ and e μ These represent the distances and temperature differences between the 5 thermocouples and the i-th correction point, respectively, and k is the distance and temperature difference between them. eμ The temperature difference is the correction amount for temperature ΔT ij The influence degree parameter. The magnitude of the unknown parameter in the above formula can be determined through temperature conduction experiments, and finally the temperature correction model can be obtained and the inner surface temperature data can be calculated.

[0107] (2) Create an initial volume texture and fill the volume texture based on the asymmetric Bresenham circular rasterization algorithm;

[0108] First, data preprocessing is performed to standardize the temperature field data for easier subsequent visualization. Based on the structural characteristics of industrial furnace walls, the temperature field data is approximately distributed in several concentric circles on a cylindrical coordinate system. After temperature correction and interpolation, the industrial furnace wall temperature field data is represented as (H, R, T), where each data point includes axial distance H, radial distance R, and temperature value T. To display the color values ​​of the temperature data, it is first mapped to grayscale values, and then converted into a color image using RGB three-channel values ​​through pseudo-color image processing.

[0109] Then, an initial volume rendering scene is created, and a cuboid model is added to the scene. This geometric model serves as the geometric carrier for volume rendering. An initial blank volume texture is created within this geometric carrier, and the volume texture size is determined to be H based on the data preprocessing results. Axial ×2R max ×2R max H Axial R represents the number of layers in the temperature field data, arranged according to axial distance. max To determine the maximum radial distance, the color value of each texel in the volume texture is initialized to (1,1,1,0). The alpha channel indicates whether volume data exists at that texel point. The visibility of volume texture texels indirectly characterizes the geometry of the volume texture object, avoiding the need to create a geometric model of the object in 3D modeling software. Based on the scale of the furnace wall temperature field data and actual visualization requirements, the volume texture storage format is set to RGBA32, the loop mode is Clamp mode, and the filtering mode is nearest neighbor filtering.

[0110] Then, a rasterization operation is performed to discretize the continuous geometric information, making it easier for computers to process. The traditional Bresenham circular rasterization algorithm draws a circle on the volume texture and determines the position of each pre-drawn pixel based on its approximation to the circle. This method only requires drawing one-eighth of the circle first, and then using a symmetric transformation to draw a complete circle. The approximation degree d can be represented by the number of pre-drawn pixels (x...). i ,y i The distance from the center to the circle is expressed as the difference between the square of the distance and the square of the radius R, as shown in equation (10). Consider the case of drawing an arc in the region 0 ≤ x < y, such as... Figure 6 (a), d H and d D These represent the approximation levels of pre-drawn pixels H and D to the circle:

[0111]

[0112] To compare the approximation of two pre-drawn pixels, let the decision parameter be p. i , represented as:

[0113]

[0114] p i The size of p determines the selection of pre-drawn pixels. i When >0, we have |d H |>|d D If point D is closer to the circle, then D is chosen as the next pixel to be drawn; otherwise, H is chosen. When drawing the next pixel, the coordinates of points H and D change, and the degree of approximation d... H and d D Change, p i As a result, the new decision parameters are changed and denoted as p. i+1 According to x i x i-1 y i y i-1 The relationship and formula (11), p i+1 With p i The relationship can be represented as:

[0115]

[0116] Specifically, for the starting point, we have:

[0117] p0 = 3 - 2R (13)

[0118] Iteratively calculate and update the decision parameter p iThe position of the next pixel to be drawn is determined by the new decision parameters.

[0119] Since the temperature field data of most industrial furnace walls is not perfectly symmetrically distributed on a circle, directly performing a symmetrical transformation on an eighth-segment arc cannot obtain volume textures with asymmetric information. To address this problem, this invention proposes an asymmetric Bresenham rasterization algorithm based on the idea of ​​this algorithm, further extending the circular rasterization algorithm in the 0≤x<y region to the entire two-dimensional coordinate system, such as... Figure 6 (b) is the derivation process for the region -x≤y<0. Therefore, the decision parameters should be recalculated in each region, and the position coordinates of the pre-drawn pixel should be determined based on the decision parameters. The color of the pre-drawn pixel is the color value in RGB space calculated from the temperature field data. The decision parameters of the Bresenham circular rasterization algorithm in the entire two-dimensional coordinate system are shown in Table 1.

