Blade wheel disc fatigue reliability analysis method, system, device and medium

Abstract

The present application belongs to the technical field of gas turbine reliability analysis, and discloses a kind of blade disc fatigue reliability analysis method, system, equipment and medium;Among them, the method includes: based on each blade disc characteristic parameter determined by screening, and the probability distribution of each blade disc characteristic parameter, sample acquisition contains the reliability test sample pool of multiple samples;The obtained reliability test sample pool is input into the MP-ACBS-Kriging proxy model that is pre-learned and trained, and the obtained indicator function is output;Based on the obtained indicator function, the failure probability of blade disc and the coefficient of variation are calculated;If the coefficient of variation meets the requirements, the fatigue reliability analysis result is obtained, otherwise, the number of samples in the reliability test sample pool is increased and recalculated.The technical scheme of the present application greatly improves the efficiency and precision of gas turbine blade disc fatigue reliability analysis.

Description

Technical Field

[0001] This invention belongs to the field of gas turbine reliability analysis technology, and specifically relates to a method, system, equipment and medium for fatigue reliability analysis of blade disks. Background Technology

[0002] Gas turbines are the core equipment of clean power generation systems. Any unplanned maintenance or downtime can lead to huge additional costs and revenue losses. Therefore, current expectations for gas turbines are not only for greater power and lower fuel consumption, but also for longer life and higher reliability.

[0003] As a critical component of gas turbines, the bladed disk system often operates at high speeds under harsh pressure and temperature conditions, making it highly susceptible to fatigue failure. Furthermore, multiple uncertainties are prevalent in the design, manufacturing, and operation of gas turbine components, leading to a degree of randomness in their actual performance. Therefore, conducting reliability analysis on the bladed disk system to ensure its reliability and safety remain within specified limits is of great significance for gas turbine systems.

[0004] For gas turbine blade disk systems, existing fatigue reliability analysis methods are unsatisfactory in terms of both accuracy and computational efficiency, suffering from low accuracy, low computational efficiency, and high computational resource requirements. Therefore, establishing an efficient, accurate, and reliable fatigue reliability analysis method for gas turbine blade disks, and improving the efficiency and accuracy of fatigue reliability analysis, is an urgent problem to be solved. Summary of the Invention

[0005] The purpose of this invention is to provide a method, system, equipment, and medium for fatigue reliability analysis of gas turbine blade disks, in order to solve one or more of the aforementioned technical problems. The technical solution provided by this invention has wide applicability to complex gas turbine blade disk models, greatly improving the efficiency and accuracy of fatigue reliability analysis of gas turbine blade disks, and has advantages such as adaptability, high efficiency, reliability, and saving computational resources.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] The first aspect of this invention provides a method for fatigue reliability analysis of a bladed disk, comprising the following steps:

[0008] Step 1: Based on multiple blade disks of the same model to be analyzed for fatigue reliability, select and determine the characteristic parameters of the blade disks, and based on the measured data of the selected characteristic parameters of each blade disk, determine the probability distribution of the characteristic parameters of each blade disk.

[0009] Step 2: Based on the selected characteristic parameters of each blade disk and the probability distribution of each characteristic parameter, a reliability test sample pool containing multiple samples is obtained, and then the process proceeds to step 3; wherein, each sample in the reliability test sample pool includes the sample value of each characteristic parameter of each blade disk selected in step 1.

[0010] Step 3: Input the obtained reliability test sample pool into the pre-trained MP-ACBS-Kriging surrogate model and output the indicator function; based on the obtained indicator function, calculate the failure probability of the blade disk.

[0011] Step 4: Based on the failure probability of the blade disk obtained in Step 3, calculate the coefficient of variation; if the coefficient of variation is lower than the preset threshold, the failure probability of the blade disk obtained in Step 3 is the fatigue reliability analysis result of multiple blade disks of the same model; otherwise, increase the number of samples in the reliability test sample pool to obtain the updated reliability test sample pool, and jump to Step 3.

[0012] A further improvement of the present invention is that, in step 1, the step of screening and determining the characteristic parameters of the blade disk includes:

[0013] Based on the pre-acquired uncertainty parameters, a global sensitivity analysis strategy is adopted to measure the impact of each parameter on the response of the blade disk model, and uncertainty parameters that meet the preset impact requirements are selected as the final blade disk characteristic parameters.

