A volatile kiln temperature field prediction method based on thermodynamic mechanism and infrared image data fusion

Abstract

The present application relates to the technical field of volatile kiln temperature field prediction, and specifically discloses a volatile kiln temperature field prediction method based on thermodynamic mechanism and infrared image data fusion, comprising the following steps: S1, a thermodynamic model is established according to the heat transfer and chemical reaction process of the volatile kiln, and a predicted temperature is obtained according to the thermodynamic model; S2, an infrared image of the head region of the volatile kiln is acquired, the infrared image is processed, and an actual solid material temperature is extracted; S3, a parameter optimization model is constructed by minimizing the error between the predicted temperature obtained by the thermodynamic model and the actual solid material temperature; S4, the model parameter value to be optimized is determined, and a parameter optimization result is obtained; and S5, the parameter optimization result is substituted into the thermodynamic model to obtain a fusion model, and a prediction result of the volatile kiln temperature field is obtained through the fusion model, so as to solve the problem that the temperature field of the volatile kiln is difficult to accurately predict in the prior art.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of volatile kiln temperature field prediction, and particularly discloses a volatile kiln temperature field prediction method based on thermodynamic mechanism and infrared image data fusion. BACKGROUND

[0002] The zinc oxide rotary volatile kiln is a core equipment for treating zinc leaching residue to recover zinc oxide dust, and is used for separating zinc oxide from a mixture of leaching residue and coke through a series of complex oxidation-reduction reactions under high-temperature reaction. The axial length of the volatile kiln is more than 60 meters, and the inner diameter is more than 4 meters, and the temperature field in the interior is the most important factor affecting technical indexes. However, it is almost impossible to measure the complete temperature field in real time due to the complex chemical mechanism, large reaction size, closed internal space and 360-degree rotation of the volatile kiln. In the actual industrial process, workers can only infer the temperature distribution in the kiln according to experience by observing the appearance and morphology of the kiln head flame image, and then controlling the temperature field. Since this control method lacks complete temperature field data as operation guidance, workers can only ensure product quality at the cost of excessive coke consumption.

[0003] The temperature field prediction method of the existing industrial rotary kiln is mostly based on the basic principles of mass conservation and energy conservation, and a pure thermodynamic model is established to realize the prediction of the temperature field. For common industrial rotary kilns such as cement and alumina, the fuel only provides heat for high-temperature reaction and hardly causes chemical reaction. However, the coke in the volatile kiln not only provides heat as a combustion agent, but also participates in chemical reaction as a reducing agent. Compared with other industrial kilns, the complex chemical reaction behavior of carbon elements in the volatile kiln makes the thermodynamic model more complex. In addition, the thermodynamic parameters of other rotary kilns will be obviously mismatched in the thermodynamic model of the volatile kiln. Due to the unique reaction mechanism of the volatile kiln, the method of relying only on pure thermodynamic modeling is difficult to accurately predict the temperature field.

[0004] Therefore, the present application provides a volatile kiln temperature field prediction method based on thermodynamic mechanism and infrared image data fusion in order to solve the above problems. SUMMARY

[0005] The present application aims to solve the problem that the temperature field of the zinc oxide rotary volatile kiln is difficult to accurately predict in the prior art.

[0006] In order to achieve the above-mentioned purpose, the basic scheme of the present application provides a volatile kiln temperature field prediction method based on thermodynamic mechanism and infrared image data fusion, comprising the following steps:

[0007] Step S1, according to the heat transfer and chemical reaction process of the volatilization kiln, a thermodynamic model involving chemical reaction heat is established, and a predicted temperature is obtained according to the thermodynamic model;

[0008] Step S2, an infrared image of the head region of the volatilization kiln is obtained, the infrared image is processed, and an actual solid material temperature is extracted;

[0009] Step S3, a parameter optimization model is constructed by minimizing the error between the predicted temperature obtained by the thermodynamic model and the actual solid material temperature;

[0010] Step S4, the model parameter value to be optimized is determined by using an optimization algorithm, and a parameter optimization result is obtained;

[0011] Step S5, the parameter optimization result is substituted into the thermodynamic model to obtain a fusion model, and a predicted result of the volatilization kiln temperature field is obtained through the fusion model.

