Aircraft engine turbine blade radiation temperature measuring device and temperature measuring method

Through neural network identification and training of the emissivity model of the turbine blades, combined with the Levenberg-Marquard algorithm, the problem of insufficient emissivity model accuracy in multispectral radiation temperature measurement technology is solved, and the accuracy and response speed of the turbine blade temperature measurement are achieved. promote.

CN115493708BActive Publication Date: 2024-12-03CHANGCHUN INST OF OPTICS FINE MECHANICS & PHYSICS CHINESE ACAD OF SCI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202211212718.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-09-30
Publication Date
2024-12-03
Estimated Expiration
2042-09-30

AI Technical Summary

Technical Problem

The existing multispectral radiation temperature measurement technology has insufficient universal accuracy of emissivity model in the temperature measurement of turbine blades, resulting in temperature measurement errors and cannot effectively solve the measurement challenges brought by the complex shape and high-speed rotation of turbine blades.

Method used

A neural network is used to identify and train the emissivity model of the thermal radiation spectrum of the turbine blades, establish a spectral emissivity model, and combine the Levenberg-Marquard algorithm to perform temperature inversion to achieve accurate inversion of the surface temperature field and heat load distribution of the turbine blades.

Benefits of technology

It effectively eliminates temperature measurement errors caused by emissivity changes, improves the accuracy and response speed of turbine blade temperature measurement, and is suitable for turbine blades of different models and surface forms.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115493708B_ABST
    Figure CN115493708B_ABST
Patent Text Reader

Abstract

The present invention provides an aeroengine turbine blade spectral radiation temperature measurement device and a temperature measurement method thereof, wherein the method comprises the following steps: S1, real-time identification of the emissivity model of the radiation spectrum of the turbine blade to be measured by an emissivity model recognition neural network; S2, establishing an evaluation deviation function Δ according to the emissivity model of the radiation spectrum of the turbine blade to be measured 2 ; S3, the temperature region of the turbine blade to be tested is divided into N regions, and the evaluation deviation function is solved by the Levenberg-Marquard algorithm to obtain N temperature solutions: T1, T2…T N ; S4, according to N temperature solutions: T1, T2...T N The evaluation deviation function value is calculated, and when the evaluation deviation function value is the minimum value, the corresponding temperature solution is used as the actual temperature value of the turbine blade to be measured. The present invention can effectively eliminate the error introduced by the emissivity change by judging the turbine blade emissivity model in real time and using the turbine blade spectral temperature inversion method based on the optimization algorithm.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of optical technology, and in particular to a radiation temperature measuring device for a turbine blade of an aero-engine and a temperature measuring method thereof. Background Art

[0002] As the most advanced equipment in modern manufacturing, aircraft engines are a reflection of a country's comprehensive national strength and high-end manufacturing level. As the "heart" of aircraft engines, the role of turbine blades is self-evident. When aircraft engines are running, turbine blades are in a state of high temperature, high pressure, and high stress for a long time, which places great demands on material strength, high temperature resistance, and fatigue resistance. When the temperature of an aircraft engine exceeds the maximum temperature that the turbine blades can withstand, the turbine blades will be damaged. Therefore, in order to eliminate long-term overload of turbine blades and improve the stability and reliability of aircraft engine operation, it is necessary to monitor the temperature of turbine blades in real time during operation.

[0003] At present, the temperature monitoring of turbine blades is mainly divided into two methods: contact temperature measurement and non-contact temperature measurement. Among them, the contact temperature measurement method mainly adopts the thermocouple temperature measurement method. The thermocouple is welded to the surface of the turbine blade, and the temperature is measured after the thermocouple and the turbine blade are in contact with each other to reach thermal equilibrium. The non-contact turbine blade temperature measurement method is currently mostly based on multi-spectral radiation temperature measurement. Multi-spectral radiation temperature measurement has the advantages of fast response speed, high upper limit of temperature measurement, and wide dynamic range. At present, the temperature measurement of aircraft engine turbine blades mostly adopts multi-spectral radiation temperature measurement. The selection of turbine blade emissivity model directly affects the accuracy of temperature measurement. Turbine blades with different materials and different surface conditions often need to select different types of emissivity models in the radiation temperature measurement process. At present, there is a lack of universal and accurate estimation methods for the emissivity model in the multi-spectral data processing algorithm of turbine blades, which limits the widespread application of turbine blade spectral temperature measurement technology.

