A method for reconstructing a gas temperature field using thermocouple measurement correction

Abstract

A gas temperature field reconstruction method using thermocouple measurement correction is disclosed. First, an Inventor 3D model and an ANSYS temperature field model are established for the gas temperature field. Then, a mapping relationship is established between the control parameters of the gas temperature field and the boundary conditions of the ANSYS temperature field model. Based on the principle of maximizing the influence of fuzzy boundary conditions on the temperature field reconstruction results, experiments are designed for precise boundary conditions and experimental data are obtained. Next, given the boundary conditions of the ANSYS temperature field model, the model is run to obtain the temperature field reconstruction results. A model quality assessment algorithm is run to calculate the fitting degree between the measured curve and the simulation curve, obtaining the algorithm fitting parameters. The algorithm fitting parameters are selected to form the loss function of the gradient descent algorithm. The learning rate, termination condition, and initial iteration parameters are determined. The gradient descent algorithm is run until convergence. Finally, the final iteration parameters are taken as the correction result of the fuzzy boundary conditions of the ANSYS temperature field model. This invention improves the spatiotemporal resolution of the temperature field reconstruction results.

Description

Technical Field

[0001] This invention relates to the field of temperature measurement and fluid simulation technology, specifically to a method for reconstructing a gas temperature field using thermocouple measurement and correction. Background Technology

[0002] With the continuous expansion of modern industrial thermal energy utilization and the deepening development of energy conservation and consumption reduction, traditional single-point or local temperature measurement methods can no longer meet the needs of thermal energy application fields in process design, condition monitoring, and control optimization. As the main pathway for heat energy generation and transfer, the gas temperature field is a key object in research on temperature measurement, modeling analysis, and regulation and control.

[0003] Gases, characterized by low specific heat capacity and high fluidity, exhibit intensified molecular thermal motion when used as fuel or heat transfer media, resulting in a dynamically changing, unsteady temperature field. This field is characterized by rapid temperature changes, large spatial temperature gradients, and susceptibility to external disturbances. Techniques for obtaining gas temperature distribution through interpolation of temperature measurement data are insufficient to accurately describe the temperature change process on a temporal scale, nor can they precisely represent the temperature gradient variation on a spatial scale. Furthermore, the susceptibility of gases to external disturbances places stringent requirements on the placement of temperature sensors, increasing the cost and difficulty of acquiring temperature distribution data. Therefore, to address the rapidly changing gas temperature field, it is necessary to research novel temperature field reconstruction methods to accurately describe and predict the distribution and changes of the gas temperature field on both temporal and spatial scales.

[0004] Numerical simulation can obtain comprehensive gas temperature distribution information with good spatiotemporal resolution. It is a technique that uses numerical methods to solve for the temperature distribution across the entire computational domain by establishing three-dimensional unsteady mathematical models of different research objects and combining information such as initial conditions, boundary conditions, and physical properties. On the one hand, numerical models require highly accurate boundary conditions, but for actual gas temperature field problems, it is difficult to provide accurate boundary conditions, which leads to large errors in the simulation results. On the other hand, the simulation results of numerical models lack effective evaluation methods. Theoretically, the mesh of numerical models can be infinitely small according to the required solution accuracy, ensuring the temporal and spatial resolution of the simulation results. However, it is difficult to obtain high spatiotemporal resolution temperature measurement data under experimental conditions. Therefore, simulation results cannot be compared with experimental results, making the evaluation of the fitting quality between the numerical model and the actual object relatively subjective and unable to determine the direction of optimization of the numerical model. By introducing high-quality thermocouple temperature measurement information and establishing effective model quality evaluation indicators, the accuracy of temperature field reconstruction by numerical simulation can be significantly improved. However, this technique has not yet been applied in temperature field reconstruction technology. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, the present invention aims to provide a gas temperature field reconstruction method using thermocouple measurement and correction, which improves the spatiotemporal resolution of the temperature field reconstruction results while ensuring the accuracy and reliability of the temperature field reconstruction.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] A method for reconstructing a gas temperature field using thermocouple measurement correction includes the following steps:

[0008] Step D1: Based on the specific structure of the gas temperature field, establish an Inventor 3D model; based on the Inventor 3D model, and combined with the control parameters of the gas temperature field, establish an ANSYS temperature field model; the Inventor 3D model includes three components: the inlet, internal structural features, and the outlet of the gas temperature field, while the ANSYS temperature field model includes five modules: Geometry, Fluent, Transient Thermal, System Coupling, and Results.

