Aircraft engine turbine blade temperature field estimation method based on machine learning

Abstract

The invention is suitable for the technical field of aero-engines, and provides an aero-engine turbine blade temperature field estimation method based on machine learning. The method comprises the following steps: selecting a calculation sample point based on an engine height-Mach number flight envelope; matching and determining the relationship between the engine key parameters and the engine key variables; determining the relationship between the engine key variable and the turbine temperature field; respectively creating an MLP neural network of which the input is the engine key parameter and the output is the engine key variable, and the input is the engine key variable and the output is the turbine temperature field. According to the method, all working condition points of the aero-engine can be covered, key variables and a turbine temperature field can be obtained at the same time by inputting key parameters of the engine, the key variables can be obtained in the engine test run process, the effectiveness of a prediction model can be conveniently checked, and compared with traditional overall performance matching and numerical calculation, the calculation efficiency can be greatly improved; the design period is shortened; and the time cost is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of aero-engines, and in particular relates to a method for estimating the temperature field of a turbine blade of an aero-engine based on machine learning. Background Art

[0002] As one of the hot end components with the highest load during the operation of aircraft engines, turbine blades are in a high temperature, high pressure, and high speed working state for a long time. Especially when the aircraft transitions from one working state to another, the rapidly changing incoming flow conditions cause the temperature of the turbine blades to deviate greatly from the design state, even exceeding the pre-designed value, which may cause blade thermal fatigue or ablation, which will directly affect the working life and safety performance of the engine. Therefore, the rapid prediction of the temperature field of the engine turbine blades is of great significance to the design and development of the engine.

[0003] At present, in the prediction method of turbine blade temperature field, on the one hand, the blade structure is greatly simplified, and although the analytical calculation method can be used for rapid prediction, the prediction accuracy is low; on the other hand, the finite element method can obtain high-fidelity results, but the calculation cycle is long. Therefore, it is difficult to quickly predict the temperature field of the turbine blades of contemporary aircraft engines with extremely complex structures. Summary of the invention

[0004] The purpose of an embodiment of the present invention is to provide a method for estimating the temperature field of an aircraft engine turbine blade based on machine learning, aiming to solve the technical problems existing in the prior art mentioned in the background technology.

[0005] The embodiment of the present invention is implemented as follows:

[0006] A method for estimating temperature field of a turbine blade of an aero-engine based on machine learning, the method specifically comprising the following steps:

[0007] Select calculation sample points based on the engine altitude-Mach number flight envelope;

[0008] Based on the overall parameters of the aircraft engine, match and determine the relationship between the key parameters of the engine and the key variables of the engine;

[0009] Based on numerical simulation methods, determine the relationship between key engine variables and turbine temperature field;

[0010] Create MLP neural networks with engine key parameters as input and engine key variables as output, and turbine temperature field as input.

[0011] As a further limitation of the technical solution of the embodiment of the present invention, the calculation sample points are selected based on the engine altitude-Mach number flight envelope, and the calculation sample points cover the upper and lower limits of the key variables, wherein the lower limits are: Ma=0, H=0; Ma is minimum, H is maximum; Ma=0, the working point of H is maximum; the upper limits are: Ma is maximum, H=0; Ma is maximum, H is maximum within the flight envelope; Ma is maximum, the working point of H is maximum within the flight envelope.

[0012] As a further limitation of the technical solution of the embodiment of the present invention, the key parameters of the engine include: the total inlet temperature T in , total inlet pressure P in , compressor rotor speed n and flow w in The key variables of the engine include: combustion chamber outlet temperature T hot , compressor outlet temperature and compressor inlet flow w; the matching determines the relationship between the key engine parameters and the key engine variables using the Newton-Raphson algorithm based aviation engine performance prediction.

[0013] As a further limitation of the technical solution of the embodiment of the present invention, the process of the numerical simulation method includes: establishing a turbine blade model, dividing the grid, determining boundary conditions, calculating and solving, and post-processing.

