Gas turbine blade water flow measuring method

Abstract

The invention discloses a gas turbine blade water flow measurement method, and relates to the technical field of gas turbine blades, and the method comprises the steps: carrying out finite element modeling simulation to obtain the theoretical flow of a target blade inner cavity; partitioning processing is performed based on the theoretical flow to obtain a first-stage flow velocity region and a second-stage flow velocity region, and a sensor group is arranged to acquire test flow data; carrying out phase alignment on the ultrasonic test data, carrying out noise reduction processing through a one-dimensional convolutional neural network, and calculating the actual water flow velocity of a second flow velocity region based on the sound velocity in the flow channel; fFT processing is carried out on radar test data, and the actual water flow velocity of each primary flow velocity area is obtained through turbulence intensity adaptive filtering and corrected radar Doppler frequency shift calculation; establishing federated Kalman filtering to carry out global fusion processing to obtain a global flow velocity; the comprehensive water flow of the target blade is obtained through calculation based on a section average flow velocity method, the corrected comprehensive water flow is obtained through calculation based on the real-time water flow friction factor, and a water flow feature set is output.

Description

Technical Field

[0001] The present invention relates to the technical field of gas turbine blades, and particularly to a method for measuring the water flow rate of a gas turbine blade. Background Art

[0002] Heavy-duty gas turbine blades are long-term exposed to extreme high temperature and high pressure environments. Their interiors are usually designed with complex cooling channels, such as serpentine channels, film holes, etc., to achieve forced cooling through the flow of high-pressure air or steam. Insufficient flow rate or uneven distribution will lead to cooling failure in local areas, causing material creep, oxidation, and even ablation, directly affecting the blade life. The internal cavity flow rate detection is used to evaluate the flow capacity of the cooling gas in the blade internal cavity, which directly affects the cooling effect of the blade. In recent years, with the continuous development of science and technology, the temperature before the turbine is also constantly increasing. This not only depends on the improvement of the process level but also requires the improvement of the structural design. Therefore, the cooling design and cooling effect test of the turbine blade are constantly being improved. The research results of domestic water flow rate detection equipment are mostly used for the detection of the internal cavity flow rate of aero-engine turbine blades, and there is little research on the detection of the internal cavity water flow rate of ground gas turbine turbine blades. At the same time, the existing domestic research schemes for water flow rate equipment still have certain limitations. For example, there are human errors in the determination of the equipment time, the flow rate measurement is easily affected by pressure fluctuations, the time interval between the completion of one blade test and the start of the next blade test is relatively long, and the efficiency is low, etc.

[0003] Currently, the Chinese invention patent application with the application number CN110823323A discloses a method for correcting the measured water flow rate of a turbine rotor blade. The application includes the following steps: defining the pressure difference relationship between the inlet and outlet of the blade during actual water flow rate measurement; uniformly selecting multiple different pressure differences between the inlet and outlet of the blade and calculating the corresponding water flow rate values; performing formula fitting through a function to obtain a theoretical flow equation; measuring the water flow rate values of the blade at multiple different pressure differences between the inlet and outlet through a measuring device; performing formula fitting through a function to obtain an actual flow equation; obtaining a flow correction coefficient; and correcting the water flow rate value of the blade at a specified inlet pressure according to the flow correction coefficient. The method for correcting the measured water flow rate of the turbine rotor blade in this application takes into account the differences in water flow rate measurement in different production lines, reasonably corrects the actual measured results of the water flow rate, and achieves the purpose of directly using the water flow rate measurement results to determine the casting compliance of the internal cavity channel of the turbine rotor blade. However, the application object of this application is the aero-engine turbine blade, and there are differences between the ground gas turbine turbine blade and the aero-engine turbine blade. Summary of the Invention

