Geometric parameter optimization method, system and equipment for turbine blade and medium

Abstract

The invention relates to the technical field of gas turbines, and discloses a geometric parameter optimization method, system and equipment for turbine blades and a medium. The method comprises the following steps: modeling the turbine blade according to geometric parameters, working condition parameters and material parameters of the turbine blade to obtain a plurality of first input samples with first fidelity and a plurality of second input samples with second fidelity; obtaining performance parameters of the turbine blade corresponding to each first input sample and each second input sample as a first response sample and a second response sample; and training the gas turbine performance prediction network, and predicting the performance parameters of the target turbine blade through the trained gas turbine performance prediction network according to the geometric parameters, the working condition parameters and the material parameters of the target turbine blade. And optimizing the geometric parameters of the target turbine blade by taking the maximum reliability and robustness of the target turbine blade indicated by the predicted performance parameters as a target.

Description

Technical Field

[0001] The present invention relates to the technical field of gas turbines, and in particular to a method, system, equipment and medium for optimizing the geometric parameters of turbine blades. Background Art

[0002] Gas turbines are the core power equipment of modern industry. They are characterized by efficient power generation, rapid response and clean energy utilization. They are an important part of the modern power system. At the same time, gas turbines can provide powerful power for aircraft engines, ship propulsion and oil and gas pipeline transportation. They are also reliable power systems for military equipment such as fighter jets and ships. They are of great significance to national energy security and national defense security.

[0003] Gas turbine blades are exposed to a harsh environment of high temperature and high speed for a long time. Due to the influence of power grid fluctuations, operating condition changes and actual environmental factors, there are many uncertainties in the operating conditions of the blades. These uncertainties not only increase the risk of blade failure under extreme conditions, but also lead to performance degradation under high-frequency conditions. Therefore, reliability design optimization of the blades is crucial to ensure safe and efficient operation of the unit.

[0004] However, in the robust optimization problem of turbine blade reliability, the input includes multi-source uncertainty parameters such as geometry and operation, and the output covers multiple response parameters such as aerodynamic performance, strength characteristics, and fatigue life. The method of obtaining a high-precision proxy model by constructing a database requires a lot of time and computing resources to obtain sufficient sample data, and the computational cost is high. Summary of the Invention

[0005] The purpose of the present invention is to provide a method, system, device and medium for optimizing the geometric parameters of turbine blades, which can solve the problem that the method of obtaining a high-precision proxy model consumes a lot of time and computing resources to obtain sufficient sample data, resulting in high computing costs.

[0006] To solve the above technical problems, an embodiment of the present invention provides a method for optimizing geometric parameters of turbine blades, comprising the following steps: Modeling the turbine blade according to geometric parameters, operating parameters, and material parameters of the turbine blade to obtain a plurality of first input samples having a first fidelity and a plurality of second input samples having a second fidelity; wherein the first fidelity is less than the second fidelity; Obtaining performance parameters of the turbine blade corresponding to each first input sample and each second input sample, respectively, as a first response sample and a second response sample; A gas turbine performance prediction network comprising a first fidelity prediction network and a correction network is trained using a plurality of first input samples, a plurality of second input samples, a plurality of first response samples, and a plurality of second response samples; wherein, during training, the first fidelity prediction network is trained using the first input samples and the first response samples so that the first fidelity prediction network predicts first response samples corresponding to the first input samples, and the correction network is trained using the first response samples predicted by the first fidelity prediction network and the second input samples so that the correction network predicts second response samples corresponding to the first input samples; The trained gas turbine performance prediction network predicts the performance parameters of the target turbine blades based on the geometric parameters, operating parameters and material parameters of the target turbine blades, and optimizes the geometric parameters of the target turbine blades with the goal of maximizing the reliability and robustness of the target turbine blades indicated by the predicted performance parameters.

[0007] Optionally, the gas turbine performance prediction network is trained using an active learning sampling method, which aims to minimize the failure probability estimation error of the target turbine blade indicated by the second response sample predicted by the gas turbine performance prediction network, and determines the location of the newly added sampling points and the fidelity of the newly added sampling points; wherein the performance parameters of the turbine blades are used to indicate the failure probability of the turbine blades.

