A method for measuring water flow rate of gas turbine blades

Abstract

The invention discloses a method for measuring water flow of gas turbine blades, which relates to the technical field of gas turbine blades. The method comprises the following steps: performing finite element modeling and simulation to obtain a theoretical flow rate in a target blade cavity; performing zoning processing based on the theoretical flow rate to obtain a primary flow velocity region and a secondary flow velocity region, and arranging a sensor group to collect and obtain test flow data; performing phase alignment on ultrasonic test data, performing noise reduction processing through a one-dimensional convolutional neural network, and obtaining an actual water flow velocity in the second flow velocity region based on the sound velocity in the flow channel; performing FFT processing on radar test data, and obtaining an actual water flow velocity in each primary flow velocity region through turbulence intensity adaptive filtering and corrected radar Doppler frequency shift calculation; establishing a federal Kalman filter to perform global fusion processing to obtain a global flow velocity; calculating the comprehensive water flow of the target blade based on a cross-sectional average flow velocity method, calculating the corrected comprehensive water flow based on a real-time water flow friction factor, and outputting a water flow feature set.

Description

Technical Field

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

[0002] Heavy-duty gas turbine blades are exposed to extreme high temperatures and high pressures for extended periods. Their internal cooling channels, such as serpentine channels and film holes, are typically designed to achieve forced cooling through the flow of high-pressure air or steam. Insufficient or uneven flow distribution can lead to localized cooling failure, causing material creep, oxidation, and even ablation, directly impacting blade life. Inner cavity flow measurement is used to assess the cooling gas flow capacity within the blade cavity, which directly impacts blade cooling effectiveness. In recent years, with the continuous advancement of science and technology, temperatures upstream of turbines have continued to rise. This requires not only improved process technology but also structural design refinements, leading to continuous refinement of turbine blade cooling design and cooling effectiveness testing. Domestic research on water flow measurement equipment has primarily focused on measuring inner cavity flow within aircraft engine turbine blades. Limited research is available on ground-based gas turbine blade inner cavity water flow measurement. Furthermore, existing domestic research solutions for water flow measurement equipment still have limitations, such as human error in timing, susceptibility of flow measurement to pressure fluctuations, and long intervals between testing one blade and the next, resulting in low efficiency.

[0003] Currently, a Chinese invention patent application with application number CN110823323A discloses a method for correcting water flow measurement values ​​for turbine rotor blades. This application includes the following steps: defining a relationship between the blade inlet and outlet pressure differentials during actual water flow measurement; uniformly selecting multiple different blade inlet and outlet pressure differentials and calculating the corresponding water flow values; fitting the formula using a function to obtain a theoretical flow equation; measuring the water flow values ​​of the blade under multiple different inlet and outlet pressure differentials using a measuring device; fitting the formula using a function to obtain an actual flow equation; obtaining a flow correction coefficient; and correcting the water flow value of the blade at a specified inlet pressure based on the flow correction coefficient. This method for correcting water flow measurement values ​​for turbine rotor blades takes into account the variability in water flow measurement across different production lines, rationally correcting the measured water flow results, and achieving the goal of directly using water flow measurement results to determine the casting conformity of the internal cavity channel of the turbine rotor blade. However, this application is intended for aircraft engine turbine blades, and there are differences between ground-based gas turbine blades and aircraft engine turbine blades. Summary of the Invention

[0004] The technical problem solved by the present invention is that the existing technology for measuring water flow 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, so the existing technology is insufficiently applicable to measuring water flow 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:

[0006] Step S1, performing finite element modeling simulation to obtain the target blade cavity theoretical flow rate;

[0007] Step S2: performing partitioning based on the theoretical flow rate to obtain a primary flow rate area and a secondary flow rate area, and arranging a sensor group to collect test flow rate data;

[0008] Step S3: performing phase alignment on the ultrasonic test data, performing noise reduction processing using a one-dimensional convolutional neural network, and calculating the actual water flow rate corresponding to the second flow rate zone based on the sound velocity in the flow channel;

[0009] Step S4: Perform FFT processing on the radar test data, filter it using a turbulence intensity adaptive filtering algorithm, and calculate the actual water flow rate corresponding to each primary flow rate area based on the corrected radar Doppler frequency shift;

[0010] Step S5: Establish a federated Kalman filter to perform local state estimation, and perform global fusion processing based on the radar weight factor and the ultrasonic weight factor to obtain the global flow velocity;

[0011] 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}.