[0009] Step D2: Analyze the modeling process in Step D1, and establish the mapping relationship between the gas temperature field control parameters and the boundary conditions of the ANSYS temperature field model; based on whether the control parameters can be accurately measured or calculated, classify the boundary conditions of the ANSYS temperature field model into precise boundary conditions. and fuzzy boundary conditions Two categories; among them, the mapping relationship includes three types: one-to-one, one-to-many, and many-to-one, with fuzzy boundary conditions. This is the source of error in temperature field reconstruction;

[0010] Step D3: To best reflect fuzzy boundary conditions The principle is to take into account the impact on the temperature field reconstruction results and the precise boundary conditions. Design experiments, conduct experiments, and obtain experimental data;

[0011] Step D4: Given the boundary conditions of the ANSYS temperature field model, run the model and obtain the temperature field reconstruction results; wherein, the initial values ​​of the boundary conditions are randomly given in the first calculation, and the subsequent calculations are given by the gradient descent algorithm;

[0012] Step D5: Run the model quality assessment algorithm on the measurement curve and simulation curves The goodness of fit is calculated to obtain the intercept estimate of the algorithm's fitting parameters. Slope estimate Correlation coefficient Among them, the measurement curve Derived from experimental data, it is a temperature curve over time measured by thermocouples, and a simulation curve. It is a temperature curve that changes over time, extracted from the temperature field reconstruction results based on the thermocouple measurement location;

[0013] Step D6: Using the model boundary condition correction algorithm, select the fitting parameters of the model quality assessment algorithm from step D5. , , The loss function that constitutes the gradient descent algorithm Determine the learning rate, termination condition, and initial iteration parameters for the gradient descent algorithm; run the gradient descent algorithm until it converges, and use the final iteration parameters as the fuzzy boundary conditions for the ANSYS temperature field model. The correction results.

[0014] The experimental data acquisition device in step D3 is a thermocouple.

[0015] The model quality assessment algorithm described in step D5 includes the following steps:

[0016] Step Q1: Assume the curve Sampling rate ,curve Sampling rate ,and ; at the lowest sampling rate Based on the curve Perform resampling;

[0017] Resampling interval The possible values ​​are as follows:

[0018] (1)

[0019] The formula for calculating the resampling time is as follows:

[0020] (2)

[0021] in, It is the sequence number of the sampling point in the resampled sequence. , It is the first Each resampling time, It is the starting point of the resampling time;

[0022] With curves Sampling time sequence Based on the following formula, the curve is interpolated using linear interpolation. Perform resampling to achieve curve and resampling curve Alignment of sampling time:

[0023] (3)

[0024] in, It is a resampled signal. It is the distance from the resampling time The most recent point in time, It is the distance from the resampling time The most recent point in time;

[0025] After resampling, the resampling curve Having curves Same sampling time and sampling rate;

[0026] Step Q2: Reassemble , And the regression parameters were calculated using univariate linear regression;

[0027] Construct a new function as follows:

[0028] (4)

[0029] in, It is the independent variable, defined as follows:

[0030] (5)

[0031] Quantitative assessment using univariate linear regression With a straight line The degree of difference is expressed by the following regression equation:

[0032] (6)

[0033] in, Let be the slope of the regression equation, and its estimated value is . ; The intercept of the regression equation is estimated to be... ; is the error variable in the regression equation.