[0008] Optionally, the active learning sampling method determines the position of the newly added sampling point and the fidelity of the newly added sampling point by the following formula: ; ; ; ; Where, x is the input vector, t represents the fidelity of the gas turbine performance prediction network, and represent the first fidelity and the second fidelity respectively, is the adaptive learning function, It represents the probability that the sample point is accurately judged by the gas turbine performance prediction network. c ( t ) is a cost function, which represents the acquisition cost ratio of the first input sample and the second input sample, is the cross-correlation function, is the probability density function, is the distance function, is the mean of the predicted values, is the variance of the predicted value,x new To add new sample points and t new is the corresponding fidelity.

[0009] Optionally, the method of predicting the performance parameters of the target turbine blades based on the geometric parameters, operating parameters, and material parameters of the target turbine blades using the trained gas turbine performance prediction network, and optimizing the geometric parameters of the target turbine blades with the goal of maximizing the reliability and robustness of the target turbine blades indicated by the predicted performance parameters, includes: The trained gas turbine performance prediction network predicts the performance parameters of the target turbine blade under different operating conditions based on the geometric parameters, material parameters, and different operating condition parameters of the target turbine blade. The performance parameters include the power, efficiency, stress, and strain of the target turbine blade. Obtaining the failure probability of the target turbine blade according to the stress and strain of the target turbine blade; With the goal of maximizing the mean and minimizing the variance of the power and efficiency of the target turbine blade under different operating conditions and the failure probability of the target turbine blade as a constraint, the geometric parameters of the target turbine blade are optimized.

[0010] Optionally, the geometric parameters include the geometric parameters of the root area and the bottom fillet area of ​​the turbine blade, the operating parameters include the total gas inlet temperature, total gas inlet pressure, gas outlet flow, total cooling air inlet temperature, total cooling air inlet pressure and rotational speed of the gas turbine, and the material parameters include the density, elastic modulus, Poisson's ratio, thermal conductivity and thermal conductivity of the turbine blade.

[0011] An embodiment of the present invention further provides a system for optimizing geometric parameters of turbine blades, comprising: an input sample acquisition module, configured to model the turbine blade according to geometric parameters, operating parameters, and material parameters of the turbine blade, and obtain a plurality of first input samples having a first fidelity and a plurality of second input samples having a second fidelity; wherein the first fidelity is less than the second fidelity; a response sample acquisition module, configured to acquire the performance parameters of the turbine blade corresponding to each first input sample and each second input sample, as a first response sample and a second response sample; a network training module, configured to train a gas turbine performance prediction network comprising a first fidelity prediction network and a correction network using a plurality of first input samples, a plurality of second input samples, a plurality of first response samples, and a plurality of second response samples; wherein during training, the first fidelity prediction network is trained using the first input samples and the first response samples so as to predict first response samples corresponding to the first input samples by the first fidelity prediction network, and the correction network is trained using the first response samples predicted by the first fidelity prediction network and the second input samples so as to predict second response samples corresponding to the first input samples by the correction network; The parameter optimization module is used to predict the performance parameters of the target turbine blades based on the geometric parameters, operating parameters and material parameters of the target turbine blades through the trained gas turbine turbine performance prediction network, and optimize the geometric parameters of the target turbine blades with the goal of maximizing the reliability and robustness of the target turbine blades indicated by the predicted performance parameters.

[0012] An embodiment of the present invention also provides a computer device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned method for optimizing the geometric parameters of turbine blades.

[0013] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the above-mentioned method for optimizing geometric parameters of turbine blades.

[0014] The method for optimizing the geometric parameters of turbine blades provided by the present invention has at least the following beneficial effects: For the reliability robust optimization problem of turbine blades, two types of sample data with different fidelity are constructed: a first input sample with a first fidelity (low-fidelity sample data) and its corresponding first response sample, and a second input sample with a second fidelity (high-fidelity sample data) and its corresponding second response sample. When training the gas turbine performance prediction network (i.e., the proxy model), the low-fidelity sample data and its response data are used to train the low-fidelity prediction network to predict the response data corresponding to the low-fidelity sample data. Then, the predicted value of the low-fidelity prediction network and the high-fidelity sample data are used to correct its predicted value, thereby outputting response data with high fidelity. That is, the low-fidelity proxy model actually performs the prediction, and only the high-fidelity sample data is used for correction. This fully utilizes the complementary advantages of high-fidelity model data and low-fidelity model data, and improves processing efficiency and reduces computational cost while ensuring model prediction accuracy.