[0004] In terms of the current development status at home and abroad, although the relevant technologies and equipment of infrared radiation temperature measurement have made great progress, most measurement systems still have technical problems such as the complex shape of turbine blades, high-speed rotation, and difficulty in positioning, and cannot be directly applied to the temperature measurement of aircraft engine turbine blades. There are no mature products in China yet. The above research has made great progress in improving the temperature inversion speed and measurement signal-to-noise ratio. However, due to the lack of a universal and accurate estimation method for the turbine blade emissivity model in the multi-spectral data processing algorithm, the widespread application of multi-spectral high-temperature measurement technology is limited. The development and research of relevant technologies for infrared radiation temperature measurement of engine turbine blades has huge economic value and social benefits in industrial production and application. The application demand for infrared radiation temperature measurement systems for aircraft engine turbine blades is also very urgent, so the multi-spectral radiation temperature measurement method still needs to be continuously improved.

[0005] There are two main algorithms for multispectral data processing that are commonly used at present: one is to assume the emissivity model of the surface of the object being measured, and bring in Planck's formula to solve the temperature, eliminating the problem of unsolvable temperature due to emissivity. However, the emissivity is usually related to the surface morphology, temperature, wavelength and measurement angle of the object being measured. The calculation results obtained by the emissivity model are consistent with the actual data, resulting in large errors in multispectral temperature measurement. The second is to convert the multi-channel temperature and emissivity solution into the problem of calculating the minimum value of the optimization function through optimization algorithms such as simulated annealing algorithms and genetic algorithms. Similarly, when there is a deviation between the emissivity model and the actual emissivity of the object being measured, the solution to the temperature of each channel will bring large errors. Summary of the invention

[0006] In view of the above problems, the purpose of the present invention is to propose an aero-engine turbine blade radiation temperature measurement device and a temperature measurement method thereof, which solves the temperature measurement error caused by the change of emissivity in the radiation temperature measurement process of turbine blades of different models at present, and measures the thermal radiation spectrum of the turbine blade. Based on the principle that the thermal radiation spectrum of different types of turbine blades is different due to different components and surface morphologies, the thermal radiation spectrum of the turbine blade is trained by neural network to identify the emissivity model, and the thermal radiation spectrum of the turbine blade and the emissivity model of the turbine blade surface are established. The distribution characteristics of the spectral emissivity of the turbine blade surface under different materials and different surface morphologies are analyzed, and a turbine blade spectral emissivity feature extraction method based on an adaptive neural network is used to construct a turbine blade spectral emissivity model training sample library, establish a turbine blade spectral radiation information and spectral emissivity theoretical model, and construct a feature transfer relationship between the original turbine blade radiation spectrum---spectral emissivity model---spectral emissivity feature recognition.

[0007] Combined with the theoretical model of turbine blade spectral emissivity, based on the measurement of turbine blade spectral radiation distribution, the characteristic model of turbine blade spectral emissivity in different scenes is analyzed to solve the problem of turbine blade spectral emissivity feature identification in complex scenes. At the same time, a turbine blade spectral radiation transmission model is established, and a multi-spectral radiation temperature measurement algorithm is designed and implemented to realize the inversion of turbine blade surface temperature field and thermal load distribution.