[0014] As a further limitation of the technical solution of the embodiment of the present invention, the MLP neural network adopts a multi-layer perceptron architecture, including an input layer, a hidden layer and an output layer, wherein: the input layer is used to receive engine key parameters and engine key variables; the hidden layer is used to extract engine information, and the output layer is used for engine prediction results.

[0015] As a further limitation of the technical solution of the embodiment of the present invention, the steps of the MLP neural network are:

[0016] The calculated sample points are standardized and divided into a training set and a test set according to a preset ratio;

[0017] Construct a multi-layer perceptron neural network model, including input layer, hidden layer and output layer;

[0018] According to a plurality of training optimization factors, the multi-layer perceptron neural network model is trained and optimized to obtain an optimal neural network prediction model;

[0019] The training set is input into the neural network prediction model for training, and the model is evaluated in combination with the test set, with the root mean square error being used as an effective evaluation indicator of the neural network prediction effect.

[0020] As a further limitation of the technical solution of the embodiment of the present invention, the formula for the normalization process is:

[0021]

[0022] Among them, y is the data after standardization, x is the calculation sample point, μ is the data mean of the calculation sample point, and θ is the data standard deviation of the calculation sample point.

[0023] As a further limitation of the technical solution of the embodiment of the present invention, the formula of the multi-layer perceptron neural network model is:

[0024]

[0025] Among them, x k is the input data, b j is the bias term, a j is the output data, w jk is the weight and σ is the activation function.

[0026] As a further limitation of the technical solution of the embodiment of the present invention, the multiple training optimization factors include: batch size, training algorithm selection, learning rate, number of iterations and loss function.

[0027] As a further limitation of the technical solution of the embodiment of the present invention, the formula of the effective evaluation index is:

[0028]

[0029] Among them, MSE is an effective evaluation index, N is the number of training samples, T i is the true value of the i-th sample, T ii is the predicted value of the i-th sample.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] The embodiment of the present invention selects calculation sample points based on the engine altitude-Mach number flight envelope; matches and determines the relationship between the key parameters of the engine and the key variables of the engine; determines the relationship between the key variables of the engine and the turbine temperature field; and respectively creates an MLP neural network whose input is the key parameters of the engine and output is the key variables of the engine, and whose input is the key variables of the engine and output is the turbine temperature field. It can cover all operating points of the aircraft engine. By inputting the key parameters of the engine, the key variables and the turbine temperature field can be obtained at the same time, and the key variables can be obtained during the engine test, which is convenient for testing the effectiveness of the prediction model. Compared with the traditional overall performance matching and numerical calculation, it can greatly improve the calculation efficiency, thereby shortening the design period and reducing the time cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1A flow chart of a method for estimating temperature field of a turbine blade of an aero-engine based on machine learning provided by an embodiment of the present invention is shown;

[0033] Figure 2 The flight envelope diagram of Mach number Ma-altitude H of an aircraft engine provided by an embodiment of the present invention is shown;

[0034] Figure 3 The structure diagram of the MLP neural network provided by the embodiment of the present invention is shown. DETAILED DESCRIPTION

[0035] 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 embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0036] It is understandable that, in the current prediction methods of turbine blade temperature field, on the one hand, the blade structure is greatly simplified, and although the analytical calculation method can be used for rapid prediction, the prediction accuracy is low; on the other hand, the finite element method can obtain high-fidelity results, but the calculation cycle is long. Therefore, it is difficult to quickly predict the temperature field of the turbine blades of contemporary aircraft engines with extremely complex structures.

[0037] To solve the above problems, the embodiment of the present invention discloses a method for estimating the temperature field of an aircraft engine turbine blade based on machine learning. The method selects calculation sample points based on the engine altitude-Mach number flight envelope; matches and determines the relationship between the key parameters of the engine and the key variables of the engine based on the overall parameters of the aircraft engine; determines the relationship between the key variables of the engine and the turbine temperature field based on the numerical simulation method; and creates an MLP neural network with the input of the key parameters of the engine and the output of the key variables of the engine, and the input of the key variables of the engine and the output of the turbine temperature field. The method can cover all operating points of the aircraft engine. By inputting the key parameters of the engine, the key variables and the turbine temperature field can be obtained at the same time, and the key variables can be obtained during the engine test, which is convenient for testing the effectiveness of the prediction model. Compared with the traditional overall performance matching and numerical calculation, the calculation efficiency can be greatly improved, thereby shortening the design period and reducing the time cost.