[0012] Preferably, the processing logic for finite element modeling and simulation includes:

[0013] 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, including the guide grooves, cooling holes and fins, to obtain an initial geometric model;

[0014] The initial geometric model was meshed with unstructured tetrahedrons and hexahedrons using the CFD platform. Local encryption was 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 were set along the flow direction to obtain wall shear stress and velocity gradient change data.

[0015] 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 convective heat transfer coefficient on the blade surface, initialize the full-field velocity, and set gravity acceleration to be enabled;

[0016] The flow terms are discretized using the QUICK algorithm, and pressure-velocity coupling is performed using the PISO algorithm. 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.

[0017] Preferably, a partition operation is performed on the target blade based on the velocity distribution map, and the processing logic of the partition operation includes:

[0018] The geometric sub-regions are obtained by segmenting the target blade flow channel based on its geometric shape, wherein the geometric shape includes 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 porous microstructure zone;

[0019] Based on the velocity distribution diagram, the average velocity of each geometric sub-area is calculated to obtain a simulated velocity data set. 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.

[0020] When the simulated flow velocity is less than the preset flow velocity threshold, the corresponding geometric sub-region is divided into a primary flow velocity region.

[0021] Preferably, a multi-dimensional sensor is set for the target blade based on the position coordinate data of the finite element model, and the processing logic includes:

[0022] High-frequency radar sensors and laser Doppler sensors are arranged for the first-level flow velocity area, ultrasonic sensors are arranged for the second-level flow velocity area, and piezoresistive sensors and temperature sensors are arranged for the target blade flow channel inlet and outlet;

[0023] Perform a hydraulic test on the target blade flow channel, clean the target blade flow channel and seal all interfaces, and inject and exhaust water into the target blade flow channel inlet based on a preset first water pressure until the flow channel is full of water and no bubbles remain;

[0024] Based on the cooling medium pressure set for the target blade, pressurized water is injected into the flow channel inlet of 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;

[0025] Preferably, the actual water flow rate in each secondary flow rate area is calculated based on the ultrasonic test data, and the processing logic includes:

[0026] The ultrasonic test data is subjected to 256-cycle phase alignment accumulation processing to obtain ultrasonic alignment data, and the ultrasonic alignment data is denoised using a one-dimensional convolutional neural network. The one-dimensional convolutional neural network structure includes three convolutional layers and one fully connected layer. The convolution kernel size of the convolution 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 the 1024-point ultrasonic test data, and the output is the ultrasonic denoised data.

[0027] Preferably, the real-time sound velocity in the flow channel is calculated based on the temperature data, and the actual water flow velocity corresponding to the second flow velocity zone is calculated by a two-way time difference weighted average method based on the sound velocity in the flow channel and the ultrasonic denoising data. The calculation expression is:

[0028] ;

[0029] ;

[0030] in, Indicates the sound velocity in the flow channel corresponding to the secondary flow velocity region with serial number i, Indicates water temperature, It represents the actual water flow rate corresponding to the secondary flow rate area with serial number i, D represents the distance in the flow channel of the ultrasonic sensor, Indicates the cosine value of the ultrasonic sound path inclination angle, Indicates the downstream receiving time, Indicates the countercurrent reception time.