[0008] To achieve the above object, the present invention adopts the following specific technical solutions:

[0009] The present invention provides a spectral radiation temperature measurement device for an aircraft engine turbine blade, comprising: a swinging reflector, a first collimating mirror, a second collimating mirror, a black body, a focusing mirror group, a spectral measurement system, a control system and a data processing system;

[0010] The radiation beam emitted by the turbine blade to be measured is incident on the swinging reflector, reflected by the swinging reflector, and then swept by cold air to be incident on the focusing lens group. The radiation beam is refracted by the focusing lens group and then incident on the semi-transparent and semi-reflective lens. After being reflected by the semi-transparent and semi-reflective lens, it is converged to the black body through the first collimator. The measuring beam reflected by the black body returns to the semi-transparent and semi-reflective lens to interfere with the radiation beam to form an interference beam.

[0011] The interference light beam is emitted horizontally and then refracted by the second collimator to enter the spectrum measurement system, thereby obtaining the radiation light beam data of the turbine blade to be measured;

[0012] The data processing system is used to process the radiation beam data to obtain the actual temperature of the turbine blade to be measured.

[0013] Preferably, the spectrum measurement system comprises: a slit, a first reflector, a grating, a second reflector, a prism and a CCD;

[0014] The interference light beam passes through the slit and is incident on the first reflector. After being reflected by the first reflector, it is incident on the grating and then emits a diffraction light beam which is incident on the prism. The diffraction light beam is reflected by the prism and the second reflector in turn and is received by the CCD to obtain the spectral radiation data of the turbine blade to be measured. The CCD transmits the spectral radiation data to the data processing system.

[0015] The present invention also provides a method for measuring the temperature of an aircraft engine turbine blade using spectral radiation, comprising the following steps:

[0016] S1, identifying the emissivity model of the radiation spectrum of the turbine blade to be tested in real time through an emissivity model recognition neural network;

[0017] S2. Establish an evaluation deviation function Δ according to the emissivity model of the radiation spectrum of the turbine blade to be tested 2 :

[0018]

[0019] S3, the temperature region of the turbine blade to be tested is divided into N regions, and the evaluation deviation function is solved by the Levenberg-Marquard algorithm to obtain N temperature solutions: T1, T2…T N ;

[0020] S4. Solve according to N temperatures: T1, T2…T N The evaluation deviation function value is calculated, and when the evaluation deviation function value is the minimum value, the corresponding temperature solution is used as the actual temperature value of the turbine blade to be measured.

[0021] Preferably, step S1 includes the following sub-steps:

[0022] S11, obtaining a radiation spectrum of a turbine blade to be measured by using an aircraft engine turbine blade spectral radiation temperature measuring device;

[0023] S12, pre-processing the radiation spectrum of the turbine blade to be measured and inputting it into the emissivity model recognition neural network;

[0024] S13. Determine in real time the emissivity model of the radiation spectrum of the turbine blade to be tested according to the sorting result of the turbine blade emissivity model.

[0025] Preferably, the radiation spectrum of the turbine blade to be measured is preprocessed as follows:

[0026] Assume that the spectral radiation temperature measurement device for aircraft engine turbine blades has a total of n effective measurement channels. For the i-th spectral measurement channel, the spectral radiation energy of the turbine blade to be measured is expressed by the Wien formula as follows:

[0027]

[0028] in,

[0029] c1=3.7418×10 -16 W·m 2 is the first radiation constant;

[0030] c2=1.4388×10 -2 m·K is the second radiation constant;

[0031] T is the temperature of the turbine blade to be measured;

[0032] λ i is the measurement wavelength corresponding to the i-th spectrum measurement channel;

[0033] ε(λi,T) is the emissivity of the turbine blade to be measured;

[0034] L(λi,T) is the spectral radiation energy of the turbine blade received by the spectral radiation temperature measurement device;

[0035] Perform spectral baseline correction on the spectral radiation energy of the turbine blade to be measured:

[0036] L s (λ i ,T)==L(λ i ,T)-min[L(λ i ,T)]

[0037] The normalized spectral radiation energy of the turbine blade to be tested is obtained as follows:

[0038]

[0039] Preferably, the training process of the emissivity model recognition neural network is:

[0040] S101, establishing a turbine blade spectral emissivity model according to spectral radiation characteristics of different turbine blades;

[0041] S102, constructing training sample data and test sample data according to the emissivity model;

[0042] S103, establishing an initial emissivity model recognition neural network, and performing sample recognition training based on the training sample data and the test sample data;

[0043] S104, judging the error convergence of the initial emissivity model recognition neural network by the gradient descent method, and improving the initial emissivity model recognition neural network according to the error size distribution;

[0044] S105. When the error convergence meets the requirements, the final emissivity model identification neural network is obtained.