[0012] Further, the thermodynamic model is represented as follows:

[0013]

[0014]

[0015] Q sh-a =Q g-ew +Q ew-es +Q cs-cw

[0016] Wherein,

[0017] Q g-∈s represents the heat transfer between the flue gas and the exposed solid material, including the convection term and the radiation term

[0018] Q g-ew represents the heat transfer between the flue gas and the exposed kiln wall, including the thermal convection term and the thermal radiation term

[0019] Q ew-es represents the heat transfer between the exposed kiln wall and the exposed material, including the thermal radiation term

[0020] Q cw-cs represents the heat transfer between the covered inner wall and the covered material, including the thermal radiation term and the thermal conduction term

[0021] Q sh-a represents the heat transfer between the shell and the external environment, including the thermal convection term and the thermal radiation term

[0022] F, C and T are mass flow, specific heat capacity and temperature, respectively;

[0023] Subscripts s and g represent solid material and flue gas, respectively.

[0024] Further, the step of processing the image in the step S2 is as follows:

[0025] Step S2.1, build an infrared thermal imaging system to obtain the infrared image of the volatile kiln head area in real time;

[0026] Step S2.2, pre-process the infrared image by using mathematical morphology operation to extract the background and foreground of the image;

[0027] Step S2.3, based on the target detection algorithm of YOLOv5s to obtain the position of invalid information in the image;

[0028] Step S2.4, fuse the image processing results of steps S2.2 and 2.3 to obtain the temperature pseudo-color map of the solid material.

[0029] Further, the specific formula for pre-processing the infrared image by using mathematical morphology operation is as follows:

[0030]

[0031] I 22 (m,n)=I 21 (m,n)○B

[0032] I 23 (m,n)=I 22 (m,n)●B

[0033] I 24 (m,n)=Bwareaopen(I 23 (m,n))

[0034] Wherein,

[0035] I1(m,n) is the original temperature pseudo-color map, each pixel coordinate (m,n) corresponds to a temperature value;

[0036] Ostu's represents the Otsu dynamic threshold segmentation method;

[0037] B is a structural element used to perform mathematical morphology operation on the image;

[0038] ○ and ● are open operation and close operation, respectively;

[0039] Bwareaopen is a function used to delete small area objects.

[0040] Further, in the step S3, the general thermodynamic parameters are determined by minimizing the error between the predicted temperature and the real temperature.

[0041] Further, in the step S3, the general thermodynamic parameters are the thermal conductivity of the material and the flue gas, and the emissivity of the material, the flue gas and the kiln wall.

[0042] Further, the parameter optimization model is expressed as follows:

[0043]

[0044]

[0045] In the formula, N1 is the sample size of the training set;

[0046] N2 is the number of temperature points of the solid material;

[0047] N3 is the number of effective temperature points in the infrared image;

[0048] d represents the depth information of the kiln head area;

[0049] L is the axial length of the kiln body;

[0050] T s (x) is the predicted temperature solved by the thermodynamic model at the axial position x;

[0051] Y t is the real solid material temperature extracted from the infrared image processing result;

[0052] is the predicted solid fluidized bed temperature, which is a function of the unknown thermodynamic parameter θ u.t and the unknown system parameter θ u.s ;

[0053] lower and upper respectively represent the upper limit and the lower limit of the optimized parameter.

[0054] Further, in the step S4, the flow rate, the kinematic viscosity and the density of the flue gas are used as the model parameters to be optimized.

[0055] The principle and effect of the scheme are:

[0056] 1. The present application aims at the problem that the temperature field of the volatilization kiln is difficult to measure in real time due to the complex chemical mechanism, large reaction size, closed internal space and 360° rotation of the volatilization kiln, and proposes a rotary volatilization kiln temperature field prediction method based on the fusion of thermodynamics and infrared images, which accurately and effectively realizes the soft measurement of the complete temperature field and provides operation guidance information for the low-carbon operation optimization of the zinc smelting volatilization kiln.