[0004] The technical problem solved by the present invention is that the existing technology for measuring water flow rate on blades is mostly used for aviation turbine engines, and the heat dissipation flow channels of ground heavy-duty gas turbine blades are relatively large, and the existing technology is insufficiently applicable to measuring water flow rate in the heat dissipation cavity of ground heavy-duty gas turbine blades.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: Step S1, performing finite element modeling simulation to obtain the target blade cavity theoretical flow rate; Step S2, performing zoning processing based on the theoretical flow to obtain a primary flow rate area and a secondary flow rate area, and arranging a sensor group to collect test flow data; Step S3, performing phase alignment on the ultrasonic test data, performing noise reduction processing through a one-dimensional convolutional neural network, and calculating the actual water flow rate corresponding to the second flow rate area based on the sound velocity in the flow channel; Step S4, performing FFT processing on the radar test data, filtering processing using a turbulence intensity adaptive filtering algorithm, and calculating the actual water flow velocity corresponding to each primary flow velocity area based on the modified radar Doppler frequency shift; Step S5, establishing a federal Kalman filter to perform local state estimation, and performing global fusion processing based on radar weight factors and ultrasonic full factors to obtain a global flow velocity; Step S6, calculate the comprehensive water flow of the target blade based on the cross-sectional average flow velocity method, calculate the corrected comprehensive water flow based on the real-time water flow friction factor, and output the water flow feature set {corrected comprehensive water flow, average flow, peak flow, flow fluctuation rate}.

[0006] Preferably, the processing logic for finite element modeling and simulation includes: Importing the design drawing data of the target turbine blade into the CAD software, and performing preliminary modeling of the key areas of the flow channel to obtain an initial geometric model, wherein the key areas of the flow channel include guide grooves, cooling holes and fins; The initial geometric model is processed by unstructured tetrahedron and hexahedron hybrid meshing through the CFD platform, and local encryption is performed on the narrow part of the flow channel to control the Y+ value near the wall to be less than the preset mesh threshold. Multi-layer boundary layer meshes are set along the flow direction to obtain the wall shear stress and velocity gradient change data; Define boundary conditions, set the inlet flow rate, inlet static pressure and total temperature, set the turbulence intensity based on the Reynolds number, set the outlet static pressure and flow rate boundaries, define the temperature, thermal conductivity and convection heat transfer coefficient on the blade surface, initialize the full-field velocity and set the gravity acceleration to be turned on; The convective terms are discretized by the QUICK algorithm, the pressure-velocity coupling is performed by the PISO algorithm, and the RANS model is calculated based on the residual convergence threshold to obtain the flow velocity contour map, the average flow velocity of the isosurface, the average flow velocity of the cross-section, analyze the jet diffusion of the cooling holes, the vortex distribution between the fins and the low-velocity stagnant zone, and generate the flow velocity distribution map.

[0007] Preferably, a partitioning operation is performed on the target blade based on the flow velocity distribution map, and the processing logic of the partitioning operation includes: The geometric shape of the target blade flow channel is segmented to obtain geometric sub-regions, and the geometric shapes include an inlet buffer zone, a serpentine main flow channel, a spoiler rib gap zone, a film hole distribution zone, a high-curvature bend zone, a cooling hole outlet diffusion zone, an outlet convergence zone, a fin tip zone, and a multi-porous microstructure zone; The average flow velocity of each geometric sub-region is calculated based on the flow velocity distribution map to obtain a simulation flow velocity data set. The simulation flow velocity data set is traversed. When the simulation flow velocity is greater than or equal to the preset flow velocity threshold, the corresponding geometric sub-region is divided into a secondary flow velocity region; When the simulation flow velocity is greater than or equal to the preset flow velocity threshold, the corresponding geometric sub-region is divided into a secondary flow velocity region.

