A multispectral temperature measurement method based on optimization idea

Abstract

The present application relates to a kind of multispectral temperature measurement method based on optimization thought, belong to the field of spectral temperature measurement and non-contact temperature measurement.The present application establishes the relationship between temperature and spectral signal, establishes multivariate temperature difference correlation function, solves correlation function, establishes temperature measurement model, optimizes model solution, realizes temperature inversion.The present application simplifies the modeling process into the optimization problem of multivariate temperature difference correlation function, avoids the relationship assumption of spectral emissivity and other physical quantities, reduces the data sample amount requirement of deep learning method, simplifies the process of multispectral temperature measurement.The present application has improved inversion accuracy compared with secondary measurement method and neural network method;Inversion speed is greatly improved compared with SMM method.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of spectral temperature measurement and non-contact temperature measurement, and particularly relates to a multi-spectral temperature measurement method based on optimization. BACKGROUND

[0002] Multi-spectral temperature measurement is based on the blackbody radiation law, and the temperature value can be inferred through the radiation light intensity and multiple wavelengths, which overcomes the constraints of requiring single spectrum and similar colorimetric spectrum in colorimetric temperature measurement, and has been widely used in engineering practice.

[0003] In the process of multi-spectral temperature inversion, the solution of spectral emissivity and multi-spectral data processing are the keys to accurate temperature measurement. At present, the solution of spectral emissivity is mostly based on the spectral emissivity assumption model as the main method. When the assumption model is close to the actual situation, the inversion temperature and spectral emissivity are highly accurate. When the two are not consistent, the inversion result is quite different from the actual situation. For temperature measurement under the condition of complex materials and dynamic changes of material properties in the combustion process, the method of spectral emissivity assumption model is blind. In recent years, the method of deep learning based on neural network is applied to multi-spectral temperature measurement, which avoids the spectral emissivity assumption model and can establish the nonlinear statistical regular relationship between temperature and multi-spectrum, but it needs a large amount of data and super strong computing power support, and the modeling process is complex. Therefore, in the process of multi-spectral temperature measurement, it is necessary to avoid the assumption of the relationship between spectral emissivity and other physical quantities and simplify the multi-spectral temperature measurement model. SUMMARY

[0004] (I) Technical problem to be solved

[0005] The technical problem to be solved by the application is how to provide a multi-spectral temperature measurement method based on optimization to solve the problems of blindness of the method of spectral emissivity assumption model, complexity of the modeling process of the method of deep learning based on neural network, and the like.

[0006] (II) Technical scheme

[0007] In order to solve the above technical problems, the application provides a multi-spectral temperature measurement method based on optimization, which comprises the following steps:

[0008] S1, establishing the relationship between temperature and spectral signal: based on the principle of multi-spectral radiation temperature measurement, the relationship between the voltage data of different channels of the multi-wavelength temperature meter and the measured temperature of each channel is established;

[0009] S2, establishing a multi-element temperature difference correlation function: using the correlation between multi-spectral signals at different temperatures, the relationship between the measured temperatures of each channel in the process of multi-spectral temperature inversion is analyzed, and based on the principle of multi-spectral radiation temperature measurement and the information correlation between the data of each channel in the temperature inversion process, a multi-element temperature difference correlation function is established.

[0010] S3. Solve the correlation function and establish a temperature measurement model: Introduce multispectral temperature measurement error, analyze the relationship between the theoretical solution and the actual solution of the function, transform the multispectral temperature inversion problem into a multivariate function optimization problem, and establish a temperature measurement model;

[0011] S4. Optimize the model solution to achieve temperature inversion: Use inequality constraints to determine the solution range of the constraint function, combine with optimization algorithms to find the optimal solution of the model, substitute the optimal solution into the model for secondary optimization, and finally achieve the solution of spectral emissivity and the inversion of true temperature.