[0057] 2. The optimal general thermodynamic parameters in the application are determined in advance, and the number of unknown parameters that need to be optimized is greatly reduced. The pure thermodynamic model has poor local optimal solution due to too many unknown parameters, and the model is difficult to quickly converge to a satisfactory solution. Compared with the temperature field prediction method based on pure thermodynamics, the application has more superior prediction performance. BRIEF DESCRIPTION OF DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0059] Figure 1 The process flow diagram of the zinc oxide rotary volatilization kiln is shown;

[0060] Figure 2 The framework diagram of the volatilization kiln temperature field prediction method based on thermodynamic mechanism and infrared image data fusion proposed in the embodiments of the present application is shown;

[0061] Figure 3 The original temperature pseudo-color image in the volatilization kiln temperature field prediction method based on thermodynamic mechanism and infrared image data fusion proposed in the embodiments of the present application is shown;

[0062] Figure 4 The image after mathematical morphological operation in the volatilization kiln temperature field prediction method based on thermodynamic mechanism and infrared image data fusion proposed in the embodiments of the present application is shown;

[0063] Figure 5 The target detection result image based on YOLOv5s in the volatilization kiln temperature field prediction method based on thermodynamic mechanism and infrared image data fusion proposed in the embodiments of the present application is shown;

[0064] Figure 6 The temperature pseudo-color image of solid material in the volatilization kiln temperature field prediction method based on thermodynamic mechanism and infrared image data fusion proposed in the embodiments of the present application is shown;

[0065] Figure 7 The temperature field prediction result comparison diagram of the pure thermodynamic model and the fusion model is shown. DETAILED DESCRIPTION

[0066] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific embodiments, structures, features and effects according to the present application will be described in detail below in combination with the drawings and preferred embodiments.

[0067] A method for predicting temperature field of a volatilization kiln based on thermodynamic mechanism and infrared image data fusion, taking a zinc oxide rotary volatilization kiln of a zinc smelting plant as an example, has a process flow as shown in FIG. Figure 1 The mixture composed of zinc leaching residue and coke undergoes a series of complex redox reactions under high temperature reaction, and then zinc oxide is separated from the mixture. The volatilization kiln is different from other rotary kilns because the coke is not only used as fuel but also as an important reducing agent for reducing zinc. When the kiln body rotates, the solid material moves from the kiln tail area to the kiln head area. At the same time, the gas moves from the kiln head area to the kiln tail area under negative pressure. The solid material enters the high temperature reaction zone after drying and preheating. The zinc oxide dust generated during volatilization is collected in the dust settling chamber. The unreacted solid material is discharged from the kiln head area.

[0068] As shown in Figure 2 , the temperature field prediction of the zinc oxide rotary volatilization kiln mainly includes four parts: thermodynamic model construction, infrared image processing, parameter optimization and temperature field prediction, which are as follows:

[0069] Step S1: Based on careful analysis of the heat transfer and chemical reaction process of the volatilization kiln, a thermodynamic model involving chemical reaction heat is established, which is as follows:

[0070] Step S1.1: The following reaction equations show the core chemical reaction process involved in the temperature field prediction method. Part of the coke provides heat as a fuel agent, and the other part reacts with ZnFe2O4 and ZnO to generate zinc vapor as a reducing agent. Zinc oxide is first reduced to zinc vapor, and then reacts with oxygen to generate zinc oxide dust.

[0071] C + O2 = CO2

[0072] 3ZnFe2O4 + 4C + 2O2 = 2Fe3O4 + 3Zn + 4CO2

[0073] 2ZnO + 2C + O2 = 2Zn(g) + 2CO2

[0074] 2Zn(g) + O2 = 2ZnO

[0075] Step S1.2: The chemical reaction heat is part of the heat composition in the thermodynamic model, which is related to the change rate of the component and the enthalpy, and can be expressed as:

[0076] Q chem，g = Σ r1(x) ΔH1

[0077] Q chem，s = ∑ r1(x) ΔH i , i = 2, 3

[0078] Where Q chem，g , Q chem，srespectively, r1, r2, r3 represent the component change rate of C, ZnFe2O4 and ZnO respectively, and ΔH represents the corresponding enthalpy.