[0045] Compared with the existing technology, the present invention can solve the problem of temperature measurement error caused by emissivity change during the measurement of radiation temperature of turbine blades of different types and different surface forms. By using a neural network to train the infrared spectrum characteristics of different types of turbine blades, the turbine blade emissivity model is judged in real time, and the turbine blade spectrum temperature inversion method based on the optimization algorithm is used to eliminate the error introduced by the emissivity change. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a structural schematic diagram of a spectral radiation temperature measurement device for an aircraft engine turbine blade provided according to an embodiment of the present invention.

[0047] Figure 2 It is a flow chart of a method for spectral radiation temperature measurement of aircraft engine turbine blades provided according to an embodiment of the present invention.

[0048] Figure 3 It is a schematic diagram of the solution process of the turbine blade temperature by neural network training provided according to an embodiment of the present invention.

[0049] Figure 4 It is a schematic diagram of the process of training the emission rate of turbine blades by a neural network according to an embodiment of the present invention.

[0050] The reference numerals therein include: high-temperature combustion gas 1, sweeping mirror 2, turbine blades 3, cold air purge 4, window 5, black body 6, focusing lens group 8, spectrum measurement system 9, slit 10, first reflector 11, grating 12, second reflector 13, prism 14, CCD 15, servo control 16, main controller 17 and data processing system 18. DETAILED DESCRIPTION

[0051] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the following description, the same modules are represented by the same reference numerals. In the case of the same reference numerals, their names and functions are also the same. Therefore, the detailed description thereof will not be repeated.

[0052] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and do not constitute a limitation of the present invention.

[0053] Figure 1 A schematic structural diagram of a spectral radiation temperature measurement device for an aircraft engine turbine blade provided according to an embodiment of the present invention is shown.

[0054] like Figure 1 As shown, the spectral radiation temperature measurement device for aircraft engine turbine blades provided in an embodiment of the present invention includes: a swinging reflector 2, a black body 6, a focusing lens group 8, a spectral measurement system 9, a control system and a data processing system 18.

[0055] The radiation light beam emitted by the turbine blade 3 is incident on the swinging reflector 2, and after being reflected by the swinging reflector and swept by the cold air 4, it is incident on the focusing lens group 8 through the window 5. After passing through the focusing lens group 8, the radiation light beam is reflected by the semi-transparent and semi-reflective lens, and then is incident on the black body 6 through the first collimating lens 7-1. The measuring light beam reflected by the black body 6 returns to the semi-transparent and semi-reflective lens to interfere with the above-mentioned radiation light beam to form an interference light beam.

[0056] The interference light beam is emitted horizontally and then enters the spectrum measurement system 9 through the second collimator 7 - 2 .

[0057] The spectrum measurement system 9 includes a slit 10 , a first reflecting mirror 11 , a grating 12 , a second reflecting mirror 13 , a prism 14 and a CCD 15 .

[0058] The interference beam passes through the slit 10 and enters the first reflector 11. After being reflected by the first reflector 11, it enters the grating 12 and emits a diffracted beam that enters the prism 14. The diffracted beam is reflected by the prism 14 and the second reflector 13 in turn and is received by the CCD 15 to obtain the radiation data of the turbine blade. The CCD 15 converts the optical signal of the diffracted beam into an electrical signal and transmits it to the data processing system 18. After processing the radiation data of the turbine blade, the temperature of the turbine blade of the aircraft engine is obtained.

[0059] The control system includes: a servo control 16 and a main controller 17 .