[0031] Preferably, the actual water flow rate of each primary flow rate area is calculated based on the radar test data, and the processing logic includes: performing fast Fourier transform processing on the radar test data to obtain radar spectrum data;

[0032] The radar spectrum data is filtered by the turbulence intensity adaptive filtering algorithm to remove high-frequency noise and turbulence offset. The frequency corresponding to the main peak is extracted to obtain the corrected radar Doppler shift. Based on the corrected radar Doppler shift and the radar wavelength, the actual water flow velocity corresponding to each first-level flow velocity area is calculated. The calculation expression includes:

[0033] ;

[0034] ;

[0035] represents the filtered radar spectrum density corresponding to the first-order velocity region with sequence number j, Indicates the spectrum density of the radar spectrum data corresponding to the first-level velocity region with sequence number j, Represents turbulence intensity test data.

[0036] Preferably, local state estimation is performed on ultrasonic and radar data respectively by using a federated Kalman filter, radar Kalman weights and ultrasonic Kalman weights are calculated based on the least squares method of local covariance, and global fusion processing is performed based on the radar weights and ultrasonic weights to obtain the global flow velocity;

[0037] ;

[0038] in, represents the global flow velocity, m represents the total number of first-level flow velocity areas, Indicates the radar Kalman weight factor corresponding to the first-level velocity region with sequence 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.

[0039] Preferably, the comprehensive water flow of the target blade is calculated based on the global flow velocity and cross-sectional average flow velocity method;

[0040] ;

[0041] Where A represents the average cross-sectional area of ​​the flow channel, represents the velocity distribution coefficient, Represents the global flow velocity.

[0042] Preferably, a target blade Reynolds number is calculated based on the comprehensive water flow rate, pressure compensation is triggered when the target blade Reynolds number is greater than or equal to a preset friction threshold, a real-time water flow friction factor is calculated based on the flow channel outlet pressure data in the piezoresistive sensor data, a theoretical friction factor under the design working condition is calculated using the Colebrook-White equation, and a corrected comprehensive water flow rate is calculated by combining the real-time water flow friction factor, the theoretical friction factor, and the comprehensive water flow rate using the Darcy-Weisbach equation;

[0043] Synchronously calculate the average flow rate, peak flow rate and flow fluctuation rate of the corrected comprehensive water flow within a 5s window, and output the water flow feature set {corrected comprehensive water flow rate, average flow rate, peak flow rate, flow fluctuation rate};

[0044] ;

[0045] in, represents the corrected integrated water flow, represents the theoretical friction factor under design conditions, Represents the real-time water flow friction factor.

[0046] Beneficial effects of the present invention: In view of the fact that the cooling channels of gas turbine blades are relatively large compared to the overall structure of aviation turbine engines and the layout of the cooling channels is more complex, this application adopts a method of simulating positioning first and then actual measurement, and completes the deployment of radar sensors and ultrasonic sensors based on velocity division, which is conducive to overcoming the local measurement error problem caused by velocity gradient in traditional methods. Based on the flow channel CFD simulation results, a high-frequency radar is deployed in the first-level flow channel in the high-speed area to capture transient flow velocity changes, and an ultrasonic sensor is used in the second-level flow channel in the low-speed area to compensate for the signal stability in a low signal-to-noise ratio environment, which is suitable for gas turbines. Combining one-dimensional CNN with adaptive filtering technology is conducive to improving the quality of collected signal data. For ultrasonic data, noise reduction based on the one-dimensional CNN network is conducive to more accurate suppression of nonlinear fluctuations; the turbulence intensity adaptive filtering mechanism designed for radar signals is conducive to ensuring the accuracy of spectral analysis affected by high-speed water flow. The innovative introduction of the federal Kalman filter framework for multi-source data fusion is conducive to retaining the local feature advantages of each sensor and avoiding the error accumulation problem caused by traditional simple weighted averaging. A dynamic friction factor compensation model, combined with real-time data from piezoresistive sensors, automatically corrects flow calculations, helping to address measurement deviations caused by variations in flow friction coefficient under high-temperature and high-pressure conditions. The system outputs a feature set containing multidimensional parameters, meeting precise measurement requirements and providing multi-dimensional data support for flow field stability analysis and fault warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A schematic diagram of the basic flow of a method for measuring water flow on a gas turbine blade provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0049] Reference Figure 1 , as one embodiment of the present invention, provides a method for measuring water flow on gas turbine blades, comprising:

[0050] Step S1, performing finite element modeling simulation to obtain the target blade cavity theoretical flow rate;

[0051] Step S2: performing partitioning based on the theoretical flow rate to obtain a primary flow rate area and a secondary flow rate area, and arranging a sensor group to collect test flow rate data;

[0052] Step S3: performing phase alignment on the ultrasonic test data, performing noise reduction processing using a one-dimensional convolutional neural network, and calculating the actual water flow rate corresponding to the second flow rate zone based on the sound velocity in the flow channel;

[0053] Step S4: Perform FFT processing on the radar test data, filter it using a turbulence intensity adaptive filtering algorithm, and calculate the actual water flow rate corresponding to each primary flow rate area based on the corrected radar Doppler frequency shift;

[0054] Step S5: Establish a federated Kalman filter to perform local state estimation, and perform global fusion processing based on the radar weight factor and the ultrasonic weight factor to obtain the global flow velocity;

[0055] 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}.

[0056] In this embodiment, the processing logic for performing finite element modeling and simulation includes:

[0057] 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, including the guide grooves, cooling holes and fins, to obtain an initial geometric model;

[0058] The initial geometric model was meshed with unstructured tetrahedrons and hexahedrons using the CFD platform. Local encryption was 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 were set along the flow direction to obtain wall shear stress and velocity gradient change data.

[0059] 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 convective heat transfer coefficient on the blade surface, initialize the full-field velocity, and set gravity acceleration to be enabled;

[0060] The flow terms are discretized using the QUICK algorithm, and pressure-velocity coupling is performed using the PISO algorithm. 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.

[0061] In this embodiment, a partition operation is performed on the target blade based on the velocity distribution map. The processing logic of the partition operation includes:

[0062] The geometric sub-regions are obtained by segmenting the target blade flow channel based on its geometric shape, wherein the geometric shape includes 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 porous microstructure zone;

[0063] Based on the velocity distribution diagram, the average velocity of each geometric sub-area is calculated to obtain a simulated velocity data set. 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.

[0064] When the simulated flow velocity is less than the preset flow velocity threshold, the corresponding geometric sub-region is divided into a primary flow velocity region.

[0065] In this embodiment, a multi-dimensional sensor is set for the target blade based on the position coordinate data of the finite element model, and the processing logic includes:

[0066] High-frequency radar sensors and laser Doppler sensors are arranged for the first-level flow velocity area, ultrasonic sensors are arranged for the second-level flow velocity area, and piezoresistive sensors and temperature sensors are arranged for the target blade flow channel inlet and outlet;

[0067] Perform a hydraulic test on the target blade flow channel, clean the target blade flow channel and seal all interfaces, and inject and exhaust water into the target blade flow channel inlet based on a preset first water pressure until the flow channel is full of water and no bubbles remain;

[0068] Based on the cooling medium pressure set for the target blade, pressurized water is injected into the flow channel inlet of 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;

[0069] In this embodiment, the actual water flow rate in each secondary flow rate area is calculated based on the ultrasonic test data. The processing logic includes:

[0070] The ultrasonic test data is subjected to 256-cycle phase alignment accumulation processing to obtain ultrasonic alignment data, and the ultrasonic alignment data is denoised using a one-dimensional convolutional neural network. The one-dimensional convolutional neural network structure includes three convolutional layers and one fully connected layer. The convolution kernel size of the convolution 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 the 1024-point ultrasonic test data, and the output is the ultrasonic denoised data.