[0079] Step S1.3: Heat transfer is another part of heat composition in the thermodynamic model, mainly including heat convection, heat radiation and heat conduction. Based on the law of conservation of energy, the thermodynamic model for temperature field prediction is expressed as follows:

[0080]

[0081]

[0082] Q sh-a = Q g-ew + Q ew-es + Q cs-cw

[0083] Where Q g-∈s represents the heat transfer between the flue gas and the exposed solid material, including the convection term and the radiation term Q g-ew represents the heat transfer between the flue gas and the exposed kiln wall, including the heat convection term and the heat radiation term Q ew-es represents the heat transfer between the exposed kiln wall and the exposed material, including only the heat radiation term Q cw-cs represents the heat transfer between the covered inner wall and the covered material, including the heat radiation term and the heat conduction term Q sh-a represents the heat transfer between the shell and the external environment, including the heat convection term and the heat radiation term F, C and T are mass flow, specific heat capacity and temperature respectively. Subscripts s and g represent solid material and flue gas respectively.

[0084] Step S2: Real-time acquisition of infrared images of the volatile kiln head area by using an infrared thermal imaging system, and on this basis, real and accurate solid material temperature is extracted by image processing method, wherein the steps of processing the infrared image are as follows:

[0085] Step S2.1: Build an infrared thermal imaging system which can acquire infrared images in real time;

[0086] Step S2.2: The infrared image is preprocessed by using mathematical morphology operation, so as to extract the background and foreground of the image, and the specific formula is as follows:

[0087]

[0088] I 22 (m,n)=I 21 (m,n)○B

[0089] I 23 (m,n)=I 22 (m,n)●B

[0090] I 24 (m,n)=Bwareaopen(I 23 (m,n))

[0091] Where I1(m,n) is the original temperature pseudo-color image, each pixel coordinate (m,n) corresponds to a temperature value. Ostu’s represents the Otsu dynamic threshold segmentation method. B is the structure element used to perform mathematical morphology operation on the image. ○ and ● are open operation and close operation respectively, and Bwareaopen is a function used to delete small area objects. 24 (m,n) is a segmented image composed of background and foreground, and the pixel positions corresponding to the values of the foreground and background are taken as 1 and 0 respectively.

[0092] Step S2.3: obtaining the position of invalid information in the image based on the target detection algorithm of YOLOv5s;

[0093] Step S2.4: obtaining the temperature pseudo-color image of the solid material by fusing the image processing results of steps S2.2 and 2.3, and the specific formula is as follows:

[0094] I4(m,n)=I1(m,n)*I 24 (m,n)*I3(m,n)

[0095] Where I4 is the temperature pseudo-color image of the solid material. I3 is the image processing result obtained based on YOLOv5s, and the value of the pixel position where the invalid information is located is 0, and the value of other positions is 1.

[0096] This embodiment selects a representative image to illustrate the effectiveness of the method, I1(m,n) represents the original temperature pseudo-color image, which is as shown in Figure 3 I 24 (m,n) is the result obtained after the mathematical morphology operation on I1(m,n), as shown in Figure 4 , which is composed of background and foreground. I3 is the image processing result obtained based on YOLOv5s, as shown in Figure 5 . By fusing the results of image processing in steps S2.2 and 2.3, the temperature pseudo-color image I4 of the solid material is obtained, and the result is shown in Figure 6 .

[0097] Step S3: Establish a parameter optimization model to extract general thermodynamic parameters applicable to any volatile kiln by minimizing the error between the predicted temperature and the real temperature. The predicted temperature in the model is calculated by the thermodynamic model, and the real temperature is extracted by the infrared image processing method; the specific formula is as follows:

[0098]

[0099]

[0100] Where N1 is the sample size of the training set, N2 is the number of temperature points of the solid material, and N3 is the number of effective temperature points in the infrared image. d is used to represent the depth information of the kiln head area, and L is the axial length of the kiln body. s (x) is the predicted temperature solved by the thermodynamic model at the axial position x. t is the real solid material temperature extracted from the infrared image processing result. is the predicted solid fluidized bed temperature, which is a function of unknown thermodynamic parameters u.t and unknown system parameters u.s . lower and upper represent the upper and lower limits of the optimized parameters, respectively.