[0015] Based on this model, the geometric parameters of the turbine blades are optimized to achieve reliability design optimization of the turbine blades, so that they maintain efficient and stable performance and solve the reliability robust optimization problem of the turbine blades. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0017] Figure 1 A schematic flow chart of a method for optimizing geometric parameters of turbine blades provided by the present invention; Figure 2 A schematic diagram of the structure of a gas turbine performance prediction network provided by the present invention; Figure 3 A schematic diagram of a training process for a gas turbine performance prediction network provided by the present invention; Figure 4 A schematic diagram of a double-layer cycle flow chart for robust optimization of reliability of gas turbine blades provided by the present invention. DETAILED DESCRIPTION

[0018] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with specific embodiments of the present invention and corresponding drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0019] At present, the geometric parameter optimization problem of turbine blades faces multiple challenges: first, its input includes multi-source uncertainty parameters such as geometry and operation, and the output covers multiple response parameters such as aerodynamic performance, strength characteristics, fatigue life, etc. The method of obtaining a high-precision proxy model by constructing a database consumes a lot of time and computing resources to obtain sufficient sample data, and the computational cost is high; second, the robust optimization problem involves multiple target parameters, and the reliability constraint involves solving probability integrals, which makes reliability robust optimization a high-dimensional, multi-peak nonlinear programming problem. Traditional optimization algorithms are inefficient and prone to falling into local optimality; finally, the physical field data of gas turbines has the characteristics of ultra-high dimensionality, strong nonlinearity and spatiotemporal coupling. Traditional proxy models often directly map input parameters to performance parameters, with low prediction accuracy and lack of physical interpretability.

[0020] Therefore, establishing a fast, accurate and efficient robust optimization method for gas turbine blade reliability is of great significance for shortening the blade design cycle, reducing the performance fluctuation of the turbine under the influence of multi-source uncertainty factors, and ensuring the safe and stable operation of the gas turbine.

[0021] The technical solutions provided by various embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0022] One embodiment of the present invention relates to a method for optimizing the geometric parameters of a turbine blade. The specific process of the method for optimizing the geometric parameters of a turbine blade in this embodiment can be as follows: Figure 1 Shown, including: Step 101: Modeling a turbine blade based on geometric parameters, operating parameters, and material parameters of the turbine blade to obtain a plurality of first input samples having a first fidelity and a plurality of second input samples having a second fidelity; wherein the first fidelity is less than the second fidelity; Step 102 : Acquire the performance parameters of the turbine blade corresponding to each first input sample and each second input sample as a first response sample and a second response sample.

[0023] Step 103: Use multiple first input samples, multiple second input samples, multiple first response samples, and multiple second response samples to train a gas turbine performance prediction network including a first fidelity prediction network and a correction network; wherein, during training, the first input samples and the first response samples are used to train the first fidelity prediction network so as to predict the first response samples corresponding to the first input samples through the first fidelity prediction network, and the first response samples predicted by the first fidelity prediction network and the second input samples are used to train the correction network so as to predict the second response samples corresponding to the first input samples through the correction network.

[0024] In step 104, the performance parameters of the target turbine blades are predicted based on the geometric parameters, operating parameters, and material parameters of the target turbine blades using the trained gas turbine performance prediction network, and the geometric parameters of the target turbine blades are optimized with the goal of maximizing the reliability and robustness of the target turbine blades indicated by the predicted performance parameters.

[0025] The following is a detailed description of the implementation details of the method for optimizing the geometric parameters of turbine blades in this embodiment. The following content is only provided for ease of understanding and is not necessary for implementing this solution.

[0026] In step 101, a reasonable numerical calculation method is first used to determine the weak parts of the turbine blade, and based on the actual situation of the blade, the turbine blade geometric parameter design space and the operating parameters and material parameter environment space are determined. First, the turbine blade is subjected to a gas-thermal-solid calculation to determine that the blade root area and the fillet area at the bottom of the blade body are weak parts. Then, the weak parts are parametrically modeled to form a geometric parameter design space. D , which is expressed as follows:

[0027] ; Where, g is the leaf root area parameter; r is the parameter of the fillet area at the bottom of the blade; the superscripts - and + indicate the lower and upper limits of the parameter, respectively.

[0028] Determine the operating parameters and material parameter environment space based on the actual operating conditions of the blade C , which is expressed as follows: ; Where, c The operating parameters include gas inlet total temperature, gas inlet total pressure, gas outlet flow, cooling air inlet total temperature, cooling air inlet total pressure, speed, etc. m are the blade material parameters, including density, elastic modulus, Poisson's ratio, thermal conductivity, thermal conductivity, etc.