[0014] The pre-acquired uncertainty parameters include at least working environment parameters, geometric parameters, material properties, and load parameters.

[0015] A further improvement of the present invention is that, in step 2, the step of sampling to obtain a reliability test sample pool containing multiple samples,

[0016] The sampling method used was Monte Carlo sampling.

[0017] A further improvement of the present invention is that the learning and training steps of the pre-trained MP-ACBS-Kriging proxy model include:

[0018] (1) Based on multiple blade disk samples, the characteristic parameters of the blade disk samples are screened and determined, and based on the measured data of the characteristic parameters of each blade disk sample screened and determined, the probability distribution of the characteristic parameters of each blade disk sample is determined; wherein, the model of the multiple blade disk samples in step (1) is the same as the model of the multiple blade disks in step 1, and the type of characteristic parameters of the blade disk samples screened and determined in step (1) is the same as the type of characteristic parameters of the blade disks screened and determined in step 1.

[0019] (2) Based on the characteristic parameters of each blade disk sample determined in step (1) and the probability distribution of the characteristic parameters of each blade disk sample, multiple initial sample points are sampled to obtain the corresponding true limit state function response set and a training sample set is constructed. Based on the training sample set, an initial Kriging surrogate model from random variables to limit state functions is constructed and step (3) is executed.

[0020] (3) Using the current Kriging model, predict the limit state function response and variance of each sample point in the candidate sample pool, and then proceed to step (4); wherein, the candidate sample pool is a set of candidate sample points obtained by sampling based on the characteristic parameters of each blade disk sample determined in step (1) and the probability distribution of the characteristic parameters of each blade disk sample.

[0021] (4) Based on the idea of ​​active learning, the optimal new training sample point x is selected through a constraint boundary active learning strategy. new1 and x new2 Calculate x new1 and x new2 The limit state equation response is obtained, and the first convergence criterion is tested; if the first convergence criterion is satisfied, then proceed to step (5); otherwise, proceed to step (3); wherein, the learning function of the constraint boundary active learning strategy is,

[0022]

[0023] In the formula, ACBS(·) is the learning function; sign(·) is the sign function; μ g (x) and σ g (x) represents the predicted value and standard deviation of the surrogate model for sample point x, respectively; φ(·) is the probability density function; D(·) is the normalized distance between samples;

[0024] x new1 =argmax(ACBS(x)); x new2 =argmin(ACBS(x));

[0025] The first convergence criterion determines convergence based on whether the current relative error meets the requirements; the expression is as follows:

[0026] argmax(e R (x new1 ),e R (x new2 ))<[e R ];

[0027] In the formula, e R For relative error; [e R [ ] represents the convergence threshold;

[0028] (5) Enrich the training sample set with the optimal training sample point x. new1 and x new2 Update the Kriging model and test the second convergence criterion; if the second convergence criterion is satisfied, the trained MP-ACBS-Kriging surrogate model is obtained; otherwise, proceed to step (3); where the second convergence criterion is determined by the fact that the number of samples in the failure domain in the candidate sample pool no longer changes significantly during the iteration process, and the expression is:

[0029]

[0030] In the formula, N kf The number of failure domain samples in the candidate sample pool; i is the iteration step; [e f [ ] represents the convergence threshold.

[0031] A further improvement of the present invention is that, in step (2), the step of obtaining the corresponding true limit state function response set includes:

[0032] Based on each set of initial sample points, a gas-thermal-structure coupled finite element analysis (GEA) of the gas turbine blade disk is conducted using a flow field order reduction method to identify key fatigue failure sites and main fatigue failure modes. The life of the gas turbine blade disk is then predicted based on a theoretical fatigue life prediction model. Based on the life prediction results and the limit state function, the corresponding set of actual limit state function responses is obtained.

[0033] The limiting state function is expressed as follows:

[0034] g(x) = N - N0;

[0035] In the formula, g(x) is the limit state function; x is the uncertainty parameter; N0 is the safe life; and N is the predicted life.