[0034] The model boundary condition correction algorithm described in step D6 includes the following steps:

[0035] Step B1: Construct the loss function ;

[0036] The model quality assessment algorithm described in steps Q1 and Q2 is used to select... and The two metrics constitute the loss function as follows:

[0037] (7)

[0038] in, Fuzzy boundary conditions from the ANSYS temperature field model composition, The definition is as follows:

[0039] (8)

[0040] in, , It is a constant;

[0041] Step B2: Determine the initial parameters for the iteration;

[0042] First, based on the range of values ​​for the temperature field model parameters, the initial parameter values ​​are given as follows:

[0043] (9)

[0044] The initial gradient direction is determined as follows:

[0045] (10)

[0046] Determine the step size for the gradient descent algorithm and termination threshold u ;

[0047] Step B3: Determine the gradient direction at the current position;

[0048] Calculate the main model separately and gradient model Obtain the specific loss function value and apply the following formula to calculate the gradient direction of the ANSYS temperature field model.

[0049] (11)

[0050] in, It is the first One parameter that needs to be corrected. This is the total number of parameters that need to be corrected. The definition is as follows:

[0051] (12)

[0052] in, It is the first The parameter of the first iteration It is a constant that is less than 1 and greater than 0;

[0053] Step B4: Determine the distance to descend from the current position;

[0054] The descent distance for all parameters is calculated sequentially using the following formula:

[0055] (13)

[0056] Step B5: Determine if the iteration termination condition has been met;

[0057] Determine if the distance of gradient descent for all parameters is less than the termination threshold. u If all values ​​are less than 1, the algorithm terminates. This is the final result; otherwise, proceed to the next step.

[0058] Step B6: Update parameters and begin the next iteration;

[0059] Based on the basic iterative formula of the gradient descent algorithm, all parameter values ​​are updated using the following formula, and the process returns to step B3.

[0060] (14)

[0061] The beneficial effects of this invention are as follows:

[0062] 1. This invention establishes a mapping relationship between the control parameters of the gas temperature field and the boundary conditions of the temperature field model, and uses the gradient descent algorithm and experimental data to correct the temperature field model, which effectively improves the reconstruction accuracy and stability of the temperature field model.

[0063] 2. This invention combines the high-quality temperature measurement data characteristics of thermocouples with the high spatiotemporal resolution advantage of numerical models, effectively improving the spatiotemporal resolution and accuracy of temperature field reconstruction. Attached Figure Description

[0064] Figure 1 This is a flowchart of the present invention.

[0065] Figure 2 This is a three-dimensional model of the temperature field of the hydrogen-oxygen flame gas in Embodiment 1 of the present invention.

[0066] Figure 3 This is the hydrogen-oxygen flame temperature field model of Embodiment 1 of the present invention.

[0067] Figure 4 This is the thermocouple temperature measurement result of Embodiment 1 of the present invention.

[0068] Figure 5 This is the fitting effect of the hydrogen-oxygen flame temperature field model before correction in Embodiment 1 of the present invention.

[0069] Figure 6 This is the fitting effect after correction of the hydrogen-oxygen flame temperature field model in Embodiment 1 of the present invention;

[0070] Figure 7This is a comparison between the temperature field model prediction results of the hydrogen-oxygen flame in Embodiment 1 of the present invention and the measurement results of the thermal imager. Detailed Implementation

[0071] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0072] Example 1 is a reconstruction of the temperature field of an oxyhydrogen flame gas. The main purpose is to reduce the error of the reconstruction results, so as to obtain a high spatiotemporal resolution temperature field reconstruction result while ensuring the accuracy of the temperature field reconstruction.

[0073] Reference Figure 1 A method for reconstructing a gas temperature field using thermocouple measurement correction includes the following steps:

[0074] Step D1: Based on the specific structure of the gas temperature field, establish an Inventor 3D model; based on the Inventor 3D model, and combined with the control parameters of the gas temperature field, establish an ANSYS temperature field model; the Inventor 3D model includes three components: the inlet, internal structural features, and the outlet of the gas temperature field, while the ANSYS temperature field model includes five modules: Geometry, Fluent, Transient Thermal, System Coupling, and Results.

[0075] In this embodiment, the Inventor 3D model and ANSYS temperature field model of the oxyhydrogen flame gas temperature field are established as follows: Figure 2 , Figure 3 As shown; in the Inventor 3D model, the top of the cylinder is the entrance, the bottom of the cylinder is the exit, and the interior of the cylinder models the external features of the thermocouple.