[0120] Table 1

[0121]

[0122]

[0123] The preprocessed furnace wall temperature field data can now be layered into the blank volume texture. In each layer, temperature correction point data and thermocouple temperature measurement data are sequentially filled using an asymmetric Bresenham rasterization algorithm. Then, the temperature data between the inner surface of the furnace wall and the cooling wall where the thermocouple is located is calculated by interpolation and filled into the volume texture. At this point, the RGB values ​​of the data filled into the volume texture are its color values, and the Alpha channel value is still 0. In the subsequent first ray projection process, the Alpha channel value will be calculated and changed to make the data visible.

[0124] (3) Calculate and modify the Alpha channel value. The first ray projection downsamples the volume texture to obtain the expected color value of the projected ray.

[0125] Raycasting is a direct volume rendering technique based on image sequences. Its calculations do not stop at the surface of the object, but rather sample along rays that pass through the object's interior. First, a ray is cast from each pixel on the screen along the line of sight. This ray passes through the object, and samples are taken at equal intervals along the ray to calculate the color value and alpha channel value (opacity value) of the sampled point. Then, the color value and opacity value are combined in either front-to-back or back-to-front order to obtain the final color value and opacity value of that point.

[0126] First, calculate and modify the Alpha channel value necessary for light projection. Referring to the definition of optical thickness, its description starts from the starting point s.start To the destination s end The optical thickness is:

[0127]

[0128] In the formula, L(t) represents the light absorption coefficient as a function of time, and t(s) start ) and t(s end ) represent the light reaching s start Point and s end The time interval of a point. Divide the optical path into several equally spaced segments, s k (k = 0, 1, 2, ..., N) RC ) represents the endpoints of each interval, N RC This represents the maximum number of samples set when sampling at high resolution. Let the Alpha channel value at this point be α(k), and let α(0) = 1, and:

[0129]

[0130] Therefore, during the calculation process, the Alpha channel value of the sampling point is updated to α(k), that is, modified to (R,G,B,α(k)). If the Alpha channel value of some volume textures is still the initial value of 0, then the visibility test shows that the texture is a transparent texture and can be ignored during the sampling process.

[0131] The first ray projection is performed, and the volume texture is downsampled at low resolution. Let N be the number of samples in this downsampling process. down s k (k = 0, 1, 2, ..., N) down ), and N down <N RC This means that the number of samples is relatively small and the sampling interval is relatively large, so the process is relatively fast.

[0132] As the ray casting algorithm flow shows, the CPU needs to transfer the data to be processed to the GPU. Therefore, the vertex positions of the geometric carrier and the camera position should be read from the application stage of the rendering process. Taking a specific vertex position P0 on the geometric carrier as an example, the direction of the ray cast at P0 can be calculated based on P0 and the camera position. To ensure spatial uniformity, the model transformation matrix M... model The camera position can be moved from world space P c Transform to model space P c Therefore, for a ray l emitted from vertex P0, its direction is P. c 'Points to P0. To display the volume rendering result on the screen, the positions of the geometric carrier vertices in model space need to be transformed in the vertex shader stage of the rendering pipeline according to the order of "model space - world space - view space - clip space".

[0133] P0'=M pers M view M model P0 (16)

[0134] Among them, M view and M pers These represent the observation transformation matrix and the perspective projection matrix, respectively. These two transformation matrices are calculated based on the relevant parameters of the geometric carrier, scene, and camera. In the subsequent stages of the vertex shader, the hardware performs NDC transformation, clipping, screen mapping, rasterization, and interpolation on the vertex information of the geometric carrier in clip space. The vertex information of the geometric carrier after the above processing is used as the input to the fragment shader, where volume texture sampling based on ray casting is performed.

[0135] During the light projection process, this invention uses a front-to-back approach (from the camera to the geometric carrier) to calculate the sampled and accumulated color values ​​and Alpha channel values, as shown in the following formula:

[0136] C ∑ (k)=C ∑ (k-1)+[1-α ∑ [(k-1)]·α(k)·f α ·C(k) (17)

[0137] α ∑ (k)=α ∑ (k-1)+[1-α ∑ [(k-1)]·α(k)·f α (18)