[0012] (III) Beneficial Effects

[0013] This invention proposes a multispectral temperature measurement method based on optimization principles. This method simplifies the modeling process into an optimization problem of a multivariate temperature difference correlation function, avoiding assumptions about the relationship between spectral emissivity and other physical quantities. It also reduces the data sample size requirements of deep learning methods and simplifies the multispectral temperature measurement process. Compared with the Secondary Measurement Method (SMM) and neural network methods, this invention offers improved inversion accuracy and significantly faster inversion speed compared to the SMM method. Attached Figure Description

[0014] Figure 1 This is an experimental structural diagram of the present invention. Detailed Implementation

[0015] To make the objectives, contents, and advantages of the present invention clearer, the specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples.

[0016] The technical problem to be solved by this invention is how to provide a multispectral temperature measurement method based on optimization ideas, so as to utilize the correlation between multispectral signals at different temperatures to establish a multivariate temperature difference correlation function, and through the optimization of the correlation function, establish a high-precision temperature measurement model, avoid the assumption of the relationship between spectral emissivity and other physical quantities, and simplify the multispectral temperature measurement process.

[0017] To address the aforementioned technical problems, this invention proposes a multispectral temperature measurement method based on optimization principles, comprising the following steps:

[0018] S1. Establish the relationship between temperature and spectral signal: Based on the principle of multispectral radiation thermometry, establish the relationship between the voltage data of different channels of the multi-wavelength thermometer and the measured temperature of each channel;

[0019] S2. Establish a multivariate temperature difference correlation function: Utilize the correlation between multispectral signals at different temperatures, analyze the relationship between the measured temperatures of each channel during the multispectral temperature inversion process, and establish a multivariate temperature difference correlation function based on the principle of multispectral radiation thermometry and the information correlation between the data of each channel during the temperature inversion process.

[0020] S3. Solve the correlation function and establish a temperature measurement model: Introduce multispectral temperature measurement error, analyze the relationship between the theoretical solution and the actual solution of the function, transform the multispectral temperature inversion problem into a multivariate function optimization problem, and establish a temperature measurement model;

[0021] S4. Optimize the model solution to achieve temperature inversion: Use inequality constraints to determine the solution range of the constraint function, combine with optimization algorithms to find the optimal solution of the model, substitute the optimal solution into the model for secondary optimization, and finally achieve the solution of spectral emissivity and the inversion of true temperature.

[0022] Further, step 1 specifically includes: According to the blackbody radiation law, the radiance of an object at absolute temperature T at wavelength λ is:

[0023]

[0024] In the formula, L(λ,T) is the radiance of the object (W·m). -2 ·μm -1 ·sr -1 ), where λ is the wavelength (μm), T is the absolute temperature (K); ε λ C1 represents the spectral emissivity of the object at temperature T; C1 = 3.7415 × 10⁻⁶. 8 W·μm 4 ·m -2 C2 = 1.43879 × 10 4 μm 4 ·K.

[0025] As can be seen from equation (1), an object at temperature T will exhibit different wavelengths λ. i Different energies L(λ,T) are emitted from the thermometer. Assume the output voltage signal of the thermometer is:

[0026]

[0027] In the formula, and λ represents the test constants for the transmittance of an optical system and the sensitivity of a photosensitive device, which are only related to wavelength and independent of temperature. i These are the wavelengths corresponding to different channels of a multi-wavelength thermometer. For the object at wavelength λ i , where , is the spectral emissivity at temperature T, C2 is the second radiation constant, and n is the number of channels of the multi-wavelength thermometer.

[0028] From equation (2), it can be seen that when a blackbody furnace produces a certain standard temperature T rb At that time, the λ corresponding to different channels of the multi-wavelength thermometer will be... i Different energies are emitted from the thermometer, and the output voltage signal of the thermometer at this time is:

[0029]

[0030] In equation (3), It is a calibration constant related to wavelength, optical system transmittance, photosensitive device sensitivity, and the first radiation constant. Since a blackbody furnace produces standard radiation, its emissivity is approximately 1. Equation (3) can be simplified to:

[0031]

[0032] Based on the mathematical model of multispectral thermometry based on reference temperature, dividing equation (2) and equation (4) yields:

[0033]

[0034] Equation (5) can be used to represent the measured temperature of the multi-wavelength thermometer i-channel.