[0038] Specifically, Figure 1 A flow chart of a method for estimating the temperature field of an aircraft engine turbine blade based on machine learning provided in an embodiment of the present invention is shown.

[0039] In a preferred embodiment of the present invention, a method for estimating the temperature field of a turbine blade of an aircraft engine based on machine learning is provided, and the method specifically comprises the following steps:

[0040] Step S101, selecting calculation sample points based on the engine altitude-Mach number flight envelope.

[0041] In the embodiment of the present invention, based on the engine altitude-Mach number flight envelope, calculation sample points are selected, and the selection of calculation sample points needs to cover the upper and lower limits of key variables, wherein the lower limits are: Ma=0, H=0; Ma minimum, H maximum; Ma=0, H maximum working point; the upper limits are: Ma maximum, H=0; Ma maximum, H maximum within the flight envelope; Ma maximum, H maximum working point within the flight envelope, such as Figure 2 A flight envelope diagram of Mach number Ma-altitude H of an aircraft engine provided by an embodiment of the present invention is shown.

[0042] Step S102: Based on the overall parameters of the aircraft engine, match and determine the relationship between the key parameters of the engine and the key variables of the engine.

[0043] In the embodiment of the present invention, based on the overall parameters of the aircraft engine, the relationship between the key parameters of the engine and the key variables of the engine is matched and determined. Specifically, the key parameters of the engine include: the total inlet temperature T in , total inlet pressure P in , compressor rotor speed n and flow w in etc.; The key variables of the engine include: combustion chamber outlet temperature T hot , compressor outlet temperature and compressor inlet flow w, etc.; match and determine the relationship between key engine parameters and key engine variables, and use the Newton-Raphson algorithm to predict the performance of aviation engines.

[0044] Step S103: Determine the relationship between the key engine variables and the turbine temperature field based on a numerical simulation method.

[0045] In an embodiment of the present invention, based on a numerical simulation method, the relationship between the key engine variables and the turbine temperature field is determined. Specifically, the process of the numerical simulation method includes: establishing a turbine blade model, dividing the grid, determining boundary conditions, calculating and solving, and post-processing.

[0046] Step S104, creating an MLP neural network whose input is the key parameters of the engine and whose output is the key variables of the engine, and whose input is the key variables of the engine and whose output is the turbine temperature field.

[0047] In the embodiment of the present invention, an MLP neural network is created whose input is the key parameters of the engine and whose output is the key variables of the engine, and an MLP neural network is created whose input is the key variables of the engine and whose output is the turbine temperature field, such as Figure 3The structure diagram of the MLP neural network provided by the embodiment of the present invention is shown. The MLP neural network adopts a multi-layer perceptron architecture, including an input layer, a hidden layer and an output layer. The input layer is used to receive key engine parameters and key engine variables; the hidden layer is used to extract engine information, and the output layer is used for engine prediction results. Specifically, the steps of the MLP neural network are:

[0048] Step 1: Standardize the calculated sample points and divide them into training set and test set according to the preset ratio. The formula for standardization is:

[0049]

[0050] Among them, y is the data after standardization, x is the calculation sample point, μ is the data mean of the calculation sample point, and θ is the data standard deviation of the calculation sample point;

[0051] Step 2: Construct a multi-layer perceptron neural network model, including input layer, hidden layer and output layer. The formula of the multi-layer perceptron neural network model is:

[0052]

[0053] Among them, x k is the input data, b j is the bias term, a j is the output data, w jk is the weight, σ is the activation function;

[0054] Step 3: According to multiple training optimization factors, the multi-layer perceptron neural network model is trained and optimized to obtain the best neural network prediction model. The multiple training optimization factors include: batch size, training algorithm selection, learning rate, number of iterations and loss function;

[0055] Step 4: Input the training set into the neural network prediction model for training, and evaluate the model with the test set. The root mean square error is used as an effective evaluation index for the prediction effect of the neural network. The formula for the effective evaluation index is:

[0056]

[0057] Among them, MSE is an effective evaluation index, N is the number of training samples, T i is the true value of the i-th sample, T ii is the predicted value of the i-th sample.