[0138] Where k is the current sampling number, and satisfies 0 ≤ k ≤ N down N down This is the maximum number of times the volume texture is sampled during downsampling. (C) Σ (k) and α ∑ (k) represents the cumulative color value and Alpha channel value obtained after the kth sampling of ray l, f α This is the alpha coefficient used for color mixing and accumulation. When the number of sampling accumulations reaches N... down That is, after the sampling process of the first ray projection is completed, the result C of the first ray projection can be obtained. P =C Σ (N down However, directly displaying the downsampled texture at low resolution on the screen may result in color cast and blurring, or aliasing and distortion. Therefore, a secondary ray casting algorithm is introduced to perform a second ray casting, which can improve the quality of the first ray casting result C. PThe expected color value is used in the color correction process of the second ray projection. Experimental analysis shows that because the first ray projection process is faster, the speed of the second ray projection mainly depends on the second ray projection process. Furthermore, the result quality of the second ray projection algorithm is better and faster than that of using only a single non-downsampling ray projection. This method combines the results of the first and second ray projections, improving speed while significantly reducing image color cast, jagged edges, aliasing, and other deviations, thereby improving image quality.

[0139] (4) The second ray projection sampled the volume texture, set the conditions that the ray projection should meet, proposed a Russian roulette algorithm based on the importance of the ray, set the ray termination condition, and performed a mixing and accumulation operation of the color value and the Alpha channel value. It was determined whether the mixed and accumulated color value was equal to or close to the expected color value, and then it was decided whether to perform correction.

[0140] During the second ray projection, the volume texture is sampled and mixed and accumulated, with a maximum sampling number of N. RC To improve the speed of light projection, a common method is to set a light termination condition. During the sampling and accumulation process, the accumulated Alpha channel value α is determined. Σ (k) and the set threshold σ·α T The magnitudes of these values ​​are related, with σ (0 < σ < 1) representing the threshold weights of the Alpha channel. The second light projection process can then be divided into two stages, as detailed in the following reference... Figure 7 :

[0141] Stage ①: If α ∑ (k)<σ·α T The light termination condition for this stage is set as follows:

[0142]

[0143] Because during the light projection process, sampling point P k The ray must be within the boundary of the geometric carrier, and if the ray l has already hit the skybox (the environment and background that need to be set during rendering), the ray l will be terminated prematurely. In the formula, the shoot() function is the function to determine whether the ray intersects with the object, and x... B y B z B Represents the bounding box boundary of the geometric carrier, with η as the criterion P. k The redundancy of whether it is within the bounding box is usually taken as a small number. l.isHit is the result of the shoot(l,SkyBox) function. Specifically, isHit is a property of the ray l, indicating whether it hits the boundary.

[0144] For certain rays in stage ①, α Σ (k) During the entire sampling and accumulation process (0≤k≤N) RC It is possible that the threshold σ·α will not be reached. T Therefore, the ray will not enter stage ② before it terminates.

[0145] Phase ②: If α ∑ (k)≥σ·α T That is, when the alpha channel value α is accumulated by mixing Σ (k) Reaching the set threshold σ·α T Record the sampling number k at this time. σ Sampling points The three-channel color values ​​R are mixed and accumulated. Σ (k σ ), G Σ (k σ B ∑ (k σ To set the ray termination condition in stage ②, this invention proposes for the first time a Russian Roulette of Significance (RROS) algorithm based on ray importance, as shown below:

[0146]

[0147] In the formula, C(k) is the RGB color value at the kth sampling point, and P is the model probability of the algorithm, representing the probability that p = P such that C(k) = C(k) / P, and the probability that p = 1-P such that C(k) = 0. N T Here, S is the sampling threshold for different cases of the algorithm, and Significance is the importance information, which is the degree of importance of the ray to the expected result of the volume rendering by judging the sampling process before the ray is projected. The formula is as follows:

[0148] Significance = [|R ∑ (k σ )-R P |+|G Σ (k σ )-G P |+|B ∑ (k σ )-B P |+β·k σ ] -1 (twenty one)

[0149] Where R P G P B P These represent the expected color value C. PThe three-channel values, where β represents the number of samples k. σ The weight of the influence of importance information is δ. The larger the Significance, the more important the projected ray is, and the greater its contribution to the final display result. S It is the threshold for determining the importance of light in different RROS situations.