[0035]

[0036] Furthermore, step 2 specifically includes: for the same point of the object to be measured at the same time, its temperature is theoretically unique, that is, the temperature measured by different channels is expressed by equation (6). They are equal, so the sum of the squares of the temperature differences measured in adjacent channels is equal. It is 0 at this time. It is also 0. Therefore, use Indicates that it is contained in unknowns in The multivariate temperature difference correlation function F is formed.

[0037] Furthermore, step 3 specifically includes: due to the existence of measurement error, the smaller the value F of the multivariate temperature difference correlation function, that is, the smaller the temperature difference measured by each channel, the lower the temperature. The closer to the true temperature, the higher the accuracy of the temperature measurement. When F reaches its minimum value of 0, The true temperature represented is unique. However, due to the existence of errors, it is difficult for F to reach its minimum value, while theoretically there are infinitely many minimum values. Therefore, the multispectral temperature inversion problem is transformed into a multivariate function extremum optimization problem.

[0038] Furthermore, the measurement error includes errors present during data acquisition, data calibration, and data processing.

[0039] Furthermore, the inequality constraint condition in step 4 specifically includes: According to the relevant theory of spectral thermometry, the spectral emissivity of the object to be measured is in the range of (0, 1). Although this constraint relationship is simple, it restricts… The range of values ​​of increases the speed of solving multivariate function extremum optimization problems and constitutes inequality constraints.

[0040] Furthermore, the optimization algorithm in step 4 specifically includes: the method for solving the optimization problem of multivariate functions under inequality constraints is called the Multi-element Extreme Value Optimization Measurement Method (MEVO), and its basic structure is as follows:

[0041]

[0042] In the formula, X represents the unknown. The variables A, B, C, and D represent parameter values ​​related to wavelength or temperature.

[0043] The above analysis shows that the established multivariate temperature difference correlation function and inequality constraints are consistent with the form of the multivariate extreme value optimization method in equation (10). Here, min F represents the optimization of the extreme value of the multivariate temperature difference correlation function. Let a represent the range of spectral emissivity, where a = 0 and b = 1 during the true temperature inversion process. Therefore, the multivariate extremum optimization method can be used to find the minimum value of the multivariate temperature difference correlation function. The code for the multivariate extremum optimization method can be implemented using algorithms such as particle swarm optimization, gradient descent, and neural networks. After solving for the extrema of the multivariate function, the spectral emissivity value can be calculated. This value is then substituted back into the multivariate temperature difference correlation function for secondary optimization, ultimately achieving the inversion of spectral emissivity and true temperature.

[0044] Example 1:

[0045] The technical problem to be solved by this invention is how to provide a multispectral temperature measurement method based on optimization ideas, so as to utilize the correlation between multispectral signals at different temperatures to establish a multivariate temperature difference correlation function, and through the optimization of the correlation function, establish a high-precision temperature measurement model, avoid the assumption of the relationship between spectral emissivity and other physical quantities, and simplify the multispectral temperature measurement process.

[0046] To address the aforementioned technical problems, this invention proposes a multispectral temperature measurement method based on optimization principles, comprising the following steps:

[0047] S1. Establish the relationship between temperature and spectral signal: Based on the principle of multispectral radiation thermometry, establish the relationship between the voltage data of different channels of the multi-wavelength thermometer and the measured temperature of each channel;

[0048] S2. Establish a multivariate temperature difference correlation function: Utilize the correlation between multispectral signals at different temperatures, analyze the relationship between the measured temperatures of each channel during the multispectral temperature inversion process, and establish a multivariate temperature difference correlation function based on the principle of multispectral radiation thermometry and the information correlation between the data of each channel during the temperature inversion process.

[0049] S3. Solve the correlation function and establish a temperature measurement model: Introduce multispectral temperature measurement error, analyze the relationship between the theoretical solution and the actual solution of the function, transform the multispectral temperature inversion problem into a multivariate function optimization problem, and establish a temperature measurement model;

[0050] S4. Optimize the model solution to achieve temperature inversion: Use inequality constraints to determine the solution range of the constraint function, combine with optimization algorithms to find the optimal solution of the model, substitute the optimal solution into the model for secondary optimization, and finally achieve the solution of spectral emissivity and the inversion of true temperature.