[0036] A further improvement of the present invention is that, in step 3, the step of calculating the failure probability of the blade disk based on the obtained indicator function,

[0037] The formula for calculating the failure probability of the blade disk is as follows:

[0038]

[0039] In the formula, denoted as , where m is the failure probability of the blade disk; m is the number of samples in the reliability test sample pool; m f This represents the number of failed samples. As an indicator function, when the limit state function g(x) i When )≤0 Select 1 if the value is 1, otherwise select 0.

[0040] A further improvement of the present invention is that, in step 4, the step of calculating the coefficient of variation based on the failure probability of the blade disk obtained in step 3,

[0041] The formula for calculating the coefficient of variation for CoV is as follows:

[0042]

[0043] A second aspect of the present invention provides a fatigue reliability analysis system for a bladed disk, comprising:

[0044] The data acquisition module is used to screen and determine the characteristic parameters of multiple blade disks of the same model that are to be analyzed for fatigue reliability, and to determine the probability distribution of the characteristic parameters of each blade disk based on the measured data of the screened characteristic parameters.

[0045] The reliability test sample pool acquisition module is used to sample and acquire a reliability test sample pool containing multiple samples based on the selected characteristic parameters of each blade disk and the probability distribution of each blade disk characteristic parameter, and then jump to execute the step of the blade disk failure probability acquisition module; wherein, each sample in the reliability test sample pool includes the sample value of each blade disk characteristic parameter selected in the data acquisition module.

[0046] The blade disk failure probability acquisition module is used to input the obtained reliability test sample pool into the pre-trained MP-ACBS-Kriging surrogate model and output an indicator function; based on the obtained indicator function, the failure probability of the blade disk is calculated.

[0047] The verification output module is used to calculate the coefficient of variation based on the failure probability of the blade disk obtained by the blade disk failure probability acquisition module. If the coefficient of variation is lower than a preset threshold, the failure probability of the blade disk obtained by the blade disk failure probability acquisition module is the fatigue reliability analysis result of multiple blade disks of the same model. Otherwise, the number of samples in the reliability test sample pool is increased to obtain an updated reliability test sample pool, and the steps of the blade disk failure probability acquisition module are executed.

[0048] A third aspect of the present invention provides an electronic device comprising:

[0049] At least one processor; and,

[0050] A memory communicatively connected to the at least one processor; wherein,

[0051] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the blade disk fatigue reliability analysis method as described in any one of the first aspects of the present invention.

[0052] The fourth aspect of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the blade disk fatigue reliability analysis method described in any one of the first aspects of the present invention.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] The blade disk fatigue reliability analysis method provided by this invention evaluates the limit state function based on the constructed Kriging surrogate model (MP-ACBS-Kriging surrogate model) based on constraint boundaries and multi-point enrichment strategy. It has wide applicability to complex gas turbine blade disk models, greatly improves the efficiency and accuracy of gas turbine blade disk fatigue reliability analysis, and has the advantages of being adaptive, efficient, reliable, and saving computing resources.

[0055] This invention considers the actual failure modes of gas turbine blade disks, selects key characteristic parameters, establishes a parameterized model of the gas turbine blade disk, conducts gas-thermal-structure coupled finite element analysis based on the flow field order reduction method, and builds a theoretical fatigue life prediction model for the gas turbine blade disk. Combining the active learning concept, it automatically selects the most valuable sample points for sequential sampling, constructs an MP-ACBS-Kriging surrogate model for evaluating the limit state function, and uses Monte Carlo sampling for reliability analysis calculation. This greatly improves the efficiency and accuracy of fatigue reliability analysis of gas turbine blade disks and has the advantages of being adaptive, fast, efficient, and reliable. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art are briefly introduced below; obviously, the drawings described below are some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without creative effort.

[0057] Figure 1 This is a flowchart illustrating a method for fatigue reliability analysis of a blade disk provided in an embodiment of the present invention;

[0058] Figure 2 This is a flowchart illustrating a fatigue reliability analysis method for blade disks based on constraint boundaries and multi-point enrichment algorithms provided in an embodiment of the present invention.

[0059] Figure 3 This is a schematic diagram of the MP-ACBS-Kriging proxy model construction process in an embodiment of the present invention. Detailed Implementation

[0060] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0061] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0062] The present invention will now be described in further detail with reference to the accompanying drawings:

[0063] Please see Figure 1 The present invention provides a method for fatigue reliability analysis of a bladed disk, comprising the following steps:

[0064] Step 1: Based on multiple blade disks of the same model to be analyzed for fatigue reliability, select and determine the characteristic parameters of the blade disks, and based on the measured data of the selected characteristic parameters of each blade disk, determine the probability distribution of the characteristic parameters of each blade disk.