[0150] Equation (20) states that for each projected ray entering stage ②, its significance is calculated and then judged. If Significance < δ S And at this time k σ <N T Then continue sampling until the number of samples reaches N. T Alternatively, sampling may terminate when the light termination condition (Equation (19)) is met, and the color value C(k) = C(k) remains unchanged during this process; if Significance < δ S And k σ ≥N T Then sampling is terminated directly; if Significance ≥ δ S Then continue from the number of times k = k σ Start sampling until the number of samples reaches N RC Alternatively, sampling may terminate when the light termination condition (Equation (22)) is met, in which case there are two possibilities:

[0151] (i) If k σ <N T If the color value C(k) = C(k) remains unchanged, sampling will terminate if the light termination condition (equation (22)) is met; otherwise, sampling will continue until k = N. T Then, it immediately proceeds to stage (ii);

[0152] (ii) If N T ≤k σ ≤N RC For each subsequent sampling process, the color value C(k) has a probability of p = P such that C(k) = C(k) / P, and a probability of p = 1 - P such that C(k) = 0. When the number of samplings reaches N... RC Alternatively, sampling may be terminated when the light termination condition (Equation (22)) is met.

[0153] Therefore, the ray projection process in stage ② should satisfy the termination condition of the Russian Roulette algorithm based on ray importance, as shown below:

[0154]

[0155] While traditional ray casting algorithms improve speed by setting ray termination conditions, ray termination can lead to color deviations in the final volume texture sampling results. Therefore, to reduce color deviations caused by ray termination in ray casting algorithms, the sampling number k is recorded when the ray terminates. T (k T ≥k σ ), sampling points and the cumulative color value C ∑ (k T We now consider using each channel (R, G, B) as the calibration standard and performing three overall calibrations. In each calibration process, the color value of that channel is used as the calibration standard. After calibration, the mixed and accumulated color value of the three channels is taken as the result. Finally, the average of the three calibration results is taken as the final result.

[0156] Correction Phase I: First, compare the red channel R of the current blended color value. Σ (k T The red channel R with the expected color value P The numerical difference, denoted as ΔR = |R ∑ (k T )-R p |

[0157] If ΔR > ε (ε is the redundancy), and R is... ∑ (k T ) < R p -ε, then let the projected ray l start from the sampling point Continuing forward, the mixed cumulative color value is calculated using formula (17) to obtain the sampling point. The result is C ∑ (k T +1), at this point the Alpha channel's blended cumulative value will not be affected by the threshold α. T Due to limitations, and ensuring that ray l will not exceed the boundary or skybox when it continues to be sampled, the calculation formula for this process is as follows:

[0158] C ∑ (k T +1)=C ∑ (k T )+[1-α ∑ (k T )]·α(k T +1)·f α ·C(k T +1) (23)

[0159] α ∑ (k T +1)=α ∑ (k T )+(1-α∑ (k T ))·α(k T +1)·f α (twenty four)

[0160] Continue sampling until ΔR' = |R ∑ (k T +n r )-R p |≤ε, and record the cumulative color value C1=C at this time. ∑ (k T +n r ).

[0161] If ΔR > ε, and is R ∑ (k T )>R p +ε, then from the sampling point Backwards. When backwards from a sampling point, the blended cumulative color value at that sampling point has already been calculated and stored in memory, so the sampling point can be read directly. The result at point C is... ∑ (k T -1), continue backing away until ΔR'=|R ∑ (k T -n' r )-R p |≤ε, and record the color value C1=C at this time. ∑ (k T -n' r ).

[0162] If ΔR ≤ ε, then no correction is needed; let C1 = C ∑ (k T ).

[0163] Correction Phase II: Similarly, the green channel G is compared with the current blended color value. ∑ (k T The green channel G with the expected color value P The corrected cumulative color value is C2.

[0164] Correction Phase III: Similarly, compare the current blended color value of the blue channel B. ∑ (k T (Blue channel B) and the expected color value P The corrected cumulative color value is C3.

[0165] Therefore, the average value of C1, C2, and C3 is finally taken as C. final= (C1+C2+C3) / 3, and display it as the final result on the screen. The reason for not individually correcting the R, G, and B channels during the correction process is that the volume texture is treated as a whole, and information from different sampling points should be considered simultaneously. This helps maintain the continuity and integrity of the volume rendering image, avoiding issues such as jagged edges and color casts that degrade the quality of the volume rendering image. Verification shows that volume rendering images sampled holistically have higher similarity and better volume rendering quality. It is also worth noting that if a ray termination condition is not set and the threshold judgment and repeated iteration of the mixed and accumulated color values ​​are performed directly, i.e., a judgment is made every time it is iterated, the time complexity of this process is O(n^2). 2 The time complexity is greater than that of the method proposed in the embodiments of the present invention (O(n)).

[0166] (5) By evaluating the speed and quality of volume rendering, the balance score of the volume rendering process is obtained, and it is determined whether to perform parameter updates and process iterations.