[0029] Then, Latin hypercube sampling is used to generate high- and low-fidelity sample sets. The proxy models constructed with traditional single-fidelity data sources often have an accuracy-efficiency trade-off problem during training. Specifically, the high-fidelity model HF can produce high-confidence results by virtue of its precise modeling of physical problems, such as refined grids in FEM, direct numerical models in CFD, full-scale experimental studies, etc., and such high-fidelity models are often accompanied by high computational / experimental costs, resulting in severe limitations on sample size. In contrast, the low-fidelity model introduces simplification methods to greatly reduce computational / experimental costs at the expense of some accuracy, such as the use of coarse grids, empirical formulas, and experiments using characteristic simulation parts. Such low-fidelity models LF can quickly generate a large amount of data, but their inherent simplification errors may lead to systematic offsets, which are particularly significant when it comes to complex problems such as strong nonlinearity and multi-field coupling. In this embodiment, in the geometric parameter design space D and working parameters and material parameters environment space C A small number of initial sample sets are constructed using the Latin hypercube sampling method YB .

[0030] In step 102, numerical calculation methods are used to solve the true responses of high-fidelity and low-fidelity samples and construct training data sets. YB , using a fine grid to solve the true response of the sample R 1. Use a coarse grid to solve the true response of the sample R 2, and process the samples and response results into the format required for network training. In this embodiment, the high-fidelity and low-fidelity samples differ only in the degree of mesh refinement. The geometric parameters, operating parameters, and material parameters are used to calculate the response of the current sample, including the power and efficiency of the turbine stage, the temperature field parameters of the turbine blades, and the stress and strain of the weak parts of the turbine blades. The format of the training data set needs to be constructed according to the actual situation. In this embodiment, the low-fidelity prediction model LF is a fully connected deep neural network, the high-fidelity prediction model HF is a graph convolutional neural network, and the low-fidelity training data set is T LF and high-fidelity training datasets T HF They are as follows:

[0031] ; ; Where, I is the sample number matrix, N is the location information and node attribute set of the nodes in each sample, A is the adjacency matrix set of the connection relationship between nodes in each sample.

[0032] In step 103, a gas turbine performance prediction network based on a multi-fidelity hybrid network is constructed and trained. Multi-fidelity modeling systematically integrates the complementary advantages of high-fidelity model HF and low-fidelity model LF data. At the same time, traditional surrogate models are usually limited by their prior assumptions, while deep neural networks still have good prediction accuracy when facing high-dimensional and strongly nonlinear problems. Establishing a gas turbine performance prediction network based on a multi-fidelity hybrid network can provide a new solution to the accuracy-efficiency trade-off problem in the process of high-precision and rapid prediction of gas turbine performance. Figure 2 The gas turbine performance prediction network GT-MFHN is composed of two sub-networks connected in series, the low-fidelity prediction network Used to predict the value of low-fidelity data, correct the network It is used to fit the complex mapping relationship from low-fidelity data to high-fidelity data. In this embodiment, the low-fidelity prediction network A fully connected deep neural network with multiple hidden layers is used. Its input is the design variables, and its output is the predicted low-fidelity turbine performance parameters. The specific mapping is:

[0033] ; Where, xis the input vector, including geometric parameters, working parameters and material parameters, is the predicted value of low-fidelity data, including turbine blade stress, strain, turbine power, efficiency, etc. For low-fidelity prediction networks The parameters to be learned in .

[0034] Correction Network A graph convolutional neural network containing multiple graph convolutional hidden layers is used, whose input is the design variables and the low-fidelity prediction network. The predicted value of , the output is the predicted high-fidelity turbine performance parameters, the specific mapping is: ; Where, is the predicted value of high-fidelity data, To correct the parameters to be learned in the network.

[0035] The gas turbine performance prediction network GT-MFHN is trained using supervised learning. Based on the phased training strategy, the specific steps are as follows:

[0036] S31, using low-fidelity training datasets T LF Updating the low-fidelity prediction network The learnable parameters , minimize the network loss function Low-fidelity prediction networks The training process is as follows:

[0037] ; Where, For low-fidelity prediction networks The loss function is defined as the true value of the low-fidelity data Predicted value with low-fidelity data The mean square error of , and the regularization loss is introduced, which is defined as follows: ; Where, M is the number of low-fidelity training samples, λ LF For low-fidelity prediction networks Regularization rate, w LF For low-fidelity prediction networks Weight coefficient. The product of the sum of the squares of the network weight coefficients and the regularization rate is added to force the model to reduce parameter redundancy while minimizing the prediction error, so as to balance the trade-off between model complexity and training error and prevent network overfitting.