[0058] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0059] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0060] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A method for estimating the temperature field of aeroengine turbine blades based on machine learning, characterized in that: The method specifically comprises the following steps: Select calculation sample points based on the engine altitude-Mach number flight envelope; Based on the overall parameters of the aircraft engine, match and determine the relationship between the key parameters of the engine and the key variables of the engine; Based on numerical simulation methods, determine the relationship between key engine variables and turbine temperature field; Create MLP neural networks with engine key parameters as input and engine key variables as output, and turbine temperature field as input.

2. The method for estimating temperature field of aircraft engine turbine blades based on machine learning according to claim 1, characterized in that: The calculation sample points are selected based on the engine altitude-Mach number flight envelope, and the calculation sample points cover the upper and lower limits of key variables, wherein the lower limits are: Ma=0, H=0; Ma is minimum, H is maximum; Ma=0, the working point of H is maximum; the upper limits are: Ma is maximum, H=0; Ma is maximum, H is maximum within the flight envelope; Ma is maximum, the working point of H is maximum within the flight envelope.

3. The method for estimating temperature field of aircraft engine turbine blades based on machine learning according to claim 1, characterized in that: The key parameters of the engine include: inlet total temperature T in , total inlet pressure P in , compressor rotor speed n and flow w in The key variables of the engine include: combustion chamber outlet temperature T hot , compressor outlet temperature and compressor inlet flow w; the matching determines the relationship between the key engine parameters and the key engine variables using the Newton-Raphson algorithm based aviation engine performance prediction.

4. The method for estimating temperature field of aircraft engine turbine blades based on machine learning according to claim 1, characterized in that: The process of the numerical simulation method includes: establishing a turbine blade model, dividing a grid, determining boundary conditions, calculating and solving, and post-processing.

5. The method for estimating temperature field of aircraft engine turbine blades based on machine learning according to claim 1, characterized in that: The MLP neural network adopts a multi-layer perceptron architecture, including an input layer, a hidden layer and an output layer, wherein: the input layer is used to receive engine key parameters and engine key variables; the hidden layer is used to extract engine information; and the output layer is used for engine prediction results.

6. The method for estimating temperature field of aircraft engine turbine blades based on machine learning according to claim 5, characterized in that: The steps of the MLP neural network are: The calculated sample points are standardized and divided into a training set and a test set according to a preset ratio; Construct a multi-layer perceptron neural network model, including input layer, hidden layer and output layer; According to a plurality of training optimization factors, the multi-layer perceptron neural network model is trained and optimized to obtain an optimal neural network prediction model; The training set is input into the neural network prediction model for training, and the model is evaluated in combination with the test set, with the root mean square error being used as an effective evaluation indicator of the neural network prediction effect.

7. The method for estimating temperature field of aircraft engine turbine blades based on machine learning according to claim 6, characterized in that: The formula for the standardization process is: Among them, y is the data after standardization, x is the calculation sample point, μ is the data mean of the calculation sample point, and θ is the data standard deviation of the calculation sample point.

8. The method for estimating temperature field of aircraft engine turbine blades based on machine learning according to claim 6, characterized in that: The formula of the multi-layer perceptron neural network model is: Among them, x k is the input data, b j is the bias term, a j is the output data, w jk is the weight and σ is the activation function.

9. The method for estimating temperature field of aircraft engine turbine blades based on machine learning according to claim 6, characterized in that: The plurality of training optimization factors include: batch size, training algorithm selection, learning rate, number of iterations, and loss function.

10. The method for estimating temperature field of aircraft engine turbine blades based on machine learning according to claim 6, characterized in that: The formula of the effective evaluation index is: Among them, MSE is an effective evaluation index, N is the number of training samples, T i is the true value of the i-th sample, T ii is the predicted value of the i-th sample.

Images