[0071] In this embodiment, the real-time sound velocity in the flow channel is calculated based on the temperature data. The actual water flow rate corresponding to the second flow rate zone is calculated by using the two-way time difference weighted average method based on the sound velocity in the flow channel and the ultrasonic denoising data. The calculation expression is:

[0072] ;

[0073] ;

[0074] in, Indicates the sound velocity in the flow channel corresponding to the secondary flow velocity region with serial number i, Indicates water temperature, It represents the actual water flow rate corresponding to the secondary flow rate area with serial number i, D represents the distance in the flow channel of the ultrasonic sensor, Indicates the cosine value of the ultrasonic sound path inclination angle, Indicates the downstream receiving time, Indicates the countercurrent reception time.

[0075] In this embodiment, the actual water flow rate of each primary flow rate area is calculated based on the radar test data, and the processing logic includes: performing fast Fourier transform processing on the radar test data to obtain radar spectrum data;

[0076] The radar spectrum data is filtered by the turbulence intensity adaptive filtering algorithm to remove high-frequency noise and turbulence offset. The frequency corresponding to the main peak is extracted to obtain the corrected radar Doppler shift. Based on the corrected radar Doppler shift and the radar wavelength, the actual water flow velocity corresponding to each first-level flow velocity area is calculated. The calculation expression includes:

[0077] ;

[0078] ;

[0079] represents the filtered radar spectrum density corresponding to the first-order velocity region with sequence number j, Indicates the spectrum density of the radar spectrum data corresponding to the first-level velocity region with sequence number j, Represents turbulence intensity test data.

[0080] In this embodiment, the local state estimation of ultrasonic and radar data is performed separately through the federated Kalman filter. The radar Kalman weight factor and the ultrasonic Kalman weight factor are calculated based on the least squares method of the local covariance. The global flow velocity is obtained by performing a global fusion process based on the radar weight factor and the ultrasonic weight factor.

[0081] ;

[0082] in, represents the global flow velocity, m represents the total number of first-level flow velocity areas, Indicates the radar Kalman weight factor corresponding to the first-level velocity region with sequence 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.

[0083] In this embodiment, the comprehensive water flow of the target blade is calculated based on the global flow velocity and cross-sectional average flow velocity method;

[0084] ;

[0085] Where A represents the average cross-sectional area of ​​the flow channel, represents the velocity distribution coefficient, Represents the global flow velocity.

[0086] In this embodiment, a target blade Reynolds number is calculated based on the integrated water flow rate. When the target blade Reynolds number is greater than or equal to a preset friction threshold, pressure compensation is triggered. A real-time water flow friction factor is calculated based on the flow channel outlet pressure data from the piezoresistive sensor data. The theoretical friction factor under the design operating conditions is calculated using the Colebrook-White equation. A corrected integrated water flow rate is calculated using the Darcy-Weisbach equation by combining the real-time water flow friction factor, the theoretical friction factor, and the integrated water flow rate.

[0087] Synchronously calculate the average flow rate, peak flow rate and flow fluctuation rate of the corrected comprehensive water flow within a 5s window, and output the water flow feature set {corrected comprehensive water flow rate, average flow rate, peak flow rate, flow fluctuation rate};

[0088] ;

[0089] in, represents the corrected integrated water flow, represents the theoretical friction factor under design conditions, Represents the real-time water flow friction factor.