[0038] S32: Maintaining Low-Fidelity Prediction Networks Parameter freezing runs as an offline proxy model, leveraging high-fidelity training datasets T HF Update and correct the network The learnable parameters , minimize the network loss function . Correction Network The training process is as follows:

[0039] ; ; Where, To correct the network The loss function is P is the number of high-fidelity training samples, is the true value of high-fidelity data, is the predicted value of high-fidelity data, λ HF To correct the network Regularization rate, w HF To correct the network Weight coefficient.

[0040] S33: Based on the active learning sampling method, an adaptive learning function AU suitable for multi-fidelity proxy models is constructed. For reliability analysis problems, the core of active learning sampling lies in how to design the optimal sampling strategy so that the proxy model can more accurately identify the boundary between the failure domain and the safety domain, that is, the limit state surface, until the convergence criteria related to the accuracy of failure probability assessment are met. Its mathematical essence is to construct a Bayesian optimization problem with the goal of minimizing the failure probability estimation error. For the gas turbine turbine performance prediction network GT-MFHN based on a multi-fidelity hybrid network, its active learning process should have a dual decision-making dimension: in addition to determining the location of the new sample points, it is also necessary to determine the fidelity level required to update the sample, that is, what fidelity model should be called in the active learning process, so as to maximize the overall proxy model accuracy while minimizing the sample acquisition cost. In order to solve the above problems, the present invention proposes an adaptive learning function AU suitable for multi-fidelity proxy models, and its mathematical expression is as follows:

[0041] ; ; Where, t represents the model fidelity, c ( t ) is the cost function, which represents the acquisition cost ratio of low-fidelity data to high-fidelity data. Indicates the probability that the sample point is accurately judged by the current proxy model. is the cross-correlation function, is the probability density function, is the distance function, is the mean of the predicted values, is the variance of the predicted value.

[0042] According to the established adaptive learning function AU, new sample points can be obtained x new and the corresponding fidelity t new , whose expression is as follows: ; The convergence criterion of the adaptive learning function AU is also important. Too aggressive will lead to insufficient prediction accuracy of the proxy model, while too conservative will cause unnecessary waste of computing resources. The present invention is based on the convergence criterion of stability. When the prediction value of high-fidelity data is If no significant changes occur during the iteration process, the model is considered to have converged and the addition of points is stopped. The mathematical expression is as follows:

[0043] in, is the convergence threshold, and its value should be reasonably selected according to the actual situation. In this embodiment, its value is 0.01.

[0044] S34: Using the adaptive learning function AU, the gas turbine performance prediction network GT-MFHN is trained. Figure 3 , based on the existing sample set, using the low-fidelity prediction network obtained in S31 and S32 and correction network Predict the results and judge the model prediction results based on the stability convergence criterion. If the model does not converge, the adaptive learning function AU is used to determine the new sample points. x new and the corresponding fidelity t new , numerical calculation methods are used to calculate the true response of the relevant fidelity model, which is added to the initial sample set and the low-fidelity prediction network is retrained and correction network , and re-use the stability convergence criterion to judge the model prediction results; if the model converges, save the current low-fidelity prediction network and correction network The learnable parameter results are used to complete the training of the gas turbine performance prediction network GT-MFHN.

[0045] In step 104, based on the trained gas turbine performance prediction network GT-MFHN, a double-layer loop framework is used to perform reliability robust optimization on the gas turbine blades. Figure 4 This paper proposes a two-layer loop framework. The inner layer uses the Monte Carlo sampling method (MC) to obtain the mean and variance of turbine performance parameters, as well as the failure probability of turbine blades. The outer layer, based on the data obtained from the inner layer, optimizes the turbine geometric parameters with the goal of maximizing the mean and minimizing the variance of the turbine performance parameters. The outer layer also sets the failure probability of turbine blades as a constraint and performs a loop optimization of the turbine geometric parameters. The specific steps are as follows:

[0046] S41: Construct the inner Monte Carlo loop. g and the fillet area parameters at the bottom of the blade r , using Monte Carlo sampling MC, in the working condition parameters and material parameter environment space C Internal working parameters c and blade material parameters m Sampling is performed, and based on the high-precision gas turbine performance prediction network GT-MFHN trained in S4, each operating point is quickly predicted to obtain the turbine power, efficiency, and turbine blade stress, strain and other parameters at each operating point.