[0051] In step 1, establishing the relationship between temperature and spectral signal specifically includes: According to the blackbody radiation law, the radiance of an object at absolute temperature T at wavelength λ is:

[0052]

[0053] In the formula, L(λ,T) is the radiance of the object (W·m). -2 ·μm -1 ·sr -1 ), where λ is the wavelength (μm), T is the absolute temperature (K); ε λ C1 represents the spectral emissivity of the object at temperature T; C1 = 3.7415 × 10⁻⁶. 8 W·μm 4 ·m -2 C2 = 1.43879 × 10 4 μm 4 ·K.

[0054] As can be seen from equation (1), an object at temperature T will exhibit different wavelengths λ. i Different energies L(λ,T) are emitted from the thermometer. Assume the output voltage signal of the thermometer is:

[0055]

[0056] In the formula, and λ represents the test constants for the transmittance of an optical system and the sensitivity of a photosensitive device, which are only related to wavelength and independent of temperature.i These are the wavelengths corresponding to different channels of a multi-wavelength thermometer. For the object at wavelength λ i , where , is the spectral emissivity at temperature T, C2 is the second radiation constant, and n is the number of channels of the multi-wavelength thermometer.

[0057] From equation (2), it can be seen that when a blackbody furnace produces a certain standard temperature T rb At that time, it will be at different wavelengths λ i Different energies are emitted from the thermometer, and the output voltage signal of the thermometer at this time is:

[0058]

[0059] In equation (3), It is a calibration constant related to wavelength, optical system transmittance, photosensitive device sensitivity, and the first radiation constant. Since a blackbody furnace produces standard radiation, its emissivity is approximately 1. Equation (3) can be simplified to:

[0060]

[0061] Based on the mathematical model of multispectral thermometry based on reference temperature, dividing equation (2) and equation (4) yields:

[0062]

[0063] Equation (5) can be used to represent the measured temperature of the multi-wavelength thermometer i-channel.

[0064]

[0065] In step 2, establishing the multivariate temperature difference correlation function specifically includes: for the same point of the object to be measured at the same time, its temperature is theoretically unique, that is, the temperature measured by different channels is expressed by equation (6). They are equal, expressed as the sum of the squares of the temperature differences measured in adjacent channels:

[0066]

[0067] Since errors are unavoidable during the measurement process, equation (7) often fails to hold. Therefore, the unknown is... The resulting multivariate temperature difference correlation function is expressed by equation (8):

[0068]

[0069] In the formula, the unknowns Included in the expression represented by equation (7) middle.

[0070] In step 3, solving the correlation function and establishing the temperature measurement model specifically includes: because of the existence of measurement error, the smaller the value F of the multivariate temperature difference correlation function, that is, the smaller the temperature difference measured by each channel, the lower the temperature. The closer to the true temperature, the higher the accuracy of the temperature measurement. When F reaches its minimum value of 0, The true temperature represented is unique. However, due to the existence of errors, it is difficult for F to reach its minimum value, while theoretically there are infinitely many minimum values. Therefore, the multispectral temperature inversion problem is transformed into a multivariate function extremum optimization problem.

[0071] The measurement error includes errors that exist during data acquisition, data calibration, and data processing.

[0072] In step 4, the inequality constraint conditions specifically include: According to the relevant theory of spectral thermometry, the spectral emissivity of the object to be measured... The range is between (0, 1). Although this constraint is simple, it restricts... The range of values ​​of increases the speed of solving multivariate function extremum optimization problems and constitutes inequality constraints.

[0073] In step 4, the optimization algorithm specifically includes: the method for solving the optimization problem of multivariate functions under inequality constraints is called the Multi-element Extreme Value Optimization Measurement Method (MEVO), and its specific structure is as follows:

[0074]

[0075] In the formula, Indicates the unknown number The variables that constitute This represents the parameter value related to wavelength or temperature. min F represents the optimized extremum of the multivariate temperature difference correlation function.