[0008] Preferably, multi-dimensional sensors are set on the target blade based on the position coordinate data of the finite element model, and the processing logic includes: High-frequency radar sensors and laser Doppler sensors are arranged in the primary flow velocity region, ultrasonic sensors are arranged in the secondary flow velocity region, and piezoresistive sensors and temperature sensors are arranged at the inlet and outlet of the target blade flow channel; A hydraulic test is performed on the target blade flow channel. The target blade flow channel is cleaned and all interfaces are sealed. Water is injected into the inlet of the target blade flow channel based on the preset first water pressure for exhaust until the flow channel is filled with water and there is no bubble residue; Pressurized water injection is carried out into the inlet of the target blade flow channel based on the cooling medium pressure set for the target blade, and a test data set is obtained through the sensors. The test data set includes radar test data, turbulent intensity test data, ultrasonic test data, flow channel temperature data, and pressure data; Preferably, the actual water flow velocity of each primary flow velocity region is calculated based on the ultrasonic test data, and the processing logic includes: The ultrasonic test data is subjected to 256-cycle phase alignment accumulation processing to obtain ultrasonic alignment data. The ultrasonic alignment data is denoised by a one-dimensional convolutional neural network. The network structure of the one-dimensional convolutional neural network includes 3 convolutional layers and 1 fully connected layer. The convolutional kernel size of the convolutional layer is set to 5*1, and the step size is set to 5. The fully connected layer is activated by the ReLU function. The input is the data corresponding to 1024-point ultrasonic test data, and the output is ultrasonic denoised data; Preferably, the speed of sound in the real-time flow channel is calculated based on the temperature data, and the actual water flow velocity corresponding to the second flow velocity region is calculated by the bidirectional time difference weighted average method based on the speed of sound in the flow channel and the ultrasonic denoising data. The calculation expression is as follows: ; ; wherein, represents the speed of sound in the flow channel corresponding to the secondary flow velocity region with serial number i, represents the water temperature, represents the actual water flow velocity corresponding to the secondary flow velocity region with serial number i, D represents the distance between ultrasonic sensors in the flow channel, represents the cosine value of the ultrasonic sound path inclination angle, represents the downstream reception time, represents the upstream reception time.

[0009] Preferably, the actual water flow velocity of each primary flow velocity region is calculated based on the radar test data. The processing logic includes: performing fast Fourier transform processing on the radar test data to obtain radar spectrum data; filtering the radar spectrum data through a turbulence intensity adaptive filtering algorithm to remove high-frequency noise and turbulence offset, and based on the filtered spectrum density extracting the frequency corresponding to the main peak to obtain the corrected radar Doppler frequency shift, and calculating the actual water flow velocity corresponding to each primary flow velocity region based on the corrected radar Doppler frequency shift and the radar wavelength. The calculation expressions include: ; ; represents the filtered radar spectrum density corresponding to the primary flow velocity region with serial number j, represents the spectrum density corresponding to the radar spectrum data of the primary flow velocity region with serial number j, represents the turbulence intensity test data.

[0010] Preferably, the ultrasonic and radar data are respectively subjected to local state estimation through the federated Kalman filter. The radar Kalman weight factor and the ultrasonic Kalman full factor are calculated based on the least squares method of local covariance, and the global flow velocity is obtained through global fusion processing based on the radar weight factor and the ultrasonic full factor; ; wherein, represents the global flow velocity, m represents the total number of primary flow velocity regions, represents the radar Kalman weight factor corresponding to the primary flow velocity region with serial number j, It represents the ultrasonic Kalman weight factor corresponding to the secondary flow velocity region with serial number i, and n represents the total number of secondary flow velocity regions.

[0011] Preferably, the comprehensive water flow rate of the target blade is calculated based on the global flow velocity and the cross-section average flow velocity method. ; Among them, A represents the average area of the flow channel cross-section. It represents the velocity distribution coefficient. It represents the global flow velocity.

[0012] Preferably, the Reynolds number of the target blade is calculated based on the comprehensive water flow rate. When the Reynolds number of the target blade is greater than or equal to the preset friction threshold, pressure compensation is triggered. The real-time water flow friction factor is calculated based on the flow channel outlet pressure data in the piezoresistive sensor data. The theoretical friction factor under the design condition is calculated through the Colebrook-White equation. The corrected comprehensive water flow rate is calculated through the Darcy-Weisbach equation in combination with the real-time water flow friction factor, the theoretical friction factor, and the comprehensive water flow rate. Synchronously calculate the average flow rate, peak flow rate, and flow rate volatility of the corrected comprehensive water flow rate within a 5s window, and output the water flow rate feature set {corrected comprehensive water flow rate, average flow rate, peak flow rate, flow rate volatility}.

[0013] ; Among them, It represents the corrected comprehensive water flow rate. It represents the theoretical friction factor under the design condition. It represents the real-time water flow friction factor.