[0167] To efficiently evaluate the speed and quality of volume rendering and achieve a better balance between the two, this invention proposes an evaluation function for calculating rendering speed and quality during the volume rendering process, as shown below:

[0168]

[0169] (Quality, Time) = F system (H Axial, ,R max, FilterMode,WrapMode,AnisoLevel)(26)

[0170] In the formula, Quality = F Similarity (Source, Target) Rendering quality metrics, F similarity This is a similarity comparison function for rendering results. It quantifies the similarity between two rendered images (Source, Target) using histogram matching, thus reflecting the rendering quality. Balance_Score is the balance evaluation score, Time is the total rendering time, and γ is the rendering speed preference; a larger γ indicates a greater emphasis on the optimization and contribution of the entire process to volume rendering speed. The final rendering result's quality (Quality) and time (Time) are determined by the function F. system (H Axial ,R max ,FilterMode,WrapMode,AnisoLevel) are used to obtain F systemThis is a parameter adjustment function for the system's rendering mode. Its parameters represent the axial height and radial length of the geometric model, the rendering system's filtering mode (nearest neighbor filtering, bilinear interpolation filtering, trilinear filtering, etc.), the loop mode (Repeat, Clamp, Mirror, etc.), and the anisotropy level (numerical parameter). By setting these rendering system parameters, the final rendering result's quality (Quality) and rendering time (Time) are obtained, which in turn significantly affect the Balance_Score.

[0171] By calculating the Balance_Score, the balance between volume rendering speed and quality during ray casting can be obtained. Using the Balance_Score as a reference, system parameters can be adjusted to achieve an optimal volume rendering state that balances both quality and speed. For example, when choosing a rendering system filtering mode, options include nearest neighbor filtering, bilinear interpolation filtering, and trilinear filtering. While nearest neighbor filtering may not be as effective as bilinear or trilinear filtering, it is simpler because the latter two are more precise but time-consuming, thus impacting volume rendering speed. Adjusting the rendering speed preference γ also affects the balance score. If rendering speed (time) is more important, γ can be increased, thus improving the F... system In terms of parameter selection, it's advisable to prioritize parameters that significantly improve speed. Therefore, by referencing the Balance_Score and adjusting the rendering system's parameters, a satisfactory balance between rendering quality and speed can be achieved.

[0172] (6) Display the rendering results

[0173] After completing the secondary ray casting, the final ray casting mixture and accumulation result C RC It is already stored in the color buffer. However, before using ray casting for volume rendering, the color buffer may already contain the previous rendering result. In this embodiment of the invention, the previous rendering result stored in the color buffer is the scene color C. BG To meet the visualization needs of volumetric data and display all volumetric details, this invention treats the volumetric drawing object as a transparent object, so that when the human eye observes the volumetric drawing object, it can not only observe the volumetric data information on the front of the volumetric drawing object, but also observe the volumetric data information on the back of the volumetric drawing object by relying on the perspective effect.

[0174] When rendering volumetric objects using transparent object rendering, stencil testing, depth testing, and transparency testing are required in the rendering pipeline, and color blending of the source and target colors must be enabled. Additionally, rendering latency and screen tearing may occur when displaying the rendered results; that is, the rendered image to be displayed may still be rendering. This issue can lead to discontinuous display of the final rendering result. Therefore, a double buffering strategy and automatic vertical synchronization are used to solve this problem, ensuring that the displayed rendering result is always continuous.

[0175] This invention aims to provide a method for visualizing the temperature field data of industrial furnace walls. To quickly and efficiently visualize high-quality industrial furnace wall temperature field data, this invention comprehensively considers the structural form of the furnace wall and the arrangement of thermocouples used for temperature measurement. It proposes a temperature correction model based on thermocouple temperature data to calculate the inner surface temperature of the furnace wall, thus realizing the calculation of the inner surface temperature of the industrial furnace wall. The temperature distribution of the inner surface of the furnace wall is crucial as one of the boundary conditions for constructing the temperature field of the industrial furnace wall. Simultaneously, addressing the problems of low quality, slow speed, and difficulty in balancing volume rendering speed and visualization quality in existing volume rendering methods, this invention proposes a Russian roulette algorithm based on ray importance, setting ray termination conditions to improve volume rendering speed, while satisfying the basic conditions of ray casting. However, to address the color deviation and reduced volume rendering image quality caused by ray termination, a secondary ray casting algorithm is proposed. This algorithm uses the mixed and accumulated color values ​​calculated by secondary ray casting for threshold judgment and color correction. Using the average brightness of the volume rendering image as a similarity index, it was found that the volume rendering image after color correction has higher similarity. Meanwhile, to assess the balance between speed and quality in the volume rendering process, this invention proposes a method for calculating the balance score between volume rendering speed and quality. After sufficient parameter tuning based on this method, satisfactory volume rendering results can be achieved. In summary, the industrial furnace wall temperature field data visualization method proposed in this invention, which integrates an internal surface temperature distribution correction model and a secondary ray projection algorithm, achieves real-time visualization of industrial furnace wall temperature field data, offering advantages such as dynamic intuitiveness, high quality, and high speed.