[0076] The above analysis shows that the established multivariate temperature difference correlation function and inequality constraints are consistent with the form of the multivariate extreme value optimization method in equation (9). Therefore, the multivariate extreme value optimization method can be used to solve for the minimum value of the multivariate temperature difference correlation function. The code for the multivariate extreme value optimization method can be implemented using algorithms such as particle swarm optimization, gradient descent, and neural networks. After solving for the extreme value of the multivariate function, the spectral emissivity value can be calculated. This value can then be substituted back into the multivariate temperature difference correlation function for secondary optimization, ultimately achieving the inversion of spectral emissivity and true temperature.

[0077] In this embodiment, taking the application of transient combustion temperature radiation measurement of explosives as an example, a simple 8-channel multispectral temperature measurement device was built, the structure of which is as follows: Figure 1 As shown, the spectrometer decomposes the radiation generated by the combustion of explosives into spectral bands of different wavelengths and converts them into electrical signals. The multispectral data collected by the spectrometer is transmitted to a PC for temperature calculation. Experimental results show that within the temperature range of 1923.15K-2273.15K, the measurement error of this invention is basically within 0.5%, with an error range of 0.07%-0.93%; the inversion time is less than 2.5S, and the fastest can reach 0.47S. Compared with the Secondary Measurement Method (SMM) and neural network methods, this invention has improved inversion accuracy; the inversion speed is significantly improved compared with the SMM method.

[0078] This invention proposes a multispectral temperature measurement method based on optimization principles. It simplifies the modeling process into an optimization problem of a multivariate temperature difference correlation function, avoiding assumptions about the relationship between spectral emissivity and other physical quantities. This reduces the data sample size requirements of deep learning methods and simplifies the multispectral temperature measurement process. Compared with the Secondary Measurement Method (SMM) and neural network methods, this invention offers improved inversion accuracy and significantly faster inversion speed compared to the SMM method.

[0079] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A multispectral temperature measurement method based on optimization principles, characterized in that, The method includes the following steps: S1. Establish the relationship between temperature and spectral signal: Based on the principle of multispectral radiation thermometry, establish the relationship between the voltage data of different channels of the multi-wavelength thermometer and the measured temperature of each channel; S2. Establish a multivariate temperature difference correlation function: Utilize the correlation between multispectral signals at different temperatures, analyze the relationship between the measured temperatures of each channel during the multispectral temperature inversion process, and establish a multivariate temperature difference correlation function based on the principle of multispectral radiation thermometry and the information correlation between the data of each channel during the temperature inversion process. S3. Solve the correlation function and establish a temperature measurement model: Introduce multispectral temperature measurement error, analyze the relationship between the theoretical solution and the actual solution of the function, transform the multispectral temperature inversion problem into a multivariate function optimization problem, and establish a temperature measurement model; S4. Optimize the model solution to achieve temperature inversion: Use the inequality constraint function to solve the range, combine it with the optimization algorithm to find the optimal solution of the model, substitute the optimal solution into the model for secondary optimization, and finally realize the solution of spectral emissivity and the inversion of true temperature. in, Step S1 specifically includes: According to the blackbody radiation law, absolute temperature... The object at wavelength The radiance is as follows: （1） In the formula, The radiance of an object, measured in W·m. -2 ·μm -1 ·sr -1 , Wavelength, in μm. This is absolute temperature, in units of; For the object at temperature The spectral emissivity; C1 = 3.7415 × 10 8 W·μm 4 ·m -2 C2 = 1.43879 × 10 4 μm 4 ·K; From equation (1), it can be seen that the temperature Objects will appear at different wavelengths Different energies are radiated from below. Assume the output voltage signal of the thermometer is: （2） In the formula, and These represent the test constants for the transmittance of an optical system and the sensitivity of a photosensitive device, respectively, which are only related to wavelength and not to temperature. These are the wavelengths corresponding to different channels of a multi-wavelength thermometer. For the object at wavelength temperature Spectral emissivity at that time It refers to the number of channels in a multi-wavelength thermometer; From equation (2), it can be seen that when a blackbody furnace produces a certain standard temperature At that time, it will be on different channels of the multi-wavelength thermometer. Different energies are emitted from the thermometer, and the output voltage signal of the thermometer at this time is: （3） In equation (3), It is a calibration constant related to wavelength, optical system transmittance, photosensitive device sensitivity, and the first radiation constant. Since a blackbody furnace produces standard radiation, its emissivity is approximately 1. Equation (3) simplifies to: （4） Based on the mathematical model of multispectral thermometry based on reference temperature, dividing equation (2) and equation (4) yields: （5） The multi-wavelength thermometer is obtained from equation (5). Channel temperature measurement : （6） In step S2, the temperature at the same point on the object being measured at the same time is theoretically unique, that is, the temperature measured by different channels as expressed by equation (6). They are equal, so the sum of the squares of the temperature differences measured in adjacent channels is equal. It is 0, therefore use Indicates that it is contained in unknowns in Constructed multivariate temperature difference correlation function ; The temperature at the same point on the object being measured at the same time is theoretically unique, that is, the temperature measured by different channels is expressed by equation (6). They are equal, expressed as the sum of the squares of the temperature differences measured in adjacent channels: （7） Since errors are unavoidable during the measurement process, equation (7) often fails to hold. Therefore, the unknown is... The resulting multivariate temperature difference correlation function is expressed by equation (8): （8） In the formula, the unknowns Included in the expression of equation (7) middle.

