CN-122020847-A - Turbine casing fatigue reliability optimization design method, device, equipment and medium

CN122020847ACN 122020847 ACN122020847 ACN 122020847ACN-122020847-A

Abstract

The application discloses a turbine casing fatigue reliability optimization design method, device, equipment and medium, which comprise the steps of S1, constructing a turbine casing axisymmetric parameterized model based on parameterized design thought, S2, applying a temperature field and a circulating load to a turbine casing service environment, performing thermosetting coupling analysis, calculating and obtaining dangerous point stress strain data, S3, performing fatigue reliability analysis by adopting an AK-MCS method, replacing an implicit function by adopting a self-adaptive proxy model in the fatigue reliability analysis process, and S4, performing optimization by adopting a nested SQP method based on the self-adaptive proxy model with the aim of meeting the maximum fatigue life average value under reliability constraint, and outputting optimal design parameters of the turbine casing. The method not only improves the efficiency of the reliability analysis method of the turbine casing, but also greatly improves the reliability analysis precision of the turbine casing, and realizes the high-efficiency and high-precision reliability optimization design of the turbine casing.

Inventors

FENG YAJUAN
DENG YITAI
LI JIAN
WEN ZHIXUN
LI MENG
WEI WEI
LU YU
CHEN RUNTUO

Assignees

中国航发湖南动力机械研究所

Dates

Publication Date: 20260512
Application Date: 20260109

Claims (10)