[0014] Advantages of the present invention: In view of the characteristics that the cooling channels of gas turbine blades are relatively large compared to the overall structure of an aero-turbine engine and the layout of the cooling channels is more complex, the present application adopts a method of first simulating and positioning and then actual measurement. Based on the velocity division, the deployment of radar sensors and ultrasonic sensors is completed, which is beneficial to overcome the problem of local measurement errors caused by the velocity gradient in the traditional method. Based on the CFD simulation results of the flow channels, high-frequency radars are deployed in the primary channels in the high-speed region to capture transient velocity changes, and ultrasonic sensors are used in the secondary channels in the low-speed region to compensate for the signal stability in the low signal-to-noise ratio environment, which is applicable to gas turbines. Combining one-dimensional CNN and adaptive filtering technology is beneficial to improving the quality of the collected signal data. For ultrasonic data, noise reduction based on a one-dimensional CNN network is beneficial to more accurately suppressing non-linear fluctuations; for the turbulence intensity adaptive filtering mechanism designed for radar signals, it is beneficial to ensure the spectral analysis accuracy affected by high-speed water flow. Innovatively introducing the federated Kalman filter framework for multi-source data fusion is beneficial to retaining the local feature advantages of each sensor and avoiding the problem of error accumulation caused by traditional simple weighted averaging. The dynamic friction factor compensation model automatically corrects the flow calculation results by combining the real-time monitoring data of piezoresistive sensors, which is beneficial to solving the measurement deviation problem of the change of the flow channel friction coefficient under high-temperature and high-pressure conditions. The system output contains a feature set of multi-dimensional parameters, which is beneficial to meeting the accurate measurement requirements and also provides multi-dimensional data support for the flow field stability analysis and fault warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 FIG. is a schematic diagram of the basic process of a method for measuring the water flow rate of a gas turbine blade provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following detailed description of the specific embodiments of the present invention will be given in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them.

[0017] Refer to Figure 1 , which is an embodiment of the present invention, and provides a method for measuring the water flow rate of a gas turbine blade, including: Step S1: Perform finite element modeling and simulation to obtain the theoretical flow rate of the inner cavity of the target blade; Step S2: Perform zoning processing based on the theoretical flow rate to obtain a primary flow velocity region and a secondary flow velocity region, and arrange a sensor group to collect test flow rate data; Step S3: Align the phases of the ultrasonic test data, perform noise reduction processing through a one-dimensional convolutional neural network, and calculate the actual water flow velocity corresponding to the second flow velocity region based on the sound velocity in the flow channel; Step S4, performing FFT processing on the radar test data, filtering processing using a turbulence intensity adaptive filtering algorithm, and calculating the actual water flow velocity corresponding to each primary flow velocity area based on the modified radar Doppler frequency shift; Step S5, establishing a federal Kalman filter to perform local state estimation, and performing global fusion processing based on radar weight factors and ultrasonic full factors to obtain a global flow velocity; Step S6, calculate the comprehensive water flow of the target blade based on the cross-sectional average flow velocity method, calculate the corrected comprehensive water flow based on the real-time water flow friction factor, and output the water flow feature set {corrected comprehensive water flow, average flow, peak flow, flow fluctuation rate}.

[0018] In this embodiment, the processing logic for finite element modeling and simulation includes: Importing the design drawing data of the target turbine blade into the CAD software, and performing preliminary modeling of the key areas of the flow channel to obtain an initial geometric model, wherein the key areas of the flow channel include guide grooves, cooling holes and fins; The initial geometric model is processed by unstructured tetrahedron and hexahedron hybrid meshing through the CFD platform, and local encryption is performed on the narrow part of the flow channel to control the Y+ value near the wall to be less than the preset mesh threshold. Multi-layer boundary layer meshes are set along the flow direction to obtain the wall shear stress and velocity gradient change data; Define boundary conditions, set the inlet flow rate, inlet static pressure and total temperature, set the turbulence intensity based on the Reynolds number, set the outlet static pressure and flow rate boundaries, define the temperature, thermal conductivity and convection heat transfer coefficient on the blade surface, initialize the full-field velocity and set the gravity acceleration to be turned on; The QUICK algorithm is used to discretize the flow terms, the PISO algorithm is used to perform pressure-velocity coupling, and the RANS model is calculated based on the residual convergence threshold to obtain the velocity cloud map, isosurface average velocity, cross-sectional average velocity, and analyze the cooling hole jet diffusion, vortex distribution between fins, and low-speed retention area to generate a velocity distribution map.