[0076] Step D2: Analyze the modeling process in Step D1, and establish the mapping relationship between the gas temperature field control parameters and the boundary conditions of the ANSYS temperature field model; based on whether the control parameters can be accurately measured or calculated, classify the boundary conditions of the ANSYS temperature field model into precise boundary conditions. and fuzzy boundary conditions Two categories; among them, the mapping relationship includes three types: one-to-one, one-to-many, and many-to-one, with fuzzy boundary conditions. It is the main source of error in temperature field reconstruction;

[0077] In this embodiment, the mapping relationship between the control parameters of the oxyhydrogen flame gas temperature field and the boundary conditions of the oxyhydrogen flame temperature field model is established as shown in Table 1. Furthermore, based on whether the control parameters can be accurately measured and given, the boundary conditions are classified into fuzzy boundary conditions. and precise boundary conditions ;

[0078] Table 1

[0079]

[0080] Step D3: To best reflect fuzzy boundary conditions The principle is to take into account the impact on the temperature field reconstruction results and the precise boundary conditions. Design experiments, conduct experiments, and obtain experimental data;

[0081] In this embodiment, the experimental data acquisition device is a thermocouple. To best reflect the effects of thermocouple response time, initial velocity of the hydrogen-oxygen mixture, and turbulence intensity, angular velocity was selected. With a sampling rate of 0.8 rad / s and a sampling frequency of 51200 Hz, and a Butterworth low-pass filter with a passband cutoff frequency of 50 Hz and a stopband cutoff frequency of 100 Hz, the thermocouple measurement curve is obtained as follows: Figure 4 As shown;

[0082] Step D4: Given the boundary conditions of the ANSYS temperature field model, run the model and obtain the temperature field reconstruction results; wherein, the initial values ​​of the boundary conditions are randomly given in the first calculation, and the subsequent calculations are given by the gradient descent algorithm;

[0083] In the first run of the gradient descent algorithm in this embodiment, the initial mixed gas velocity of the hydrogen-oxygen flame temperature field model is given as 16 m / s, the initial mixed gas turbulence intensity is 5.00%, and the initial average diameter of the thermocouple sensing nodes is 200 mm. ;

[0084] Step D5: Run the model quality assessment algorithm on the measurement curve and simulation curves The goodness of fit is calculated to obtain the intercept estimate of the algorithm's fitting parameters. Slope estimate Correlation coefficient Among them, the measurement curve Derived from experimental data, it is a temperature curve over time measured by thermocouples, and a simulation curve. It is a temperature curve that changes over time, extracted from the temperature field reconstruction results based on the thermocouple measurement location;

[0085] The model quality assessment algorithm in this embodiment specifically includes the following steps:

[0086] Step Q1: For the measurement curve and simulation curves The sampling rates are 51200Hz and 333.3Hz, respectively. Using the lowest sampling rate of 333.3Hz as the benchmark, the measurement curve... Perform resampling;

[0087] Resampling interval as follows:

[0088] (1)

[0089] The resampling time is calculated as follows:

[0090] (2)

[0091] in, It is the sequence number of the sampling point in the resampled sequence. , It is the first Each resampling time, It is the starting point of the resampling time;

[0092] set up After resampling, it is recorded as The linear interpolation is calculated as follows:

[0093] (3)

[0094] in, It is a resampled curve. It is the distance from the resampling time The most recent point in time, It is the distance from the resampling time The most recent point in time;

[0095] After resampling, the resampling curve It has simulation curves Same sampling time and sampling rate;

[0096] Step Q2: Reassemble , And the regression parameters were calculated using univariate linear regression;

[0097] Construct a new function as follows:

[0098] (4)

[0099] in, It is the independent variable, defined as follows:

[0100] (5)

[0101] The regression equation for univariate linear regression is as follows:

[0102] (6)

[0103] in, Let be the slope of the regression equation, and its estimated value is . ; The intercept of the regression equation is estimated to be... ; The error variable in the regression equation;