[0060] The servo control 16 is mainly used to control the movement of the focusing lens group 8 and the swing of the scanning mirror 2, wherein the servo system controls the movement of the focusing lens group to achieve the purpose of focusing, and controls the movement of the scanning mirror 2 to achieve measurement of different positions of the turbine blades.

[0061] The main controller 17 is mainly used for parameter control of the spectrometer, black box and other equipment in the measuring equipment.

[0062] Figure 2 A schematic flow chart of a method for spectral radiation temperature measurement of an aircraft engine turbine blade provided in accordance with an embodiment of the present invention is shown.

[0063] Figure 3 A schematic diagram of a process for solving the temperature of turbine blades by training a neural network according to an embodiment of the present invention is shown.

[0064] like Figure 2 and Figure 3 As shown, the spectral radiation temperature measurement method for aircraft engine turbine blades provided by the embodiment of the present invention includes the following steps:

[0065] S1, identifying the emissivity model of the radiation spectrum of the turbine blade to be tested in real time through an emissivity model recognition neural network;

[0066] Since the spectral emissivity models of turbine blades of different types and temperatures have different forms during the temperature measurement of turbine blades, when the emissivity model of the turbine blades changes due to factors such as surface morphology and temperature changes during the temperature measurement process, traditional spectral radiation temperature measurement will produce errors.

[0067] Therefore, the spectral emissivity model of the turbine blade can be identified and determined in real time through the measured spectral radiation intensity distribution curve of the turbine blade, and then the temperature of the target turbine blade can be solved to eliminate the temperature measurement error caused by the change of spectral emissivity.

[0068] Thermal radiation temperature measurement generally assumes the turbine blades to be black bodies (ε=1) or gray bodies (ε<1), but the thermal emissivity ε of actual turbine blades varies. When measuring turbine blades, the emissivity of the turbine blade surface is related to the surface state of the turbine blades, the temperature, whether there is a thermal barrier coating on the turbine blade surface, and the wavelength of radiation temperature measurement.

[0069] In order to solve the above problems, a neural network is used to identify the spectral emissivity model of turbine blades. Due to the differences between the spectral curves obtained by different emissivity models, the classification function of the neural network can be used to identify and classify these spectral curves, so that the original emissivity model can be accurately identified.

[0070] Figure 4A schematic diagram of a process of training the emission rate of turbine blades using a neural network according to an embodiment of the present invention is shown.

[0071] like Figure 4 As shown, step S1 includes the following sub-steps:

[0072] S11, obtaining a radiation spectrum of a turbine blade to be measured by using an aircraft engine turbine blade spectral radiation temperature measuring device;

[0073] S12, pre-processing the radiation spectrum of the turbine blade to be measured and inputting it into the emissivity model recognition neural network;

[0074] The preprocessing process of the radiation spectrum of the turbine blade to be tested is as follows:

[0075] Assume that the spectral radiation temperature measurement device for aircraft engine turbine blades has a total of n effective measurement channels. For the i-th spectral measurement channel, the spectral radiation energy of the turbine blade to be measured can be approximately expressed using the Wien formula as follows:

[0076]

[0077] in,

[0078] c1=3.7418×10 -16 W·m 2 is the first radiation constant;

[0079] c2=1.4388×10 -2 m·K is the second radiation constant;

[0080] T is the temperature of the turbine blade to be measured;

[0081] λ i is the measurement wavelength corresponding to the i-th spectrum measurement channel;

[0082] ε(λ i ,T) is the emissivity of the measured turbine blades;

[0083] L(λ i ,T) is the spectral radiation energy of the turbine blade received by the spectral radiation temperature measurement system.

[0084] In order to solve the spectral shift and drift caused by the unstable environment inside the aircraft engine during the actual turbine blade spectrum measurement process, it is necessary to perform spectral baseline correction on the spectral radiation energy of the turbine blade to be measured:

[0085] L s (λ i ,T)==L(λ i ,T)-min[L(λ i ,T)]

[0086] In order to eliminate the problem that weak spectral signals are easily ignored during the training process, the measured spectrum needs to be normalized to:

[0087]

[0088] S13. Determine in real time the emissivity model of the radiation spectrum of the turbine blade to be tested according to the sorting result of the turbine blade emissivity model.