[0019] In this embodiment, a partition operation is performed on the target blade based on the velocity distribution diagram, and the processing logic of the partition operation includes: Segmentation is performed based on the geometric shape of the target blade flow channel to obtain geometric sub-regions, wherein the geometric shape includes an inlet buffer zone, a serpentine mainstream channel, a spoiler rib gap zone, an air film hole distribution zone, a high curvature bend zone, a cooling hole outlet diffusion zone, an outlet convergence zone, a fin tip zone, and a porous microstructure zone; Based on the velocity distribution diagram, the average velocity of each geometric sub-area is calculated to obtain a simulated velocity data set, and the simulated velocity data set is traversed. When the simulated velocity is greater than or equal to a preset velocity threshold, the corresponding geometric sub-area is divided into a secondary velocity area; When the simulated flow velocity is greater than or equal to the preset flow velocity threshold, the corresponding geometric sub-region is divided into a secondary flow velocity region.

[0020] In this embodiment, multi-dimensional sensors are set for the target blade based on the position coordinate data of the finite element model, and the processing logic includes: A high-frequency radar sensor and a laser Doppler sensor are arranged in the primary flow velocity region, an ultrasonic sensor is arranged in the secondary flow velocity region, and a piezoresistive sensor and a temperature sensor are arranged at the inlet and outlet of the target blade flow channel; Perform a hydraulic test on the target blade flow channel, clean the target blade flow channel and seal all interfaces, and inject water and exhaust air into the inlet of the target blade flow channel based on the preset first water pressure until the flow channel is filled with water and there is no bubble residue; Pressurized water injection is carried out into the inlet of the target blade flow channel based on the cooling medium pressure set for the target blade, and a test data set is obtained through the sensor. The test data set includes radar test data, turbulence intensity test data, ultrasonic test data, flow channel temperature data, and pressure data; In this embodiment, the actual water flow velocity of each primary flow velocity region is calculated based on the ultrasonic test data, and the processing logic includes: Perform 256-cycle phase alignment accumulation processing on the ultrasonic test data to obtain ultrasonic alignment data, and perform noise reduction processing on the ultrasonic alignment data through a one-dimensional convolutional neural network. The network structure of the one-dimensional convolutional neural network includes 3 convolutional layers and 1 fully connected layer. The convolution kernel size of the convolutional layer is set to 5*1, and the step size is set to 5. The fully connected layer is activated by the ReLU function. The input is the data corresponding to 1024-point ultrasonic test data, and the output is ultrasonic denoised data; In this embodiment, the speed of sound in the real-time flow channel is calculated based on the temperature data, and the actual water flow velocity corresponding to the second flow velocity region is calculated through the two-way time difference weighted average method based on the speed of sound in the flow channel and the ultrasonic denoised data. The calculation expression is: ; ; Wherein, represents the speed of sound in the flow channel corresponding to the secondary flow velocity region with serial number i, represents the water temperature, represents the actual water flow velocity corresponding to the secondary flow velocity region with serial number i, D represents the in-channel spacing of the ultrasonic sensor, represents the cosine value of the ultrasonic sound path inclination angle, represents the downstream reception time, represents the upstream reception time.

[0021] In this embodiment, the actual water flow velocities of each first-level flow velocity region are calculated based on radar test data, and the processing logic includes: performing fast Fourier transform processing on the radar test data to obtain radar spectrum data; Filtering the radar spectrum data through a turbulence intensity adaptive filtering algorithm to remove high-frequency noise and turbulence offset, and based on the filtered spectrum density extracting the frequency corresponding to the main peak to obtain the corrected radar Doppler shift, and calculating the actual water flow velocity corresponding to each first-level flow velocity region based on the corrected radar Doppler shift and the radar wavelength. The calculation expression includes: ; ; represents the filtered radar spectrum density corresponding to the first-level flow velocity region with serial number j, represents the spectrum density corresponding to the radar spectrum data corresponding to the first-level flow velocity region with serial number j, represents the turbulence intensity test data.