2. The multispectral temperature measurement method based on optimization as described in claim 1, characterized in that, Step S3 specifically includes: multivariate temperature difference correlation function values. The smaller the value, the smaller the temperature difference measured by each channel, indicating a lower temperature. The closer the temperature is to the true temperature, the higher the accuracy of the temperature measurement. When the minimum value of 0 is reached, The true temperature represented is unique; due to the existence of error, It is difficult to find the minimum value, and theoretically there are infinitely many minimum values. Therefore, the multispectral temperature inversion problem is transformed into a multivariate function extremum optimization problem.

3. The multispectral temperature measurement method based on optimization as described in claim 2, characterized in that, The measurement error includes errors that exist during data acquisition, data calibration, and data processing.

4. The multispectral temperature measurement method based on optimization as described in claim 2, characterized in that, In step S4, the inequality constraint conditions specifically include: According to the relevant theory of spectral thermometry, the spectral emissivity of the object to be measured... The range is between (0, 1), and this constraint governs... The range of values ​​of increases the speed of solving multivariate function extremum optimization problems and constitutes inequality constraints.

5. The multispectral temperature measurement method based on optimization as described in claim 4, characterized in that, The optimization algorithm in step S4 specifically includes: the method for solving the optimization problem of multivariate functions under inequality constraints is called the Multi-element Extreme Value Optimization Measurement Method (MEVO), and its basic structure is as follows: （10） In the formula, Indicates the unknown number The variables that constitute This represents a parameter value related to wavelength or temperature; where This represents optimizing the extrema of the multivariate temperature difference correlation function. This represents the range of spectral emissivity during the true-temperature inversion process. ; After solving for the extrema of the multivariate function, the spectral emissivity value can be calculated. Substituting it back into the multivariate temperature difference correlation function for secondary optimization, the inversion of spectral emissivity and true temperature can be finally achieved.

6. The multispectral temperature measurement method based on optimization as described in claim 4, characterized in that, The optimization algorithm in step S4 specifically includes: the method for solving the optimization problem of multivariate functions under inequality constraints is called the Multi-element Extreme Value Optimization Measurement Method (MEVO), and its specific structure is as follows: （9） In the formula, Indicates the unknown number The variables that constitute , , , Indicates parameter values ​​related to wavelength or temperature; This represents optimizing the extrema of the multivariate temperature difference correlation function; The minimum value of the multivariate temperature difference correlation function is obtained by using the multivariate extreme value optimization method. After finding the extreme value of the multivariate function, the spectral emissivity value can be calculated. The value is then substituted back into the multivariate temperature difference correlation function for secondary optimization, and finally the inversion of spectral emissivity and true temperature is achieved.

7. The multispectral temperature measurement method based on optimization as described in claim 5 or 6, characterized in that, Multivariate extremum optimization methods are implemented using particle swarm optimization, gradient descent, or neural network algorithms.

Images