[0176] Specifically, this invention focuses on the furnace wall of an industrial furnace and proposes a method for visualizing the temperature field data of the furnace wall. This method includes an inner surface temperature correction model, an asymmetric Bresenham circular rasterization method, and a secondary ray projection algorithm. It also designs a Russian roulette algorithm based on ray importance, sets ray termination conditions, and proposes a balanced score evaluation criterion for assessing the speed and quality of volume rendering. This invention fully considers the structural characteristics of the industrial furnace wall and the arrangement of thermocouples. Combining a multi-influence factor fusion method and heat transfer theory, it constructs a three-dimensional geometric structure for temperature correction, calculates the magnitudes of each influencing factor and correction parameter, and proposes a temperature correction model based on thermocouple temperature detection to calculate the inner surface temperature of the furnace wall. This enables the calculation of the inner surface temperature of the industrial furnace wall and provides boundary conditions for constructing the temperature field of the industrial furnace wall. To address the non-strict asymmetry of industrial furnace wall structures, this invention proposes an asymmetric Bresenham circular rasterization algorithm to handle asymmetric furnace wall structures, based on the traditional Bresenham rasterization algorithm's approach to selecting the next rendering point. This solves the problem of incomplete rasterization of asymmetric furnace wall structures in industrial furnaces, enabling the creation and filling of volume textures. To address the difficulty in balancing volume rendering speed and quality during ray casting, this invention proposes a Russian roulette algorithm based on ray importance. This algorithm sets ray termination conditions to accelerate volume rendering. However, ray termination may lead to color deviations, affecting volume rendering quality. Therefore, this invention proposes a secondary ray casting algorithm, which improves volume rendering quality by color correction of the mixed and accumulated color values, achieving a balance between volume rendering speed and quality. To better evaluate the balance between volume rendering speed and quality, this invention proposes a balance score evaluation criterion. By referring to this evaluation criterion and fully adjusting the volume rendering system parameters, satisfactory volume rendering results can be achieved. Meanwhile, this invention addresses the color buffering issue during volume rendering by treating the volume rendering object as a transparent object and enabling color blending in the rendering pipeline, thus showcasing all the volumetric details of the temperature field data. To address rendering latency and screen tearing, a dual buffering and automatic vertical synchronization strategy is employed to ensure continuous display of the rendering results. Ultimately, this invention enables the visualization of industrial furnace wall temperature field data, offering advantages such as dynamic intuitiveness, high quality, and high speed.

[0177] Example 3

[0178] This third embodiment uses a 1050m iron smelting plant as an example. 3 Using the blast furnace as the experimental platform, several sets of blast furnace wall temperature field data were selected for visualization, and the visualization results were displayed and analyzed. Figure 8The visualization results of the blast furnace wall temperature field data are displayed. The visualization results are obtained using ray casting without ray termination, ray casting with ray termination, and secondary ray casting methods, respectively. The visualization results are then compared from the perspectives of volume rendering speed and volume rendering image similarity. The effectiveness and correctness of the secondary ray casting method are verified through the comparison of these two indicators.

[0179] (1) Volume Rendering Speed ​​Comparison: Volume rendering speed is evaluated using frame rate (unit: Hz). A higher frame rate indicates a shorter rendering time per frame, resulting in faster volume rendering. The frame rate of ray casting without ray termination is used as a control, and different alpha channel thresholds α are set for each method. T , to different α T The frame rates of ray casting with ray termination, ray casting with secondary ray casting, and the control group were compared. The comparison results under different alpha channel thresholds are as follows: Figure 9 As shown, it can be seen that with α T As the ray terminating condition is reduced, the ray casting frame rate with ray termination is a multiple of the ray casting frame rate without ray termination, gradually increasing from 1.089 times to 1.538 times. It can be seen that after setting the ray termination condition, the ray casting speed is significantly improved.