[0022] In this embodiment, local state estimation is respectively performed on ultrasonic and radar data through federated Kalman filtering, the radar Kalman weight factor and the ultrasonic Kalman full factor are calculated based on the least squares method of local covariance, and global fusion processing is performed based on the radar weight factor and the ultrasonic full factor to obtain the global flow velocity; ; wherein, represents the global flow velocity, m represents the total number of first-level flow velocity regions, represents the radar Kalman weight factor corresponding to the first-level flow velocity region with serial number j, represents the ultrasonic Kalman weight factor corresponding to the second-level flow velocity region with serial number i, and n represents the total number of second-level flow velocity regions.

[0023] In this embodiment, the comprehensive water flow rate of the target blade is calculated based on the global flow velocity and the cross-sectional average velocity method, ; wherein, A represents the average cross-sectional area of the flow channel, represents the velocity distribution coefficient, represents the global flow velocity.

[0024] In this embodiment, the target blade Reynolds number is calculated based on the comprehensive water flow rate. When the target blade Reynolds number is greater than or equal to the preset friction threshold, pressure compensation is triggered. The real-time water flow friction factor is calculated based on the flow channel outlet pressure data in the piezoresistive sensor data. The theoretical friction factor under the design condition is calculated through the Colebrook-White equation. The corrected comprehensive water flow rate is calculated through the Darcy-Weisbach equation in combination with the real-time water flow friction factor, the theoretical friction factor, and the comprehensive water flow rate. The average flow rate, peak flow rate, and flow rate volatility of the corrected comprehensive water flow rate within a 5s window are calculated synchronously, and the water flow rate feature set {corrected comprehensive water flow rate, average flow rate, peak flow rate, flow rate volatility} is output.

[0025] ; Among them, represents the corrected comprehensive water flow rate, represents the theoretical friction factor under the design condition, represents the real-time water flow friction factor.

[0026] Among them, considering the characteristics that the cooling flow channel of the gas turbine blade is relatively large relative to the overall structure of the aero-turbine engine and the cooling flow channel layout is more complex, this application adopts the method of first simulating and positioning and then actual measurement. Based on the velocity division, the deployment of radar sensors and ultrasonic sensors is completed, which is beneficial to overcoming the local measurement error problem caused by the velocity gradient in the traditional method. Based on the CFD simulation results of the flow channel, high-frequency radars are deployed in the primary flow channels in the high-speed area to capture the transient velocity changes, and ultrasonic sensors are used in the secondary flow channels in the low-speed area to compensate for the signal stability in the low signal-to-noise ratio environment, which is applicable to gas turbines. Combining one-dimensional CNN and adaptive filtering technology is beneficial to improving the quality of the collected signal data. For ultrasonic data, noise reduction based on the one-dimensional CNN network is beneficial to more accurately suppressing non-linear fluctuations; the turbulence intensity adaptive filtering mechanism designed for radar signals is beneficial to ensuring the spectral analysis accuracy affected by high-speed water flows. Innovatively introducing the federated Kalman filter framework for multi-source data fusion is beneficial to retaining the local feature advantages of each sensor and avoiding the error accumulation problem caused by traditional simple weighted averaging. The dynamic friction factor compensation model automatically corrects the flow rate calculation result by combining the real-time monitoring data of the piezoresistive sensor, which is beneficial to solving the measurement deviation problem of the change of the flow channel friction coefficient under high-temperature and high-pressure conditions. The system output contains a feature set of multi-dimensional parameters, which is beneficial to meeting the accurate measurement requirements and also provides multi-dimensional data support for the flow field stability analysis and fault warning.

[0027] Those skilled in the art should understand that the embodiments of the present invention may provide a method, a system or a computer program product. Therefore, the present invention may be implemented in the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present invention may be implemented in the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. Among them, the storage medium may be implemented based on any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. These computer program instructions may also be stored in a computer-readable memory capable of guiding a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the process Figure 1 in one process or multiple processes and / or blocks Figure 1 the functions specified in one block or multiple blocks.