[0065] Step 2: Based on the selected characteristic parameters of each blade disk and the probability distribution of each characteristic parameter, a reliability test sample pool containing multiple samples is obtained, and then the process proceeds to step 3; wherein, each sample in the reliability test sample pool includes the sample value of each characteristic parameter of each blade disk selected in step 1.

[0066] Step 3: Input the obtained reliability test sample pool into the pre-trained MP-ACBS-Kriging surrogate model and output the indicator function; based on the obtained indicator function, calculate the failure probability of the blade disk.

[0067] Step 4: Based on the failure probability of the blade disk obtained in Step 3, calculate the coefficient of variation; if the coefficient of variation is lower than the preset threshold, the failure probability of the blade disk obtained in Step 3 is the fatigue reliability analysis result of multiple blade disks of the same model; otherwise, increase the number of samples in the reliability test sample pool to obtain the updated reliability test sample pool, and jump to Step 3.

[0068] The technical solution provided by this invention evaluates the limit state function based on the constructed MP-ACBS-Kriging surrogate model. It has wide applicability to complex gas turbine blade disk models, greatly improving the efficiency and accuracy of fatigue reliability analysis of gas turbine blade disks. It has the advantages of being adaptive, efficient, reliable, and saving computing resources.

[0069] Please see Figure 2 The present invention provides a method for fatigue reliability analysis of bladed disks based on constraint boundaries and multi-point enrichment algorithms, comprising the following steps:

[0070] S1: Determine the main characteristic parameters and corresponding probability distributions of the gas turbine blade disk, and construct a parameterized model of the gas turbine blade disk; whereby...

[0071] Specifically, based on statistical analysis of sample measurement data, past experience, or engineering examples, uncertainty parameters and probability distribution models are selected. These uncertainty parameters include working environment parameters, geometric parameters, material properties, and load parameters. The probability distribution of these uncertainty parameters can be Weibull distribution, log-normal distribution, or normal distribution. Subsequently, global sensitivity analysis strategies such as the Sobol' method are used to measure the impact of each parameter on the model response, and the main uncertainty parameters are extracted to achieve feature dimensionality reduction of the model. Furthermore, the Sobol' sensitivity measurement is as follows:

[0072]

[0073] In the formula, The total Sobol' index; For the domain Ω n =[x|0≤x i ≤1; Sensitivity measure on i = 1, ..., n; D is the total variance; For partial variance; i1,…,i s The subscript is 1 ≤ i1 < ... s ≤n,s=1,…,n;

[0074] The key uncertainties, such as the selected geometric dimensions and material properties, are abstracted into numerical parameters to construct a flexible and controllable parameterized model of the gas turbine blade disk.

[0075] ​In this embodiment of the invention, the main uncertainties of the gas turbine blade disk are extracted by sensitivity analysis, and the model is parameterized and managed, which reduces the complexity of reliability analysis and improves the efficiency of reliability analysis.

[0076] S2: Based on the flow field order reduction method, a gas-thermal-structure coupled finite element analysis was conducted on the gas turbine blade disk to identify key fatigue failure sites and main fatigue failure modes, and a theoretical fatigue life prediction model for the gas turbine blade disk was constructed; among which...

[0077] Specifically, a high-fidelity simulation model of the solid and fluid domains of the gas turbine blade disk is established. In the fluid domain, a flow field reduction method (ROM) such as the intrinsic orthogonal decomposition (POD) is used to reduce the degrees of freedom of the flow field. First, the numerical parameters described in step S1 are sampled multiple times according to the probability distribution as input for experiments or numerical calculations, and a series of flow field snapshots are obtained. Then, the optimal ordered orthogonal basis of the system is obtained by performing basic mode decomposition on the flow field snapshots. The optimal orthogonal basis is truncated to obtain the reduced-order model of the flow field. Finally, the reduced-order model is solved and mapped to the solution of the original flow field through linear combination of basic modes.