1. The turbine casing fatigue reliability optimization design method is characterized by comprising the following steps: s1, constructing a turbine casing axisymmetric parameterized model based on a parameterized design idea; s2, aiming at the service environment of the turbine casing, applying a temperature field and a circulating load, carrying out thermosetting coupling analysis, and calculating and obtaining dangerous point stress strain data; s3, carrying out fatigue reliability analysis by adopting an AK-MCS method, and replacing an implicit function by using a self-adaptive proxy model in the fatigue reliability analysis process; S4, optimizing by using a nested SQP method based on the self-adaptive agent model with the aim of meeting the maximum fatigue life average value under the reliability constraint, and outputting the optimal design parameters of the turbine casing.
2. The method for optimizing the fatigue reliability of a turbine casing according to claim 1, wherein in step S1, when the turbine casing axisymmetric parameterized model is constructed, the model retains all core features of the original high-pressure turbine casing including, for example, reinforcing ribs, mounting edges, hanging rings and mounting holes, and omits non-critical structures including supports to simplify the calculation.
3. The turbine casing fatigue reliability optimization design method according to claim 1, wherein the step S3 specifically includes the steps of: s31, defining a function, namely setting fatigue failure critical damage D CR =1, and defining a cumulative damage D= Σ (n i /N fi ) based on a Miner linear cumulative damage theory: ; Wherein N i is the number of cycles under multiple loads, N fi is the fatigue life under the load corresponding to multiple loads, N represents the number of cycles under a single load, and N f represents the fatigue life corresponding to multiple loads; s32, determining a random input variable, namely selecting a model size variable which has obvious influence on fatigue life as the random input variable, and determining a distribution form of the model size variable, wherein the distribution form comprises normal distribution, uniform distribution and exponential distribution; S33, carrying out fatigue reliability analysis by adopting an AK-MCS method, and calculating failure probability and variation coefficient based on a Kriging agent model and self-adaptive iterative update during the fatigue reliability analysis.
4. The turbine casing fatigue reliability optimization design method according to claim 3, wherein the fatigue life N f under multiple loads is calculated by a molo model: ; In the formula, As the average strain amplitude of the hazard point, E is the elastic modulus of the material and is the stress average value of the dangerous point, Is the fatigue strength coefficient), For a fatigue ductility factor of b to be the fatigue strength index and c to be the fatigue ductility index, the above parameter values can be obtained from GH4169 material fatigue test data by linear heteroscedastic regression.
5. The turbine casing fatigue reliability optimization design method according to claim 3 or 4, wherein step S33 specifically includes the steps of: S331, generating a sample pool, namely generating a Monte Carlo sample pool S by using Latin hypercube sampling based on random variable joint distribution; S332, constructing an initial training set, namely randomly selecting n 0 samples from the S, calculating the fatigue life through thermosetting coupling analysis and a Morro model, and obtaining a function value g (x) to form an initial training set T; S333, constructing a Kriging agent model, namely selecting a 0-order regression polynomial and a Gaussian correlation function, determining a correlation parameter theta through maximum likelihood estimation, constructing a Kriging agent model g K (x), and outputting a prediction mean mu g (x) and a prediction standard deviation sigma g (x); S334, self-adaptive iterative updating, namely calculating U learning function values (U= |mu g (x)/σ g (x) |) of all samples in the sample pool S, selecting a sample x new with the smallest U, calculating a real function value g (x new ) of the sample, adding a training set T, updating a Kriging proxy model, and repeating the process until the minU (x) is more than or equal to 2; S335, calculating failure probability, namely judging the failure state of each sample in the sample pool S through a converged Kriging proxy model, wherein g K (x) is less than or equal to 0 and is failure, and calculating the failure probability Coefficient of variation , Is the number of failed samples.
6. The turbine casing fatigue reliability optimization design method according to claim 1, wherein the step S4 specifically includes the steps of: S41, constructing an optimization model, wherein the optimization model comprises design variables, an objective function and constraint conditions, the design variables are that the average value of the required optimization variables is selected as the design variables, a value range and a variation coefficient are set, the objective function is that the maximum fatigue life average value E (N f ) is calculated by a Kriging agent model, the constraint conditions define reliability constraint and design variable boundary constraint, and the mathematical expression of the optimization model is as follows: ; Wherein, the Representing design variables of the reliability optimization design, Representing random input dimensional variables of a model, which are related to design variables In relation to the use of a liquid crystal display device, The random input environment variables representing the model, Representing the variables by design The random input variable that is generated is selected, Representing fatigue life value, objective function, of turbine casing model based on Moro model Expressed as a mean fatigue life, which is a function of design variables The function of the change is that, Functional function representing fatigue failure of turbine casing, judging whether sample point of input variable fails or not by setting fatigue life threshold as target cycle number A, namely ; Representing fatigue failure probability of a case model, which is a design variable The relative amounts of the variables that are involved, A threshold value representing fatigue reliability constraints of the receiver model, And Representing design variables respectively Upper and lower bound constraints of (2); S42, solving and outputting optimal design parameters by adopting a proxy model, a SQP sub-problem solving method and a nested SQP optimizing method based on a self-adaptive proxy model.
7. The turbine casing fatigue reliability optimization design method according to claim 6, wherein the step S42 specifically includes the steps of: S421, constructing a proxy model, namely respectively constructing a fatigue life proxy model N f K (x) and a failure boundary proxy model g K (x) in a design variable expansion space; SQP sub-problem solving, namely approximating the optimization problem as a quadratic programming sub-problem at the iteration point mu X k of the current design variable: ; wherein H k is a Hessian matrix, C k is a gradient vector, the gradient vector is obtained by N f K (x) through finite difference, A k is a constraint gradient matrix, and B k is a constraint value, and the failure probability is obtained by calculating by gK (x) and AK-MCS; S423, adaptively updating the proxy model, namely generating a new sample pool at each iteration point mu X k , updating gK (X) and N f K (X) through a U/H learning function, and ensuring the accuracy of the proxy model; S424, step size searching and convergence judging, namely carrying out one-dimensional searching along the QP sub-problem optimal solution direction S k , determining the step size alpha k , updating the design variable mu X (k+1) =μX k +α k S k , and converging when mu X (k+1) -μX k ||<10⁻ 4 is detected, and outputting the optimal design parameters of the turbine casing.
8. Turbine casing fatigue reliability optimal design device, its characterized in that includes: The model construction module is used for constructing an axisymmetric parameterized model of the turbine casing based on the parameterized design idea; the stress-strain data calculation module is used for applying a temperature field and a cyclic load to the service environment of the turbine casing, performing thermosetting coupling analysis, and calculating and acquiring dangerous point stress-strain data; the reliability analysis module is used for carrying out fatigue reliability analysis by adopting an AK-MCS method, and replacing an implicit function by the self-adaptive agent model in the fatigue reliability analysis process; And the design parameter optimization module is used for carrying out optimization by adopting a nested SQP method based on the self-adaptive agent model with the aim of maximizing the fatigue life mean value under the reliability constraint and outputting the optimal design parameters of the turbine casing.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the turbine casing fatigue reliability optimization design method according to any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the turbine casing fatigue reliability optimization design method according to any one of claims 1 to 7.