[0028] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention may be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for measuring the water flow rate of a gas turbine blade, characterized in that, include: Step S1, performing finite element modeling simulation to obtain the target blade cavity theoretical flow rate; Step S2, performing zoning processing based on the theoretical flow to obtain a primary flow rate area and a secondary flow rate area, and arranging a sensor group to collect and obtain test flow data; Step S3, performing phase alignment on the ultrasonic test data, performing noise reduction processing through a one-dimensional convolutional neural network, and calculating the actual water flow rate corresponding to the second flow rate area based on the sound velocity in the flow channel; Step S4, performing FFT processing on the radar test data, filtering processing using a turbulence intensity adaptive filtering algorithm, and calculating the actual water flow velocity corresponding to each primary flow velocity area based on the modified radar Doppler frequency shift; Step S5, establishing a federal Kalman filter to perform local state estimation, and performing global fusion processing based on radar weight factors and ultrasonic full factors to obtain a global flow velocity; Step S6: Calculate the comprehensive water flow of the target blade based on the cross-sectional average flow velocity method, calculate the corrected comprehensive water flow based on the real-time water flow friction factor, and output the water flow feature set.

2. The water flow rate measurement method for a gas turbine blade according to claim 1, wherein: The processing logic for finite element modeling and simulation includes: Importing the design drawing data of the target turbine blade into the CAD software, and performing preliminary modeling of the key areas of the flow channel to obtain an initial geometric model, wherein the key areas of the flow channel include guide grooves, cooling holes and fins; The initial geometric model is processed by unstructured tetrahedron and hexahedron hybrid meshing through the CFD platform, and local encryption is performed on the narrow part of the flow channel to control the Y+ value near the wall to be less than the preset mesh threshold. Multi-layer boundary layer meshes are set along the flow direction to obtain the wall shear stress and velocity gradient change data; Define boundary conditions, set the inlet flow rate, inlet static pressure and total temperature, set the turbulence intensity based on the Reynolds number, set the outlet static pressure and flow rate boundaries, define the temperature, thermal conductivity and convection heat transfer coefficient on the blade surface, initialize the full-field velocity and set the gravity acceleration to be turned on; The QUICK algorithm is used to discretize the flow terms, the PISO algorithm is used to perform pressure-velocity coupling, and the RANS model is calculated based on the residual convergence threshold to obtain the velocity cloud map, isosurface average velocity, cross-sectional average velocity, and analyze the cooling hole jet diffusion, vortex distribution between fins, and low-speed retention area to generate a velocity distribution map.

3. The water flow measurement method for a gas turbine blade according to claim 1, wherein: The target blade is partitioned based on the velocity distribution diagram, and the processing logic of the partition operation includes: Segmentation is performed based on the geometric shape of the target blade flow channel to obtain geometric sub-regions, wherein the geometric shape includes an inlet buffer zone, a serpentine mainstream channel, a spoiler rib gap zone, an air film hole distribution zone, a high curvature bend zone, a cooling hole outlet diffusion zone, an outlet convergence zone, a fin tip zone, and a porous microstructure zone; Based on the velocity distribution diagram, the average velocity of each geometric sub-area is calculated to obtain a simulated velocity data set, and the simulated velocity data set is traversed. When the simulated velocity is greater than or equal to a preset velocity threshold, the corresponding geometric sub-area is divided into a secondary velocity area; When the simulated flow velocity is greater than or equal to a preset flow velocity threshold, the corresponding geometric sub-region is divided into a secondary flow velocity region.

4. The water flow measurement method for a gas turbine blade according to claim 3, characterized in that: The multi-dimensional sensor is set for the target blade based on the position coordinate data of the finite element model. The processing logic includes: Arrange high-frequency radar sensors and laser Doppler sensors in the first flow velocity region, ultrasonic sensors in the second flow velocity region, and piezoresistive sensors and temperature sensors at the inlet and outlet of the target blade flow channel; Conduct a hydraulic test on the target blade flow channel, clean the target blade flow channel and seal all interfaces, and inject water and exhaust air into the inlet of the target blade flow channel based on a preset first water pressure until the flow channel is filled with water and there is no bubble residue; Pressurize and inject water into the inlet of the target blade flow channel based on the cooling medium pressure set for the target blade, and obtain a test data set through the sensors. The test data set includes radar test data, turbulence intensity test data, ultrasonic test data, flow channel temperature data, and pressure data.