[0089] The commonly used spectral emissivity models of aircraft engine turbine blades are mainly the following:

[0090] (1) Linear emissivity model: ε(λ i ,T)=a0+a1λ i

[0091] (2) Polynomial emissivity model:

[0092] (3) Trigonometric emissivity model: ε(λ i ,T)=sin(a0+a1λ i )

[0093] (4) Exponential emissivity model: ε(λ i ,T)=exp(a0+a1λ i )

[0094] The training process of the above firing rate model recognition neural network is:

[0095] S101, establishing a turbine blade spectral emissivity model according to spectral radiation characteristics of different turbine blades;

[0096] S102, constructing training sample data and test sample data according to the emissivity model;

[0097] S103, establishing an initial emissivity model recognition neural network, and performing sample recognition training;

[0098] S104, judging the error convergence of the initial emissivity model recognition neural network by the gradient descent method, and improving the initial emissivity model recognition neural network according to the error size distribution;

[0099] The spectral training parameters, spectral preprocessing methods and spectral feature selection methods are adjusted according to the error size distribution.

[0100] Spectral training parameters include: number of hidden layers and layers, activation function category, output layer weights and other parameters.

[0101] S105. When the error convergence meets the requirements, the final emissivity model identification neural network is obtained.

[0102] Meeting the requirement means that the recognition rate using the test set sample emission rate is approximately 1.

[0103] In actual measurement, the radiation spectrum distribution of the aircraft engine turbine blades in each spectral channel is measured, and the radiation spectrum of the turbine blades is classified into emissivity models using the trained emissivity model recognition neural network, so as to obtain the spectral emissivity model corresponding to the turbine blades at this time.

[0104] When the turbine blade spectral emissivity model is introduced, the number of spectral emissivity coefficients and the unknown quantities in the multi-spectral temperature measurement model equation group are determined. It is difficult to directly solve the equation group by converting it into an implicit function problem of turbine blade temperature and spectral emissivity model coefficients. Therefore, this project uses the optimal deviation function algorithm to solve the true temperature of the turbine blade.

[0105] S2. Establish the evaluation deviation function Δ according to the selected emissivity model 2 :

[0106]

[0107] S3, the temperature region of the turbine blade to be tested is divided into N regions, and the evaluation deviation function is solved by the Levenberg-Marquard algorithm to obtain N temperature solutions: T1, T2…T N ;

[0108] S4. Solve according to N temperatures: T1, T2…T N The evaluation deviation function value is calculated, and when the evaluation deviation function value is the minimum value, the corresponding temperature solution is used as the actual temperature value of the turbine blade to be measured.

[0109] Therefore, when the spectral emissivity model of the turbine blade to be tested is determined, the deviation function Δ 2 When the minimum value of the deviation function approaches zero, the corresponding spectral emissivity of the turbine blade to be measured and the temperature of the turbine blade to be measured can be obtained.

[0110] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

[0111] The above specific implementations of the present invention do not constitute a limitation on the protection scope of the present invention. Any other corresponding changes and modifications made based on the technical concept of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A spectral radiation temperature measurement device for aircraft engine turbine blades, characterized in that: include: Swinging reflector, first collimator, second collimator, black body, focusing lens group, spectrum measurement system, control system and data processing system; The radiation beam emitted by the turbine blade to be measured is incident on the swinging reflector, and after being reflected by the swinging reflector, it is swept by cold air and incident on the focusing lens group. After being refracted by the focusing lens group, the radiation beam is incident on the semi-transparent and semi-reflective lens, and after being reflected by the semi-transparent and semi-reflective lens, it is converged to the black body through the first collimator. The measuring beam reflected by the black body returns to the semi-transparent and semi-reflective lens to interfere with the radiation beam to form an interference beam. The interference light beam is emitted horizontally and then refracted by the second collimator to enter the spectrum measurement system, thereby obtaining radiation light beam data of the turbine blade to be measured; The data processing system is used to process the radiation beam data to obtain the actual temperature of the turbine blade to be measured.