Description

Turbine casing fatigue reliability optimization design method, device, equipment and medium Technical Field The application relates to the technical field of aeroengine structural design and reliability engineering, in particular to a turbine casing fatigue reliability optimization design method, device, equipment and medium. Background The turbine casing is a core bearing part of the aeroengine, bears the functions of supporting a rotor, forming an airflow channel and containing high-temperature fuel gas, and the service environment is subjected to high temperature (600-650 ℃), complex cyclic load (internal and external pressure difference and thermal stress) and multisource uncertainty (material fluctuation, dimensional error and load fluctuation). More than 70% of aero-engine case accidents are caused by fatigue failure, for example, the failure rate of weld cracks of a case mounting seat after a certain type of engine reaches 3%, the maximum crack length is 100mm, and the flight safety is seriously threatened. The existing reliability calculation method comprises traditional design, monte Carlo simulation, FORM/AFOSM and other simplified methods, AK-MCS method and the like, however, the existing turbine casing fatigue reliability design method has the following defects: 1) The traditional design adopts a deterministic method, and based on allowable stress and experience safety coefficient, parameter uncertainty is not considered, and over-design or under-design is easily caused; 2) In the existing reliability analysis method, the direct Monte Carlo Simulation (MCS) requires massive finite element calculation (single-case thermosetting coupling analysis for more than 10 hours), and the efficiency is extremely low; 3) The simplified methods such as FORM/AFOSM have poor adaptability to implicit function functions; 4) The AK-MCS method lacks application and adaptation optimization algorithm of the AK-MCS method aiming at the turbine casing, and is difficult to meet the requirements of high reliability and long service life. Disclosure of Invention The application provides an optimized design method for fatigue reliability of a turbine casing, and solves the technical problems of low efficiency, low precision, poor adaptability and lack of pertinence in the prior art. The application is realized by the following scheme: the turbine casing fatigue reliability optimization design method comprises the following steps: s1, constructing a turbine casing axisymmetric parameterized model based on a parameterized design idea; s2, aiming at the service environment of the turbine casing, applying a temperature field and a circulating load, carrying out thermosetting coupling analysis, and calculating and obtaining dangerous point stress strain data; s3, carrying out fatigue reliability analysis by adopting an AK-MCS method, and replacing an implicit function by using a self-adaptive proxy model in the fatigue reliability analysis process; S4, optimizing by using a nested SQP method based on the self-adaptive agent model with the aim of meeting the maximum fatigue life average value under the reliability constraint, and outputting the optimal design parameters of the turbine casing. Further, in step S1, when the turbine casing axisymmetric parameterized model is constructed, the model retains all core features of the original high-pressure turbine casing, including, for example, reinforcing ribs, mounting edges, hanging rings, and mounting holes, and omits non-critical structures including standoffs to simplify the calculation. Further, the step S3 specifically includes the steps of: s31, defining a function, namely setting fatigue failure critical damage D CR =1, and defining a cumulative damage D= Σ (n i/Nfi) based on a Miner linear cumulative damage theory: ; Wherein N i is the number of cycles under multiple loads, N fi is the fatigue life under the load corresponding to multiple loads, N represents the number of cycles under a single load, and N f represents the fatigue life corresponding to multiple loads; s32, determining a random input variable, namely selecting a model size variable which has obvious influence on fatigue life as the random input variable, and determining a distribution form of the model size variable, wherein the distribution form comprises normal distribution, uniform distribution and exponential distribution; S33, carrying out fatigue reliability analysis by adopting an AK-MCS method, and calculating failure probability and variation coefficient based on a Kriging agent model and self-adaptive iterative update during the fatigue reliability analysis. Further, the fatigue life N f at the corresponding multiple loads is calculated by the molo model (taking into account the mean stress correction): ; In the formula, As the average strain amplitude of the hazard point,E is the elastic modulus of the material and is the stress average value of the dangerous point,Is the fatigue strength coefficie