[0101] For the parameter optimization model described in step S3, a total of 200 data in the actual industrial process were collected, of which the first 150 were used to optimize the unknown parameters in the thermodynamic model, and the last 50 were used to test the accuracy of the parameter optimization model. The results after parameter optimization are shown in Table 1, where λ s and λ g represent the thermal conductivity of the material and the flue gas, respectively, ε s , ε w and ε g represent the emissivity of the material, the flue gas and the kiln wall, respectively. These extracted general thermodynamic parameters are applicable to all volatile kilns. g , u g and ρ g represent the flow rate, kinematic viscosity and density of the flue gas, respectively, which will change with the process conditions in different volatile kilns.

[0102] Table 1: Optimization results of unknown parameters in the thermodynamic model

[0103]

[0104] Step S4: apply the general thermodynamic parameters extracted in step S3 to another volatilization kiln with different physical dimensions, the volatilization kiln in step S4 has different physical dimensions compared with the volatilization kiln in step S3, but they have the same heat transfer mechanism. In order to determine the unknown parameters to be optimized in the temperature field prediction method based on pure thermodynamics (pure thermodynamics model) and the temperature field prediction method of fusion of thermodynamic mechanism and infrared image (fusion model), two comparative experiments of the two models are designed in this embodiment to show the effect of the present application, as follows:

[0105] Step S4.1: for the pure thermodynamics model, since the general thermodynamic parameters are not determined in advance, the total number of parameters to be optimized is 8, which are λ s , λ g , ε s , ε w , ε g , v g , u g , ρ g ;

[0106] Step S4.2: for the fusion model, since the general thermodynamic parameters are determined in advance by the infrared image processing method, as shown in Table 1. Therefore, the total number of parameters to be optimized of the fusion model is only 3, which are v g , u g , ρ g ;

[0107] Step S4.3: identify the model parameters to be optimized in the two models by using the optimization algorithm, and the optimization results are shown in Table 2.

[0108] Table 2: parameter optimization results of temperature field prediction model

[0109]

[0110] Step S5: substitute the parameter optimization results into the thermodynamic model to obtain the fusion model, the fusion model is based on the thermodynamic model, and the unknown parameters in the thermodynamic model are determined in advance by the parameter optimization model. The thermodynamic model after parameter optimization is the fusion model, and the prediction result of the temperature field of the volatilization kiln can be obtained through the fusion model, so as to realize the accurate prediction of the temperature field of the volatilization kiln.

[0111] In this embodiment, the parameter optimization results in Table 2 are substituted into the thermodynamic model, so as to obtain the temperature field prediction results of the pure thermodynamics model and the fusion model, as shown in Table 3. Figure 7 Correspondingly, Table 3 shows that compared with the pure thermodynamics model, the fusion model of this embodiment has smaller MSE and MARE, which means that the fusion model has better temperature field prediction accuracy.

[0112] Table 3: Temperature field prediction results

[0113]

[0114] where MSE represents the mean square error, MARE represents the mean relative error absolute value, and the specific calculation formula is as follows:

[0115]

[0116]

[0117] where n represents the sample number, y i and respectively represent the true value and the predicted value of the i-th sample.

[0118] From the above results, compared with the temperature field prediction method based on pure thermodynamics, the present application has superior prediction performance, and the reason is that the optimal general thermodynamic parameters of the fusion model are determined in advance, and the number of unknown parameters to be optimized is greatly reduced. The pure thermodynamic model has poor local optimal solution due to too many unknown parameters, and the model is difficult to quickly converge to a satisfactory solution.

[0119] The present application aims at the problem that the chemical mechanism of the volatilization kiln is complex, the reaction size is large, the internal space is closed, and the 360° rotation leads to the difficulty in real-time measurement of the temperature field, and proposes a rotary volatilization kiln temperature field prediction method based on thermodynamics and infrared image fusion, which accurately and effectively realizes the soft measurement of the complete temperature field, and provides operation guidance information for the low-carbon operation optimization of the zinc smelting volatilization kiln.

[0120] The above is only a preferred embodiment of the present application, and does not limit the present application in any form. Although the present application has been disclosed as above with a preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to obtain equivalent embodiments with equivalent changes, without departing from the technical solution of the present application. Any modification, change, equivalent change and modification of the above embodiments made according to the technical essence of the present application are still within the scope of the technical solution of the present application.