[0180] (2) Comparison of similarity between volume-drawn images

[0181] Because under certain conditions, the human eye has difficulty distinguishing the similarity between volume rendering images obtained by different methods and volume rendering images without ray termination, this invention employs histogram matching to quantify and analyze the similarity of volume rendering images. A comparison of volume rendering image similarity under different alpha channel thresholds is shown below. Figure 10 As shown in the figure, a comparison reveals that with α T As α increases, the image similarity of the secondary ray casting algorithm is always greater than that of the ordinary ray casting algorithm, only when α... T When = 1, the similarity of the images obtained by the two methods is the same, so it can be concluded that the visualization image obtained by the secondary ray casting method is of better quality.

[0182] In summary, the experiments show that, under different alpha channel thresholds, setting ray termination conditions results in faster volume rendering with a significant speed improvement, and the secondary ray casting algorithm yields results with higher similarity and better visualization quality.

[0183] Reference Figure 11The industrial furnace wall temperature field data visualization system proposed in this embodiment includes a memory 10, a processor 20, and a computer program stored on the memory 10 and run on the processor 20. When the processor 20 executes the computer program, it implements the steps of the industrial furnace wall temperature field data visualization method proposed in this embodiment.

[0184] The specific working process and working principle of the industrial furnace wall temperature field data visualization system in this embodiment can be referred to the working process and working principle of the industrial furnace wall temperature field data visualization method in this embodiment.

[0185] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for visualizing temperature field data of industrial furnace walls, characterized in that, The method includes: Based on thermocouple temperature measurement data and heat transfer theory, the temperature field data of the industrial furnace wall is obtained. This includes: Thermocouples are arranged in a ring within the cooling wall of the industrial furnace wall structure. Obtain the preset temperature correction point on the inner surface of the industrial furnace wall corresponding to the thermocouple; Based on heat transfer theory, a temperature correction model is established to calculate the inner surface temperature of the furnace wall in industrial furnaces. Based on the temperature correction model, the temperature data at the temperature correction points are iteratively corrected according to the sum of squared temperature differences at all thermocouple embedding points in each layer, thereby obtaining the temperature field data of the industrial furnace wall. The specific form of the temperature correction model is as follows: Among them, T ij k and T ij k+1 Let be the temperature values ​​of the i-th temperature correction point after k and k+1 iterations, respectively. Let ΔT be the temperature correction amount for the i-th temperature correction point corresponding to the j-th thermocouple in the (k+1)-th iteration. ij Let be the temperature correction amount at the i-th temperature correction point, and U be the set of the 5 thermocouples closest to the i-th correction point, where U = [j, j_left, j_right, j_up, j_down]. iμ The temperature correction ΔT is the distance between the μ-th thermocouple (one of the five nearest thermocouples) and the i-th temperature correction point. ij The degree of influence parameter, r μ and e μ Let k be the distance and temperature difference between the μ-th thermocouple and the ith correction point, respectively, among the 5 nearest neighbor thermocouples. eμ The temperature difference is the correction amount for temperature ΔT ij The degree of influence parameter, F R Let j_left, j_right, j_up, and j_down represent the thermocouples that are on the left and right sides of the same horizontal circular plane and on the same vertical circular plane as the j-th thermocouple, respectively. Fill the industrial furnace wall temperature field data into the preset volume texture; The volume texture is downsampled using the first ray projection to obtain the expected color value of the first projected ray; Based on the preset second ray projection termination condition, the volume texture is sampled using the second ray projection; The mixed and accumulated color value of the second projected ray after the second ray projection sampling process is terminated is corrected based on the expected color value of the first projected ray; The volumetric rendering result is obtained by shading and rendering based on the mixed and accumulated color values ​​of the corrected second projected light rays.

2. The method for visualizing industrial furnace wall temperature field data according to claim 1, characterized in that, Filling the industrial furnace wall temperature field data into a preset volume texture includes: Create an initial volume texture; Each layer of the initial volume texture is sequentially filled with temperature correction point data and thermocouple temperature measurement data using the Bresenham rasterization algorithm to obtain the first volume texture. The temperature data between the inner surface of the furnace wall and the cooling wall where the thermocouple is located is calculated by interpolation and then filled into the first volume texture to obtain a filled volume texture.