[0047] S42: Calculate the current geometric parameters in the working condition parameters and material parameter environment space C The uncertainty response of turbine performance parameters is quantified by the mean and variance. At the same time, the life of the turbine blades at each operating point is calculated based on the relevant life prediction model, and the failure probability of the turbine blades is calculated. The mean and variance calculation formulas of the performance parameters are as follows:

[0048] ; ; Where, Performance parameters l The uncertainty response mean, Performance indicators l Uncertainty response standard deviation, performance parameters l Including turbine power W and efficiency wait, The number of samples for Monte Carlo sampling MC.

[0049] The fatigue life of gas turbine blades can be expressed as an implicit function of stress or strain N f ( z ), then the limit state function of its fatigue reliability evaluation is g ( z ) can be expressed as: ; Where, The defined safe life represents the design life under a certain safety margin. Its specific value needs to be determined based on multiple factors such as design requirements, risk assessment, and historical data.

[0050] Failure probability P f It can be expressed as: ; S43: Construct an outer robustness optimization loop. Select an appropriate optimization method and, based on the data obtained from the inner layer, optimize the turbine blade geometry parameters by maximizing the mean and minimizing the variance of the turbine performance parameters. At the same time, set the constraint as the failure probability of the turbine blades and perform a cyclic optimization of the turbine blade geometry parameters. W and turbine efficiency The mathematical description of robust multi-objective optimization can be expressed as:

[0051] , ; Where, is the defined failure probability threshold.

[0052] A heuristic non-dominated sorting genetic algorithm is used for robust reliability optimization. It divides individuals in the population into different non-dominated levels and calculates the crowding distance of individuals in each non-dominated level. d i , and select individuals to form a new parent population based on the current non-dominated level and crowding distance, and finally obtain the Pareto optimal solution set through continuous iteration until the termination condition is met. It has a strong global search capability, can effectively explore the entire design space, and is naturally suitable for multi-condition parallel computing, and has significant advantages in the robust optimization problem of gas turbine blade reliability. d i The expression is as follows:

[0053] ; Where, and For individuals iIn the objective function f k The function value of the adjacent individuals, and They are respectively all individuals in the non-dominated level in the objective function f k The maximum and minimum values ​​on .

[0054] The relevant parameters in the optimization algorithm should be selected based on the optimization efficiency and actual computing resources. In this embodiment, considering the optimization efficiency and computing cost, the population size is set to 200, the maximum number of iterations is set to 100, the crossover probability is set to 0.9, the mutation probability is set to 0.1, and the crossover distribution index and mutation distribution index are both set to 20.

[0055] S44: A double-layer loop framework is used to optimize the performance of the gas turbine. When the optimal solution set does not change or the maximum number of iterations is reached, the optimization is stopped and the final optimal solution set is output.

[0056] The method for optimizing the geometric parameters of turbine blades of this embodiment has at least the following beneficial effects: 1. By constructing an adaptive learning function, it proactively acquires the relevant parameters and fidelity of newly added sample points and automatically updates the proxy model until the accuracy requirements are met. The training process is highly versatile and requires no manual intervention, achieving efficient utilization of training samples and significantly reducing the time and computational costs required to build the dataset. 2. The multi-fidelity hybrid network includes a low-fidelity prediction network and a correction network. It fully utilizes the complementary advantages of high-fidelity model data and low-fidelity model data to achieve multi-source data fusion. It can provide a new solution to the accuracy-efficiency trade-off problem in the process of high-precision and rapid prediction of gas turbine turbine performance. At the same time, the correction network takes into account the physical field information of the model, which is more accurate and more interpretable than traditional proxy models. 3. A two-layer loop architecture is used for robust reliability optimization. The inner layer uses a trained multi-fidelity hybrid network to quickly obtain responses under different operating conditions and calculate the reliability and robustness parameters of the turbine blades. The outer layer performs robust reliability optimization of the blades in the entire design space based on the responses of the inner layer. This can obtain a design solution that combines aerodynamic robustness and strength reliability, ensuring the efficiency and reliability of turbine operation.

[0057] 4. Using (optimization method) to conduct multi-objective optimization for multiple optimization variables of reliability optimization problem, maintaining the orthogonality of objective function, the Pareto optimal solution set can be fully obtained, providing decision makers with a multi-dimensional optimization selection space with fast calculation speed, low complexity and good convergence.