[0078] Taking multi-parameter steady-state flow prediction as an example, the flow field approximation fitting form based on the POD method is as follows:

[0079]

[0080] In the formula, Φ(X,α) j () represents a high-dimensional system at the j-th, j = 1, 2, ..., N-th dimension. p Group parameter α j Positional spatial variables (such as pressure, temperature, velocity, etc.) under boundary conditions; A l (α j ) represents the modal coefficients corresponding to the l-th POD basis; φ l (X) is the POD basis vector, and L is the number of POD bases.

[0081] Subsequently, the flow field information was used as the boundary condition input for the finite element method (FEM) calculation of the solid domain of the gas turbine blade disk. Gas-thermal-structure interaction analysis was conducted to obtain the temperature field, stress-strain distribution, and other parameters of the blade disk. Based on the temperature and strength distribution characteristics of the blade disk and the actual operating conditions of the gas turbine, key fatigue failure sites and main failure modes were identified. The life prediction of the gas turbine blade disk was then achieved based on a theoretical fatigue life prediction model. This theoretical life prediction model includes methods such as the nominal stress method and the local stress-strain method.

[0082] Taking the low-cycle fatigue failure mode of a gas turbine blade disk as an example, the local stress-strain method uses the stress and strain of the local structure instead of the overall stress and strain of the structure for life calculation, making it more suitable for low-cycle fatigue life prediction. The SWT life prediction model can be used, as detailed below:

[0083]

[0084] In the formula, ε a σ is the cyclic strain amplitude; max The maximum stress is σ'. f ε' is the fatigue strength coefficient; b is the fatigue strength index; ε' f is the fatigue ductility coefficient; c is the fatigue ductility index; E is the elastic modulus; N f This refers to fatigue life.

[0085] In this embodiment of the invention, when predicting the fatigue life of a gas turbine blade disk based on the gas-thermal-solid coupling finite element method, a flow field order reduction method is used to reconstruct the physical field. While accurately predicting the flow characteristics, the computational degrees of freedom are significantly reduced, thereby saving the huge computational resources required for repeated solutions of the flow field under multiple operating conditions and further improving computational efficiency.

[0086] S3: Generate an initial training sample set (DoE). Based on constraint boundaries and a multi-point enrichment active learning algorithm, gradually enrich samples into the DoE during the surrogate model construction process to construct an MP-ACBS-Kriging surrogate model for gas turbine blade disk reliability analysis; whereby...

[0087] Specifically, based on the gas turbine blade disk life prediction framework obtained in step S2 using gas-thermal-structure coupled finite element analysis, a limit state function (LSF) is established, which is as follows:

[0088] g(x) = N - N0

[0089] In the formula, g(x) is the limit state function; x is the uncertainty parameter; N0 is the safe life; and N is the predicted life.

[0090] Please see Figure 3 The construction process of the MP-ACBS-Kriging surrogate model for reliability analysis of gas turbine blade disks is as follows:

[0091] ① Using sampling methods such as Latin hypercube (LHS), N samples are drawn according to the probability distribution of the numerical parameters described in step S1. kAn initial set of sample points is generated, followed by gas-thermal-structure coupling analysis, lifetime prediction, and limit state function evaluation to obtain the corresponding true limit state function response set. This is then used to establish a training sample set (DoE), thereby constructing an initial Kriging surrogate model from random variables to the limit state function. The LHS method is used to generate a set containing N... c A candidate sample pool of samples is used for enrichment of experimental sample points.

[0092] ②Predict the limit state function response and variance of each sample in the LHS candidate pool using the current Kriging model.

[0093] ③ Based on the concept of active learning, the optimal new training sample point x is selected through the Constrained Boundary Active Learning Strategy (ACBS). new1 and x new2 Calculate x new1 and x new2 The limit state equation response is obtained, and the first convergence criterion is tested. The constrained boundary algorithm can sequentially locate sampling points near the limit state surface, improving the accuracy of the surrogate model, while ensuring that the added sample points are as sparse and non-clustered as possible and all fall in the high-probability region. Its learning function is:

[0094]

[0095] In the formula, ACBS(·) is the learning function; sign(·) is the sign function; μ g (x) and σ g (x) represents the predicted value and standard deviation of the surrogate model for sample point x, respectively; φ(·) is the probability density function; and D(·) is the normalized distance between samples.