[0090] Given the relatively large size and complex layout of gas turbine blade cooling channels compared to the overall structure of aircraft turbine engines, this application employs a method of simulated positioning followed by actual measurement. This method utilizes velocity-based segmentation to deploy radar and ultrasonic sensors, effectively overcoming the local measurement errors caused by velocity gradients in traditional methods. Based on CFD simulation results, a high-frequency radar is deployed in the high-speed primary channel to capture transient velocity changes. Ultrasonic sensors are used in the low-speed secondary channel to compensate for signal instability in low signal-to-noise ratio environments, making this method suitable for gas turbines. Combining one-dimensional CNN with adaptive filtering technology improves the quality of collected signal data. For ultrasonic data, noise reduction based on a one-dimensional CNN network facilitates more precise suppression of nonlinear fluctuations. An adaptive turbulence intensity filtering mechanism designed for radar signals ensures accurate spectral resolution of high-speed flow. The innovative introduction of a federated Kalman filter framework for multi-source data fusion preserves the local characteristics of each sensor and avoids the error accumulation caused by simple weighted averaging. A dynamic friction factor compensation model, combined with real-time monitoring data from piezoresistive sensors, automatically corrects flow calculations, effectively addressing measurement errors associated with changes in flow friction coefficient under high-temperature and high-pressure conditions. The system output contains a feature set of multi-dimensional parameters, which is conducive to meeting the needs of precise measurement and also provides multi-dimensional data support for flow field stability analysis and fault warning.

[0091] Those skilled in the art will appreciate that embodiments of the present invention may provide methods, systems, or computer program products. Therefore, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage media may be 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 storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0092] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, and all of these should be included in the scope of the claims of the present invention.

Claims

1. A method for measuring water flow on gas turbine blades, characterized in that: include: Step S1, performing finite element modeling simulation to obtain the target blade cavity theoretical flow rate; Step S2: performing partitioning based on the theoretical flow rate to obtain a primary flow rate area and a secondary flow rate area, and arranging a sensor group to collect test flow rate data; Step S3: performing phase alignment on the ultrasonic test data, performing noise reduction processing using a one-dimensional convolutional neural network, and calculating the actual water flow rate corresponding to the second flow rate zone based on the sound velocity in the flow channel; Step S4: Perform FFT processing on the radar test data, filter it using a turbulence intensity adaptive filtering algorithm, and calculate the actual water flow rate corresponding to each primary flow rate area based on the corrected radar Doppler frequency shift; Step S5: Establish a federated Kalman filter to perform local state estimation, and perform global fusion processing based on the radar weight factor and the ultrasonic weight factor to obtain the 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; In step S2, a high-frequency radar sensor and a laser Doppler sensor are arranged for the primary flow velocity area, an ultrasonic sensor is arranged for the secondary flow velocity area, and a piezoresistive sensor and a temperature sensor are arranged for the target blade flow channel inlet and flow channel outlet; The test data set is obtained through sensor acquisition, and the test data set includes radar test data, turbulence intensity test data, ultrasonic test data, flow channel temperature data and pressure data.

2. A method for measuring water flow on gas turbine blades according to claim 1, characterized in that: 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, including the guide grooves, cooling holes and fins, to obtain an initial geometric model; The initial geometric model was meshed with unstructured tetrahedrons and hexahedrons using the CFD platform. Local encryption was 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 were set along the flow direction to obtain 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 convective heat transfer coefficient on the blade surface, initialize the full-field velocity, and set gravity acceleration to be enabled; The flow terms are discretized using the QUICK algorithm, and pressure-velocity coupling is performed using the PISO algorithm. 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 method for measuring water flow on a gas turbine blade according to claim 1, wherein: The target blades are partitioned based on the velocity distribution map. The processing logic of the partitioning operation includes: The geometric sub-regions are obtained by segmenting the target blade flow channel based on its geometric shape, wherein the geometric shape includes 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 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. 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 less than the preset flow velocity threshold, the corresponding geometric sub-region is divided into a primary flow velocity region.

4. A method for measuring water flow on gas turbine blades according to claim 3, characterized in that: A multi-dimensional sensor is set for the target blade based on the position coordinate data of the finite element model. The processing logic includes: High-frequency radar sensors and laser Doppler sensors are arranged for the first-level flow velocity area, ultrasonic sensors are arranged for the second-level flow velocity area, and piezoresistive sensors and temperature sensors are arranged for the target blade flow channel inlet and outlet; Perform a hydraulic test on the target blade flow channel, clean the target blade flow channel and seal all interfaces, and inject and exhaust water into the target blade flow channel inlet based on a preset first water pressure until the flow channel is full of water and no bubbles remain; Based on the cooling medium pressure set for the target blade, pressurized water is injected into the flow channel inlet of 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.