[0058] The steps of the various methods above are divided only for the purpose of clear description. When implemented, they can be combined into one step or some steps can be split and decomposed into multiple steps. As long as they include the same logical relationship, they are within the scope of protection of the present invention. Adding insignificant modifications or introducing insignificant designs to the algorithm or process without changing the core design of the algorithm and process are all within the scope of protection of the invention.

[0059] Another embodiment of the present invention relates to a system for optimizing the geometric parameters of turbine blades. The implementation details of the system for optimizing the geometric parameters of turbine blades of this embodiment are described in detail below. The following content is provided for ease of understanding only and is not essential for implementing this solution. The system for optimizing the geometric parameters of turbine blades of this embodiment includes: an input sample acquisition module, configured to model the turbine blade according to geometric parameters, operating parameters, and material parameters of the turbine blade, and obtain a plurality of first input samples having a first fidelity and a plurality of second input samples having a second fidelity; wherein the first fidelity is less than the second fidelity; a response sample acquisition module, configured to acquire the performance parameters of the turbine blade corresponding to each first input sample and each second input sample, as a first response sample and a second response sample; a network training module, configured to train a gas turbine performance prediction network comprising a first fidelity prediction network and a correction network using a plurality of first input samples, a plurality of second input samples, a plurality of first response samples, and a plurality of second response samples; wherein during training, the first fidelity prediction network is trained using the first input samples and the first response samples so as to predict first response samples corresponding to the first input samples by the first fidelity prediction network, and the correction network is trained using the first response samples predicted by the first fidelity prediction network and the second input samples so as to predict second response samples corresponding to the first input samples by the correction network; The parameter optimization module is used to predict the performance parameters of the target turbine blades based on the geometric parameters, operating parameters and material parameters of the target turbine blades through the trained gas turbine turbine performance prediction network, and optimize the geometric parameters of the target turbine blades with the goal of maximizing the reliability and robustness of the target turbine blades indicated by the predicted performance parameters.

[0060] It is not difficult to find that this embodiment is a system embodiment corresponding to the above-mentioned method embodiment, and this embodiment can be implemented in conjunction with the above-mentioned method embodiment. The relevant technical details and technical effects mentioned in the above-mentioned embodiment are still valid in this embodiment, and to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied to the above-mentioned embodiment.

[0061] It is worth noting that all modules involved in this embodiment are logical modules. In actual applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, to highlight the innovations of the present invention, this embodiment does not include units that are not closely related to solving the technical problems proposed by the present invention. However, this does not mean that other units do not exist in this embodiment.

[0062] Another embodiment of the present invention relates to a computer device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method for optimizing the geometric parameters of turbine blades in the above-mentioned embodiments.

[0063] The memory and processor are connected using a bus, which can include any number of interconnected buses and bridges. The bus connects various circuits of one or more processors and memories. The bus can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits. These are all well known in the art and are therefore not described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single component or multiple components, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over a wireless medium via an antenna. Furthermore, the antenna receives data and transmits it to the processor.

[0064] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory can be used to store data used by the processor when performing operations.

[0065] Another embodiment of the present invention relates to a computer-readable storage medium storing a computer program, which implements the above method embodiment when executed by a processor.

[0066] That is, those skilled in the art will understand that all or part of the steps in the above-described method embodiments can be implemented by instructing the relevant hardware through a program. The program is stored in a storage medium and includes a number of instructions for causing a device (such as a microcontroller or chip) or a processor to execute all or part of the steps in the method embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0067] Those skilled in the art will appreciate that the above embodiments are specific embodiments for implementing the present invention, and that in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present invention.

Claims

1. A method for optimizing geometric parameters of turbine blades, characterized in that: The method comprises: Modeling the turbine blade according to geometric parameters, operating parameters, and material parameters of the turbine blade to obtain a plurality of first input samples having a first fidelity and a plurality of second input samples having a second fidelity; wherein the first fidelity is less than the second fidelity; Obtaining performance parameters of the turbine blade corresponding to each first input sample and each second input sample, respectively, as a first response sample and a second response sample; A gas turbine performance prediction network comprising a first fidelity prediction network and a correction network is trained using a plurality of first input samples, a plurality of second input samples, a plurality of first response samples, and a plurality of second response samples; wherein, during training, the first fidelity prediction network is trained using the first input samples and the first response samples so that the first fidelity prediction network predicts first response samples corresponding to the first input samples, and the correction network is trained using the first response samples predicted by the first fidelity prediction network and the second input samples so that the correction network predicts second response samples corresponding to the first input samples; The trained gas turbine performance prediction network predicts the performance parameters of the target turbine blades based on the geometric parameters, operating parameters and material parameters of the target turbine blades, and optimizes the geometric parameters of the target turbine blades with the goal of maximizing the reliability and robustness of the target turbine blades indicated by the predicted performance parameters.