[0096] Based on the multi-point enrichment (MP) principle, points that contribute the most to the accuracy of the surrogate model in both the safe and failure domains are selected as new training sample points. This ensures a balance of training sample points in both the safe and failure domains, improving the computational efficiency of the algorithm while increasing the accuracy of fitting the limit state surface. The mathematical expression is as follows:

[0097] x new1 =argmax(ACBS(x));

[0098] x new2 =argmin(ACBS(x));

[0099] The first convergence criterion determines whether convergence has been achieved by checking whether the current relative error meets the requirements, and its formula is as follows:

[0100] argmax(e R (x new1 ),e R (xnew2 ))<[e R ];

[0101] In the formula, e R For relative error; [e R [ ] represents the convergence threshold.

[0102] ④ Enrich the optimal training sample points x in the DoE. new1 and x new2 Update the current Kriging model and test the second convergence criterion. The second convergence criterion determines the convergence of the algorithm based on whether the number of samples in the failure domain in the candidate sample pool no longer changes significantly during the iteration process. The formula is as follows:

[0103]

[0104] In the formula, N kf The number of failure domain samples in the DoE; i is the number of iteration steps; [e f [ ] represents the convergence threshold.

[0105] ⑤ Check if the active learning termination condition has been met, i.e., the first and second convergence criteria are satisfied simultaneously. If so, terminate the active learning algorithm; otherwise, return to step ②.

[0106] This completes the construction of the MP-ACBS-Kriging proxy model that meets the accuracy requirements.

[0107] This invention proposes an active learning strategy for constraint boundaries and a multi-point enrichment and addition principle, thereby constructing an MP-ACBS-Kriging surrogate model. The most valuable candidate sample points are automatically selected along both sides of the limit state surface for adaptive updating of the surrogate model, which improves the accuracy of the surrogate model used for reliability analysis. At the same time, it reduces the number of evaluations of the original limit state function, further reducing the required computational resources and improving computational efficiency.

[0108] S4: Fatigue reliability analysis of gas turbine blade disks is performed using the MP-ACBS-Kriging surrogate model; among which...

[0109] Specifically, based on the probability distribution of the numerical parameters described in step S1, a reliability test sample pool containing m samples is generated through Monte Carlo sampling (MCS).

[0110] The MP-ACBS-Kriging surrogate model obtained in step S3 is used to replace experiments or the finite element method to evaluate the limit state function and calculate the failure probability of the gas turbine blade disk. And the coefficient of variation for CoV, the formula is as follows:

[0111]

[0112]

[0113] In the formula, m f This represents the number of failed samples. For indicator functions; when g(x) i When )≤0 Select 1 if the value is 1, otherwise select 0.

[0114] Check if the current coefficient of variation is lower than the threshold of 0.05. If it is, the reliability assessment ends and the calculation result is output. Otherwise, estimate the number of samples required for the reliability test sample pool based on the current failure probability and coefficient of variation, update the MCS reliability test sample pool, and repeat step S4.

[0115] The method of this invention uses the MP-ACBS-Kriging surrogate model to replace experiments or the finite element method for evaluating the limit state function when performing reliability calculations. This overcomes the high computational cost of Monte Carlo sampling while ensuring computational accuracy, further reducing computational costs and improving the efficiency of reliability analysis.

[0116] The following are embodiments of the apparatus of the present invention, which can be used to execute embodiments of the method of the present invention. For details not disclosed in the apparatus embodiments, please refer to the embodiments of the method of the present invention.

[0117] In another embodiment of the present invention, a fatigue reliability analysis system for bladed disks is provided, comprising:

[0118] The data acquisition module is used to screen and determine the characteristic parameters of multiple blade disks of the same model that are to be analyzed for fatigue reliability, and to determine the probability distribution of the characteristic parameters of each blade disk based on the measured data of the screened characteristic parameters.

[0119] The reliability test sample pool acquisition module is used to sample and acquire a reliability test sample pool containing multiple samples based on the selected characteristic parameters of each blade disk and the probability distribution of each blade disk characteristic parameter, and then jump to execute the step of the blade disk failure probability acquisition module; wherein, each sample in the reliability test sample pool includes the sample value of each blade disk characteristic parameter selected in the data acquisition module.

[0120] The blade disk failure probability acquisition module is used to input the obtained reliability test sample pool into the pre-trained MP-ACBS-Kriging surrogate model and output an indicator function; based on the obtained indicator function, the failure probability of the blade disk is calculated.