[0104] Step D6: Using the model boundary condition correction algorithm, select the fitting parameters of the model quality assessment algorithm from step D5. , , The loss function that constitutes the gradient descent algorithm The specific expression is described in step B1 of this embodiment; the initial learning rate for the mixed airflow velocity is 0.01, and the termination threshold is 0.1 m / s; the initial learning rate for the mixed airflow turbulence intensity is 0.002, and the termination threshold is 0.01%; the initial learning rate for the average diameter of the thermocouple temperature measuring node is 0.001, and the termination threshold is 5. After 11 iterations, the algorithm converged, and the final iteration parameters are as follows:

[0105]

[0106] The boundary condition correction algorithm includes the following steps:

[0107] Step B1: Construct the loss function

[0108] Based on the model quality assessment algorithm described in steps Q1 and Q2 of this embodiment, the following algorithm is selected: and The two metrics constitute the loss function as follows:

[0109] (7)

[0110] in, Fuzzy boundary conditions from the ANSYS temperature field model composition, In this embodiment, the following definitions apply:

[0111] (8)

[0112] Step B2: Determine the initial parameters for the iteration. First, based on the reasonable range of values ​​for the ANSYS temperature field model parameters, the initial parameter values ​​are given as follows:

[0113] (9)

[0114] The initial gradient direction is determined as follows:

[0115] (10)

[0116] Determine the step size for the gradient descent algorithm and termination threshold u For the specific value, please refer to step D6 of this embodiment;

[0117] Step B3: Determine the gradient direction at the current position;

[0118] Calculate the main model separately and gradient model Obtain the specific loss function value and apply the following formula to calculate the gradient direction of the hydrogen-oxygen flame temperature field model.

[0119] (11)

[0120] in, It is the first One parameter that needs to be corrected. In this embodiment, it is 3. The definition is as follows:

[0121] (12)

[0122] in, It is the first The parameter of the first The next iteration;

[0123] Step B4: Determine the distance to descend from the current position;

[0124] The descent distance for all parameters is calculated sequentially using the following formula:

[0125] (13)

[0126] Step B5: Determine if the iteration termination condition has been met;

[0127] Determine if the distance of gradient descent for all parameters is less than the termination threshold. u If all values ​​are less than 1, the algorithm terminates. This is the final result; otherwise, proceed to the next step.

[0128] Step B6: Update parameters and begin the next iteration;

[0129] Based on the basic iterative formula of the gradient descent algorithm, all parameter values ​​are updated using the following formula, and the process returns to step B3.

[0130] (14)

[0131] Reference Figure 5 , Figure 6Before correction, the loss function value of the oxyhydrogen flame temperature field model was 51.063, and after correction, the loss function value was 35.133, a decrease of 31.2%. This shows that the gradient descent algorithm can effectively improve the reconstruction accuracy of the ANSYS temperature field model.

[0132] Reference Figure 7 The results of thermal imaging measurements were compared with those of the corrected oxyhydrogen flame temperature field model. The loss function value was 19.735, indicating that the reconstruction results of the temperature field reconstruction method involved in this invention have good accuracy.