2. The method for optimizing geometric parameters of turbine blades according to claim 1, wherein: The gas turbine performance prediction network is trained using an active learning sampling method. The active learning sampling method aims to minimize the failure probability estimation error of the target turbine blade indicated by the second response sample predicted by the gas turbine performance prediction network, and determines the location of the newly added sampling points and the fidelity of the newly added sampling points. The performance parameters of the turbine blade are used to indicate the failure probability of the turbine blade.

3. The method for optimizing geometric parameters of turbine blades according to claim 2, wherein: The active learning sampling method determines the location of the newly added sampling points and the fidelity of the newly added sampling points by the following formula: ； ； ； ； Where, x is the input vector, t represents the fidelity of the gas turbine performance prediction network, and represent the first fidelity and the second fidelity respectively, is the adaptive learning function, It represents the probability that the sample point is accurately judged by the gas turbine performance prediction network. c ( t ) is a cost function, which represents the acquisition cost ratio of the first input sample and the second input sample, is the cross-correlation function, is the probability density function, is the distance function, is the mean of the predicted values, is the variance of the predicted value, x new To add new sample points and t new is the corresponding fidelity.

4. The method for optimizing geometric parameters of turbine blades according to claim 3, wherein: The method includes: predicting the performance parameters of the target turbine blades based on the geometric parameters, operating parameters, and material parameters of the target turbine blades using the trained gas turbine performance prediction network; and optimizing the geometric parameters of the target turbine blades with the goal of maximizing the reliability and robustness of the target turbine blades indicated by the predicted performance parameters. The trained gas turbine performance prediction network predicts the performance parameters of the target turbine blade under different operating conditions based on the geometric parameters, material parameters, and different operating condition parameters of the target turbine blade. The performance parameters include the power, efficiency, stress, and strain of the target turbine blade. Obtaining the failure probability of the target turbine blade according to the stress and strain of the target turbine blade; With the goal of maximizing the mean and minimizing the variance of the power and efficiency of the target turbine blade under different operating conditions and the failure probability of the target turbine blade as a constraint, the geometric parameters of the target turbine blade are optimized.

5. The method for optimizing geometric parameters of turbine blades according to any one of claims 1 to 4, characterized in that: The geometric parameters include the geometric parameters of the root area and the bottom fillet area of ​​the turbine blade; the operating parameters include the total gas inlet temperature, total gas inlet pressure, gas outlet flow, total cooling air inlet temperature, total cooling air inlet pressure and rotational speed of the gas turbine; the material parameters include the density, elastic modulus, Poisson's ratio, thermal conductivity and thermal conductivity of the turbine blade.

6. A turbine blade geometric parameter optimization system, characterized in that: The system comprises: an input sample acquisition module, configured to model the turbine blade according to geometric parameters, operating parameters, and material parameters of the turbine blade, and obtain a plurality of first input samples having a first fidelity and a plurality of second input samples having a second fidelity; wherein the first fidelity is less than the second fidelity; a response sample acquisition module, configured to acquire the performance parameters of the turbine blade corresponding to each first input sample and each second input sample, as a first response sample and a second response sample; a network training module, configured to train a gas turbine performance prediction network comprising a first fidelity prediction network and a correction network using a plurality of first input samples, a plurality of second input samples, a plurality of first response samples, and a plurality of second response samples; wherein during training, the first fidelity prediction network is trained using the first input samples and the first response samples so as to predict first response samples corresponding to the first input samples by the first fidelity prediction network, and the correction network is trained using the first response samples predicted by the first fidelity prediction network and the second input samples so as to predict second response samples corresponding to the first input samples by the correction network; The parameter optimization module is used to predict the performance parameters of the target turbine blades based on the geometric parameters, operating parameters and material parameters of the target turbine blades through the trained gas turbine turbine performance prediction network, and optimize the geometric parameters of the target turbine blades with the goal of maximizing the reliability and robustness of the target turbine blades indicated by the predicted performance parameters.

7. A computer device, characterized in that: include: at least one processor; And, a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method for optimizing the geometric parameters of a turbine blade as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for optimizing geometric parameters of a turbine blade according to any one of claims 1 to 5 is implemented.

Images