2. The spectral radiation temperature measuring device for aircraft engine turbine blades according to claim 1, characterized in that: The spectrum measurement system comprises: a slit, a first reflector, a grating, a second reflector, a prism and a CCD; The interference light beam is incident on the first reflector through the slit, is reflected by the first reflector, is incident on the grating, and then emits a diffraction light beam which is incident on the prism. The diffraction light beam is reflected by the prism and the second reflector in turn and is received by the CCD to obtain the spectral radiation data of the turbine blade to be measured. The CCD transmits the spectral radiation data to a data processing system.

3. A temperature measurement method for an aircraft engine turbine blade spectral radiation temperature measurement device according to any one of claims 1 or 2, characterized in that: The following steps are involved: S1, identifying the emissivity model of the radiation spectrum of the turbine blade to be measured in real time through an emissivity model recognition neural network; S2, establishing an evaluation deviation function Δ according to the emissivity model of the radiation spectrum of the turbine blade to be measured 2 : S3, dividing the temperature region of the turbine blade to be tested into N regions, solving the evaluation deviation function by Levenberg-Marquard algorithm, and obtaining N temperature solutions: T1, T2, ..., T N ; S4. Solve according to the N temperature: T1, T2...T N The evaluation deviation function value is calculated, and when the evaluation deviation function value is the minimum value, the corresponding temperature solution is used as the actual temperature value of the turbine blade to be measured.

4. The method for measuring temperature of a turbine blade of an aircraft engine by spectral radiation according to claim 3, characterized in that: The step S1 comprises the following sub-steps: S11, obtaining the radiation spectrum of the turbine blade to be measured by the aircraft engine turbine blade spectral radiation temperature measurement device; S12, preprocessing the radiation spectrum of the turbine blade to be tested and inputting it into the emissivity model recognition neural network; S13. Determine in real time the emissivity model of the radiation spectrum of the turbine blade to be tested according to the sorting result of the turbine blade emissivity model.

5. The method for measuring temperature of a turbine blade of an aircraft engine by spectral radiation according to claim 4, characterized in that: The radiation spectrum of the turbine blade to be tested is preprocessed as follows: Assume that the aircraft engine turbine blade spectral radiation temperature measurement device has a total of n effective measurement channels. For the i-th spectral measurement channel, the spectral radiation energy of the turbine blade to be measured is expressed by the Wien formula as follows: in, c1=3.7418×10 -16 W·m 2 is the first radiation constant; c2=1.4388×10 -2 m·K is the second radiation constant; T is the temperature of the turbine blade to be measured; λ i is the measurement wavelength corresponding to the i-th spectrum measurement channel; ε(λ i , T) is the emissivity of the turbine blade to be tested; L(λ i , T) is the spectral radiation energy of the turbine blade received by the spectral radiation temperature measuring device; Perform spectral baseline correction on the spectral radiation energy of the turbine blade to be measured: L s (l i ,T)==L(λ i ,T)-min[L(λ i ,T)] The normalized spectral radiation energy of the turbine blade to be tested is obtained as follows:

6. The method for measuring temperature of a turbine blade of an aircraft engine by spectral radiation according to claim 5, characterized in that: The training process of the firing rate model recognition neural network is: S101, establishing a turbine blade spectral emissivity model according to spectral radiation characteristics of different turbine blades; S102, constructing training sample data and test sample data according to the emissivity model; S103, establishing an initial emissivity model recognition neural network, and performing sample recognition training according to the training sample data and the test sample data; S104, judging the error convergence of the initial emissivity model recognition neural network by a gradient descent method, and improving the initial emissivity model recognition neural network according to the error size distribution; S105. When the error convergence meets the requirements, the final emissivity model identification neural network is obtained.