Claims

1. A gas temperature field reconstruction method using thermocouple measurement correction, characterized by, Includes the following steps: Step D1: Based on the specific structure of the gas temperature field, establish an Inventor 3D model; based on the Inventor 3D model, and combined with the control parameters of the gas temperature field, establish an ANSYS temperature field model; the Inventor 3D model includes three components: the inlet, internal structural features, and the outlet of the gas temperature field, while the ANSYS temperature field model includes five modules: Geometry, Fluent, Transient Thermal, System Coupling, and Results. Step D2: analyze the modeling process of step D1, establish the mapping relationship between the gas temperature field control parameters and the boundary conditions of the ANSYS temperature field model; according to whether the control parameters can be accurately measured or calculated, the boundary conditions of the ANSYS temperature field model are divided into accurate boundary conditions and fuzzy boundary conditions Two categories; wherein, the mapping relationship includes one-to-one, one-to-many and many-to-one, and the fuzzy boundary condition is the source of temperature field reconstruction error; Step D3: To best reflect fuzzy boundary conditions The principle is to take into account the impact on the temperature field reconstruction results and the precise boundary conditions. Design experiments, conduct experiments, and obtain experimental data; Step D4: Given the boundary conditions of the ANSYS temperature field model, run the model and obtain the temperature field reconstruction results; wherein, the initial values ​​of the boundary conditions are randomly given in the first calculation, and the subsequent calculations are given by the gradient descent algorithm; Step D5: Run the model quality assessment algorithm on the measurement curve and simulation curves The goodness of fit is calculated to obtain the intercept estimate of the algorithm's fitting parameters. Slope estimate Correlation coefficient Among them, the measurement curve Derived from experimental data, it is a temperature curve over time measured by thermocouples, and a simulation curve. It is a temperature curve that changes over time, extracted from the temperature field reconstruction results based on the thermocouple measurement location; Step D6: Using the model boundary condition correction algorithm, select the fitting parameters of the model quality assessment algorithm from step D5. , , The loss function that constitutes the gradient descent algorithm Determine the learning rate, termination condition, and initial iteration parameters for the gradient descent algorithm; run the gradient descent algorithm until it converges, and use the final iteration parameters as the fuzzy boundary conditions for the ANSYS temperature field model. The correction results.

2. The method according to claim 1, characterized in that, The experimental data acquisition device in step D3 is a thermocouple.

3. The method according to claim 1, characterized in that, The model quality assessment algorithm described in step D5 includes the following steps: Step Q1: Assume the curve Sampling rate ,curve Sampling rate ,and ; at the lowest sampling rate Based on the curve Perform resampling; Resampling interval The possible values ​​are as follows: （1） The formula for calculating the resampling time is as follows: （2） in, It is the sequence number of the sampling point in the resampled sequence. , It is the first Each resampling time, It is the starting point of the resampling time; With curves Sampling time sequence Based on the following formula, the curve is interpolated using linear interpolation. Resampling is performed to achieve the curve and resampling curve Alignment of sampling time: （3） in, It is a resampled signal. It is the distance from the resampling time The most recent point in time, It is the distance from the resampling time The most recent point in time; After resampling, the resampling curve Having curves Same sampling time and sampling rate; Step Q2: Reassemble , And the regression parameters were calculated using univariate linear regression; Construct a new function as follows: （4） in, It is the independent variable, defined as follows: （5） Quantitative assessment using univariate linear regression With a straight line The degree of difference is expressed by the following regression equation: （6） in, Let be the slope of the regression equation, and its estimated value is . ; The intercept of the regression equation is estimated to be... ; is the error variable in the regression equation.

4. The method according to claim 3, characterized in that, The model boundary condition correction algorithm described in step D6 includes the following steps: Step B1: Construct the loss function ; The model quality assessment algorithm described in steps Q1 and Q2 is used to select... and The two metrics constitute the loss function as follows: （7） in, Fuzzy boundary conditions from the ANSYS temperature field model composition, The definition is as follows: （8） in, , It is a constant; Step B2: Determine the initial parameters for the iteration; First, based on the range of values ​​for the temperature field model parameters, the initial parameter values ​​are given as follows: （9） The initial gradient direction is determined as follows: （10） Determine the step size for the gradient descent algorithm and termination threshold u ; Step B3: Determine the gradient direction at the current position; Calculate the main model separately and gradient model Obtain the specific loss function value and apply the following formula to calculate the gradient direction of the ANSYS temperature field model. （11） in, It is the first One parameter that needs to be corrected. This is the total number of parameters that need to be corrected. The definition is as follows: （12） in, It is the first The parameter of the first iteration It is a constant that is less than 1 and greater than 0; Step B4: Determine the distance to descend from the current position; The descent distance for all parameters is calculated sequentially using the following formula: （13） Step B5: Determine if the iteration termination condition has been met; Determine if the distance of gradient descent for all parameters is less than the termination threshold. u If all values ​​are less than 1, the algorithm terminates. This is the final result; otherwise, proceed to the next step. Step B6: Update parameters and begin the next iteration; Based on the basic iterative formula of the gradient descent algorithm, all parameter values ​​are updated using the following formula, and the process returns to step B3. （14）。

Images