5. The method for measuring water flow on a gas turbine blade according to claim 1, wherein: The actual water flow rate in each secondary flow rate area is calculated based on the ultrasonic test data. The processing logic includes: The ultrasonic test data is subjected to 256-cycle phase alignment accumulation processing to obtain ultrasonic alignment data, and the ultrasonic alignment data is denoised by a one-dimensional convolutional neural network. The one-dimensional convolutional neural network structure includes 3 convolutional layers and 1 fully connected layer. The convolution kernel size of the convolution 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 the 1024-point ultrasonic test data, and the output is the ultrasonic denoised data.

6. A method for measuring water flow on gas turbine blades according to claim 5, characterized in that: The real-time sound velocity in the flow channel is calculated based on the temperature data. The actual water flow velocity corresponding to the second flow velocity zone is calculated using the two-way time difference weighted average method based on the sound velocity in the flow channel and the ultrasonic denoising data. The calculation expression is: ； ； in, Indicates the sound velocity in the flow channel corresponding to the secondary flow velocity region with serial number i, Indicates water temperature, It represents the actual water flow rate corresponding to the secondary flow rate area with serial number i, D represents the distance in the flow channel of the ultrasonic sensor, Indicates the cosine value of the ultrasonic sound path inclination angle, Indicates the downstream receiving time, Indicates the countercurrent reception time.

7. The method for measuring water flow on gas turbine blades according to claim 1, wherein: The actual water flow rate of each primary flow rate area 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; The radar spectrum data is filtered by the turbulence intensity adaptive filtering algorithm to remove high-frequency noise and turbulence offset. The frequency corresponding to the main peak is extracted to obtain the corrected radar Doppler shift. Based on the corrected radar Doppler shift and the radar wavelength, the actual water flow velocity corresponding to each first-level flow velocity area is calculated. The calculation expression includes: ； ； represents the filtered radar spectrum density corresponding to the first-order velocity region with sequence number j, Indicates the spectrum density of the radar spectrum data corresponding to the first-level velocity region with sequence number j, Represents turbulence intensity test data.

8. The method for measuring water flow on gas turbine blades according to claim 1, wherein: The local state of ultrasonic and radar data is estimated separately through federated Kalman filtering. The radar Kalman weight factor and ultrasonic Kalman weight factor are calculated based on the least squares method of local covariance. The global flow velocity is obtained by global fusion processing based on the radar weight factor and ultrasonic weight factor. ； in, represents the global flow velocity, m represents the total number of first-level flow velocity areas, Indicates the radar Kalman weight factor corresponding to the first-level velocity region with sequence 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.

9. The method for measuring water flow on gas turbine blades according to claim 1, wherein: The comprehensive water flow of the target blade is calculated based on the global flow velocity and 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. The method for measuring water flow on gas turbine blades according to claim 1, wherein: 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 from the piezoresistive sensor data. The theoretical friction factor under the design working condition is calculated using the Colebrook-White equation. The corrected comprehensive water flow rate is calculated by combining the real-time water flow friction factor, the theoretical friction factor, and the comprehensive water flow rate using the Darcy-Weisbach equation; Synchronously calculate the average flow rate, peak flow rate and flow fluctuation rate of the corrected comprehensive water flow within a 5s window, and output the water flow feature set {corrected comprehensive water flow rate, average flow rate, peak flow rate, flow fluctuation rate}; The calculation expression of the corrected comprehensive water flow is: ； in, represents the corrected integrated water flow, represents the theoretical friction factor under design conditions, Represents the real-time water flow friction factor.

Images