[0121] The verification output module is used to calculate the coefficient of variation based on the failure probability of the blade disk obtained by the blade disk failure probability acquisition module. If the coefficient of variation is lower than a preset threshold, the failure probability of the blade disk obtained by the blade disk failure probability acquisition module is the fatigue reliability analysis result of multiple blade disks of the same model. Otherwise, the number of samples in the reliability test sample pool is increased to obtain an updated reliability test sample pool, and the steps of the blade disk failure probability acquisition module are executed.

[0122] In another embodiment of the present invention, a computer device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions from the computer storage medium to achieve a corresponding method flow or corresponding function. The processor described in this embodiment of the present invention can be used for the operation of a blade disk fatigue reliability analysis method.

[0123] In another embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory). This computer-readable storage medium is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the blade disk fatigue reliability analysis method in the above embodiments.

[0124] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0125] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0126] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0127] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method of bladed disc fatigue reliability analysis, characterized in that, The method comprises the following steps: Step 1, based on a plurality of blade discs of the same type for fatigue reliability analysis, screening and determining blade disc characteristic parameters, and based on the measured data of the screening and determining blade disc characteristic parameters, determining the probability distribution of each blade disc characteristic parameter; Step 2, based on the screening and determining of each blade disc characteristic parameter and the probability distribution of each blade disc characteristic parameter, sampling to obtain a reliability test sample pool containing a plurality of samples, and jumping to Step 3; wherein each sample in the reliability test sample pool comprises a sample value of each blade disc characteristic parameter screened and determined in Step 1; Step 3, inputting the obtained reliability test sample pool into the pre-learned MP-ACBS-Kriging surrogate model to output an obtained indicator function; based on the obtained indicator function, calculating an obtained blade disc failure probability; Step 4, based on the blade disc failure probability obtained in Step 3, calculating an obtained coefficient of variation; if the coefficient of variation is lower than a preset threshold, the blade disc failure probability obtained in Step 3 is the fatigue reliability analysis result of the plurality of blade discs of the same type, otherwise, increasing the sample quantity of the reliability test sample pool, obtaining an updated reliability test sample pool, and jumping to Step 3; The learning and training steps of the pre-learned MP-ACBS-Kriging surrogate model comprise: (1) based on a plurality of blade disc samples, screening and determining blade disc sample characteristic parameters, and based on the measured data of the screening and determining blade disc sample characteristic parameters, determining the probability distribution of each blade disc sample characteristic parameter; wherein the type of the plurality of blade disc samples in Step (1) is the same as the type of the plurality of blade discs in Step 1, and the type of the blade disc sample characteristic parameters screened and determined in Step (1) is the same as the type of the blade disc characteristic parameters screened and determined in Step 1; (2) based on the screening and determining of each blade disc sample characteristic parameter in Step (1) and the probability distribution of each blade disc sample characteristic parameter, sampling to obtain a plurality of groups of initial sample points, obtaining a corresponding true limit state function response set, and constructing an obtained training sample set; based on the training sample set, constructing an initial Kriging surrogate model of the random variable to the limit state function, and jumping to Step (3); (3) predicting the limit state function response and variance of each sample point in the candidate sample pool through the current Kriging model, and jumping to Step (4); wherein the candidate sample pool is a plurality of groups of candidate sample points sampled based on the screening and determining of each blade disc sample characteristic parameter in Step (1) and the probability distribution of each blade disc sample characteristic parameter; (4) based on the idea of active learning, the best new training sample points are selected through the constraint boundary active learning strategy x new1 and x new2 , the limit state equation response of x new1 and x new2 is calculated, and the first convergence criterion is checked; if the first convergence criterion is met, step (5) is executed, otherwise step (3) is executed; wherein the learning function of the constraint boundary active learning strategy is ； wherein is a learning function; sign(·) is a sign function; The first convergence criterion is whether the current relative error meets the requirement, and the expression is g ( x ) and In Step 1, the step of screening and determining blade disc characteristic parameters comprises: g ( x ) are the predicted value and the predicted standard deviation of the proxy model for the sample point x , respectively; the function (·) is a probability density function; D (·) is a sample-wise normalized distance; ； ； Based on the pre-obtained uncertainty parameters, using a global sensitivity analysis strategy to measure the influence of each parameter on the response of the blade disc model, and screening and determining the uncertainty parameters meeting the preset influence requirement as the final blade disc characteristic parameters; ； wherein is the relative error; is the convergence threshold; (5) enriching the best training sample points in the training sample set x new1 and x new2 updating the Kriging model, and checking a second convergence criterion; if the second convergence criterion is satisfied, a learning trained MP-ACBS-Kriging surrogate model is obtained, otherwise, jumping to step (3); wherein the second convergence criterion is to judge the convergence of the algorithm according to whether the number of samples in the failure domain in the candidate sample pool no longer changes obviously in the iteration process, and the expression is ； In the formula, is the number of invalid domain samples in the candidate sample pool; i is the number of iteration steps; is the convergence threshold.