3. The method for visualizing industrial furnace wall temperature field data according to any one of claims 1-2, characterized in that, Based on a preset second ray projection termination condition, sampling the volume texture using second ray projection includes: Determine whether the cumulative Alpha channel value of the second projected ray in the second ray projection is less than a preset channel threshold. If so, terminate the sampling of the second projected ray according to the first termination condition. If not, the sampling of the second projected light is terminated according to the second termination condition. At the same time, the sampling count, sampling point, and three-channel mixed-cumulative color record values ​​are recorded when the mixed-cumulative Alpha channel value of the second projected light reaches the preset channel threshold. The light importance is calculated based on the sampling count, the three-channel mixed-cumulative color record values, and the expected color values ​​of the three channels. If the light importance is determined to be less than the preset importance threshold, sampling continues until the sampling count reaches the preset first sampling count while keeping the color values ​​of the three channels unchanged. If the light importance is determined to be greater than the preset importance threshold, the color values ​​of the three channels are adjusted according to the Russian Roulette algorithm model based on the light importance.

4. The method for visualizing industrial furnace wall temperature field data according to claim 3, characterized in that, The first termination condition is as follows: The second termination condition is as follows: Where, N RC The maximum number of samples is given, σ is the threshold weight of the alpha channel, and 0 < σ < 1, α T α is the threshold for the alpha channel. Σ (k) represents the accumulated Alpha channel value of the k-th sample, the shoot(l,SkyBox) function is the function to determine whether the ray l intersects with the object SkyBox, and l.isHit is the result of the shoot(l,SkyBox) function. Let P be the coordinate point. k Spatial coordinates, x B y B z B Represents the coordinates of the bounding box boundary where the geometric carrier is located, and η is the judgment P. k Whether it is redundancy within the bounding box.

5. The method for visualizing industrial furnace wall temperature field data according to claim 4, characterized in that, The specific formula for adjusting the color values ​​of the three channels based on the Russian Roulette algorithm model of light importance is as follows: Significance=[|R ∑ (k σ )-R P |+|G ∑ (k σ )-G P |+|B ∑ (k σ )-B P |+β·k σ ] -1 , Where C(k) is the RGB color value of the kth sample, P is the preset model probability based on the ray importance Russian roulette algorithm, and N T To preset the first sampling number, δ S The preset importance threshold is defined, where Significance represents importance information, and R... Σ (k σ ), G Σ (k σ ) and B Σ (k σ R represents the three-channel blended cumulative color recording value. P G P B P The expected color values ​​for each of the three channels, where β represents the number of samples recorded, k. σ The weight of the impact on importance information.

6. The method for visualizing industrial furnace wall temperature field data according to claim 5, characterized in that, Correcting the mixed and accumulated color value of the second projected ray after the second ray projection sampling process is terminated, based on the expected color value of the first projected ray, includes: Compare the red channel R of the current blended cumulative color value Σ (k T The red channel R with the expected color value P The numerical difference, denoted as ΔR = |R ∑ (k T )-R p |, if ΔR>ε, and R Σ (k T ) < R p -ε, then let the projected ray l start from the sampling point Continue forward and calculate the blended cumulative color value to obtain the sampling point. The cumulative color value is C ∑ (k T +1), continue forward until ΔR' = |R ∑ (k T +n r )-R p |≤ε, and record the corrected blending cumulative color value of the red channel at this time as C1=C ∑ (k T +n r If ΔR > ε, and is R Σ (k T )>R p +ε, then from the sampling point Backtrack and directly read sampling point P. kT-1 The cumulative color value is C ∑ (k T -1), and continue backtracking until ΔR'=|R ∑ (k T -n' r )-R p |≤ε, and record C1=C ∑ (k T -n' r If ΔR ≤ ε, then no correction is needed, and let C1 = C ∑ (k T ), where ε is the preset error threshold, k T n represents the current sampling count. r and n' r These represent the number of steps forward and backward when the correction is complete, respectively. Obtain the corrected mixed cumulative color values ​​of the green and blue channels, and obtain the corrected mixed cumulative color value of the second projected ray based on the average of the corrected mixed cumulative color values ​​of the three channels.

7. A data visualization system for the temperature field of an industrial furnace wall, the system comprising: The memory (10), the processor (20), and the computer program stored in the memory (10) and executable on the processor (20) are characterized in that the processor (20) implements the steps of the method according to any one of claims 1 to 6 when executing the computer program.

Images