Claims

1. A method for predicting the temperature field of a volatilization kiln based on the fusion of thermodynamic mechanisms and infrared image data, characterized in that, Includes the following steps: Step S1: Based on the heat transfer and chemical reaction process of the volatilization kiln, establish a thermodynamic model involving the heat of chemical reaction, and obtain the predicted temperature based on the thermodynamic model; Step S2: Obtain an infrared image of the kiln head area of ​​the volatilization kiln, process the infrared image and extract the actual solid material temperature. The image processing steps are as follows: Step S2.1: Build an infrared thermal imaging system to acquire infrared images of the kiln head area of ​​the volatilization kiln in real time; Step S2.2: Preprocess the infrared image using mathematical morphology operations to extract the background and foreground of the image; Step S2.3: Use the YOLOv5s object detection algorithm to obtain the location of invalid information in the image; Step S2.4: The image processing results of steps S2.2 and 2.3 are fused to obtain a pseudo-color temperature map of the solid material; The specific formula for preprocessing infrared images using mathematical morphology operations is as follows: ； ； ； ； in, This is the original temperature pseudo-color image, where each pixel coordinate (m,n) corresponds to a temperature value; This represents the Otsu dynamic threshold segmentation method; B is a structuring element used to perform mathematical morphological operations on images; and These are opening and closing operations, respectively. It is a function used to delete small regions of objects; Step S3: Construct a parameter optimization model by minimizing the error between the predicted temperature obtained from the thermodynamic model and the actual temperature of the solid material. The formula for the parameter optimization model is as follows: ； ； In the formula, It is the sample size of the training set; This refers to the number of temperature points of the solid material. This is the number of valid temperature points in the infrared image; This indicates the depth information of the kiln head area; It is the axial length of the kiln body; It is the predicted temperature obtained by the thermodynamic model at the axial position x; It is the actual temperature of solid materials extracted from infrared image processing results; The predicted temperature of the solid fluidized bed is given by unknown thermodynamic parameters. and unknown system parameters The function; and These represent the upper and lower limits of the parameter being optimized, respectively; Step S4: Use the optimization algorithm to determine the model parameter values ​​that need to be optimized and obtain the parameter optimization results; Step S5: Substitute the parameter optimization results into the thermodynamic model to obtain the fusion model, and use the fusion model to obtain the predicted results of the volatilization kiln temperature field.

2. The method for predicting the temperature field of a volatilization kiln based on the fusion of thermodynamic mechanisms and infrared image data according to claim 1, characterized in that, The thermodynamic model is represented as follows: ； ； ； in, This represents the heat transfer between flue gas and exposed solid material, including convection terms. and radiation items ; This refers to the heat transfer between the flue gas and the exposed kiln wall, including the heat convection term. and thermal radiation ; This refers to the heat transfer between the exposed kiln wall and the exposed materials, including the thermal radiation term. ; This refers to the heat transfer between the covered inner wall and the covered material, including the thermal radiation term. and heat conduction term ; This represents the heat transfer between the casing and the external environment, including the heat convection term. and thermal radiation ; F, C, and T are the mass flow rate, specific heat capacity, and temperature, respectively. The subscripts s and g represent solid materials and flue gas, respectively.

3. The method for predicting the temperature field of a volatilization kiln based on the fusion of thermodynamic mechanisms and infrared image data according to claim 1, characterized in that, In step S3, the general thermodynamic parameters are determined by minimizing the error between the predicted temperature and the actual temperature.

4. The method for predicting the temperature field of a volatilization kiln based on the fusion of thermodynamic mechanisms and infrared image data according to claim 1, characterized in that, In step S3, the thermal conductivity of the material and flue gas, as well as the emissivity of the material, flue gas, and kiln wall, are used as general thermodynamic parameters.

5. The method for predicting the temperature field of a volatilization kiln based on the fusion of thermodynamic mechanisms and infrared image data according to claim 1, characterized in that, In step S4, the flow rate, kinematic viscosity, and density of the flue gas are the model parameters that need to be optimized.

Images