2. The method of blade disk fatigue reliability analysis according to claim 1, wherein, ​ ​ The pre-acquired uncertainty parameters at least include working environment parameters, geometric parameters, material properties and load parameters.

3. The method of claim 1, wherein, In step 2, the step of sampling to obtain a reliability verification sample pool containing a plurality of samples, The selected sampling method is Monte Carlo sampling.

4. The method of claim 1, wherein, In step (2), the step of obtaining a corresponding true limit state function response set comprises: According to each group of initial sample points, based on the flow field reduction method, the gas turbine blade disc is subjected to aero-thermal-mechanical coupling finite element analysis to identify the key fatigue failure positions and main fatigue failure forms, and the life prediction of the gas turbine blade disc is carried out based on the theoretical fatigue life prediction model; based on the life prediction result and the limit state function, a corresponding true limit state function response set is obtained; wherein, The limit state function is expressed as, ； wherein is the limit state function; is the safety life; is the predicted life.

5. The method of claim 1, wherein, In step 3, based on the obtained indicator function, the step of calculating the blade disc failure probability comprises: The calculation expression of the blade disc failure probability is ； wherein is the blade disk failure probability; m is the sample size of the reliability verification sample pool; is the number of failure samples; is the indicator function, which takes the value 1 when the limit state function and 0 otherwise. ​ 6. The method of blade disk fatigue reliability analysis of claim 5, wherein, In step 4, based on the blade disc failure probability obtained in step 3, the step of calculating the coefficient of variation comprises Coefficient of variation The calculation expression for the coefficient of variation is, 。 7. A bladed disk fatigue reliability analysis system, characterized by, The blade disc fatigue reliability analysis method of claim 1 comprises: A data acquisition module is configured to filter and determine blade disc characteristic parameters based on a plurality of blade discs of the same type to be subjected to fatigue reliability analysis, and determine the probability distribution of each blade disc characteristic parameter based on the measured data of each blade disc characteristic parameter filtered and determined. A reliability verification sample pool acquisition module is configured to sample and obtain a reliability verification sample pool containing a plurality of samples based on each blade disc characteristic parameter filtered and determined and the probability distribution of each blade disc characteristic parameter, and jump to execute the steps of the blade disc failure probability acquisition module; wherein each sample in the reliability verification sample pool comprises sample values of each blade disc characteristic parameter filtered and determined in the data acquisition module. A blade disc failure probability acquisition module is configured to input the obtained reliability verification sample pool into a pre-learned MP-ACBS-Kriging surrogate model to output an indicator function, and calculate a blade disc failure probability based on the obtained indicator function. A verification output module is configured to calculate a coefficient of variation based on the blade disc failure probability obtained by the blade disc failure probability acquisition module, and if the coefficient of variation is lower than a preset threshold, the blade disc failure probability obtained by the blade disc failure probability acquisition module is the fatigue reliability analysis result of the plurality of blade discs of the same type, otherwise, the number of samples in the reliability verification sample pool is increased, an updated reliability verification sample pool is obtained, and the steps of the blade disc failure probability acquisition module are executed.

8. An electronic device, comprising: Comprise: At least one processor; And The memory is in communication connection with 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 blade disc fatigue reliability analysis method of any one of claims 1 to 6.

9. A computer readable storage medium storing a computer program, characterized in that, The computer program is executed by a processor to implement the blade disc fatigue reliability analysis method in any one of claims 1 to 6.

Images