5. A method for measuring the water flow rate of a gas turbine blade according to claim 1, characterized in that: Calculate the actual water flow velocity of each first flow velocity region based on the ultrasonic test data. The processing logic includes: Perform 256-cycle phase alignment and accumulation processing on the ultrasonic test data to obtain ultrasonic alignment data, and perform noise reduction processing on the ultrasonic alignment data through a one-dimensional convolutional neural network. The network structure of the one-dimensional convolutional neural network includes 3 convolutional layers and 1 fully connected layer. The convolution kernel size of the convolutional layer is set to 5*1, and the stride is set to 5. The fully connected layer is activated by the ReLU function. The input is the data corresponding to 1024-point ultrasonic test data, and the output is ultrasonic denoised data.

6. A method for measuring the water flow rate of a gas turbine blade according to claim 5, wherein: Calculate the sound speed in the real-time flow channel based on the temperature data, and calculate the actual water flow velocity corresponding to the second flow velocity region through the bidirectional time difference weighted average method based on the sound speed in the flow channel and the ultrasonic denoised data. The calculation expression is: ； ； Among them, represents the sound speed in the flow channel corresponding to the secondary flow velocity region with serial number i, represents the water temperature, represents the actual water flow velocity corresponding to the secondary flow velocity region with serial number i, D represents the distance between ultrasonic sensors in the flow channel, represents the cosine value of the ultrasonic sound path inclination angle, represents the downstream reception time, represents the upstream reception time.

7. A method for measuring the water flow rate of a gas turbine blade according to claim 1, wherein: Calculate the actual water flow velocity of each first flow velocity region based on the radar test data. The processing logic includes: performing fast Fourier transform processing on the radar test data to obtain radar spectrum data; Filter the radar spectrum data through the turbulence intensity adaptive filtering algorithm to remove high-frequency noise and turbulence offset. Based on the filtered spectral density Extract the frequency corresponding to the main peak to obtain the corrected radar Doppler shift. Calculate the actual water flow velocity corresponding to each first-level flow velocity region based on the corrected radar Doppler shift and the radar wavelength. Its calculation expression includes: ； ； It represents the filtered radar spectral density corresponding to the first-level flow velocity region with serial number j. It represents the spectral density corresponding to the radar spectral data corresponding to the first-level flow velocity region with serial number j. It represents the test data of turbulence intensity.

8. A method for measuring the water flow rate of a gas turbine blade according to claim 1, characterized in that: Perform local state estimation on the ultrasonic and radar data respectively through the federated Kalman filter, calculate the radar Kalman weight factor and the ultrasonic Kalman full factor based on the least squares method of the local covariance, and perform global fusion processing based on the radar weight factor and the ultrasonic full factor to obtain the global flow velocity; ； Among them, represents the global flow velocity, m represents the total number of primary flow velocity regions, represents the radar Kalman weight factor corresponding to the primary flow velocity region with serial number j, represents the ultrasonic Kalman weight factor corresponding to the secondary flow velocity region with serial number i, and n represents the total number of secondary flow velocity regions.

9. The method for measuring the water flow rate of a gas turbine blade according to claim 1, wherein: The comprehensive water flow rate of the target blade is calculated based on the global flow velocity and the cross-sectional average flow velocity method, ; where A represents the average cross-sectional area of the flow channel, represents the velocity distribution coefficient, represents the global flow velocity.

10. A method for measuring the water flow rate of a gas turbine blade according to claim 1, characterized in that: Calculate the Reynolds number of the target blade based on the comprehensive water flow rate. When the Reynolds number of the target blade is greater than or equal to the preset friction threshold, trigger pressure compensation. Calculate the real-time water flow friction factor based on the flow channel outlet pressure data in the piezoresistive sensor data, calculate the theoretical friction factor under the design condition through the Colebrook-White equation, and calculate the corrected comprehensive water flow rate through the Darcy-Weisbach equation in combination with the real-time water flow friction factor, the theoretical friction factor, and the comprehensive water flow rate; Synchronously calculate the average flow rate, peak flow rate, and flow rate volatility of the corrected comprehensive water flow rate within a 5s window, and output the water flow rate feature set {corrected comprehensive water flow rate, average flow rate, peak flow rate, flow rate volatility}; The calculation expression of the corrected comprehensive water flow rate is: ； Among them, represents the corrected comprehensive water flow rate, represents the theoretical friction factor under the design condition, represents the real-time water flow friction factor.

Images