A method and system for analyzing factors affecting fatigue performance of a turbine blade, and a medium

Abstract

The application provides a turbine blade fatigue performance influence factor analysis method, system and medium, and belongs to the turbine blade performance prediction technical field, and the method comprises the following steps: generating a small amount of training samples, constructing a back propagation neural network (BP neural network) describing the relationship between the turbine blade fatigue life and the input variables; sequentially increasing the training samples, updating the BP neural network, and until the calculated failure probability reaches convergence; based on the failure samples obtained in the process of solving the failure probability, the conditional failure probability estimate value at different sample points is estimated using the Bayesian inference theory; and according to the average difference between the fatigue failure probability estimate value of the turbine blade and the conditional failure probability estimate value at each failure sample point, the fatigue reliability sensitivity estimate value of each input variable is obtained. The method solves the problems that the existing proxy model method has large sample demand, low efficiency when solving the reliability sensitivity of the turbine blade, and cannot be applied to high-dimensional problems.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of turbine blade performance prediction, and particularly relates to a turbine blade fatigue performance influencing factor analysis method, system and medium. BACKGROUND

[0002] As a core component of an aero-engine and a gas turbine, a turbine blade works in a complex environment of high temperature, high pressure, centrifugal load and alternating stress for a long time, and its fatigue failure problem directly affects the safety and reliability of equipment. Therefore, the fatigue reliability sensitivity of the turbine blade needs to be analyzed in a timely manner to solve the fatigue failure problem of the turbine blade. The traditional turbine blade fatigue failure problem is expanded by using a Monte Carlo simulation method and a first-order / two-order reliability method. The Monte Carlo simulation evaluates the failure probability by random sampling, but requires thousands of times of finite element analysis, and the calculation cost is extremely high. The first-order / two-order reliability method approximates the limit state function by Taylor expansion, but the precision is insufficient for high-dimensional nonlinear problems.

[0003] In recent years, machine learning methods have been introduced to replace traditional methods, mainly including proxy models that use training data to establish the mapping relationship between input variables and fatigue life to replace expensive physical simulation. However, the existing machine learning methods need a large number of sample simulations to establish the relationship between input variables and fatigue life, especially when facing high-dimensional input variables, the calculation cost increases dramatically, which is difficult to meet the engineering efficiency requirements. At the same time, the relationship between the fatigue life of the turbine blade and the input variables is strongly nonlinear, and traditional models or statistical methods are difficult to accurately fit this complex mapping, resulting in insufficient precision of failure probability and sensitivity calculation, which cannot provide reliable support for design optimization. SUMMARY

[0004] In order to solve the problem of how to establish a high-fidelity mapping between turbine blade multivariables and fatigue life with a small number of samples, reduce the number of proxy model calls under the premise of ensuring analysis accuracy, and improve the efficiency of global sensitivity analysis, the application provides a turbine blade fatigue performance influencing factor analysis method.

[0005] In order to achieve the above purpose, the application provides the following technical scheme:

[0006] A turbine blade fatigue performance influencing factor analysis method, comprising:

[0007] Obtain a plurality of structural parameters, material performance parameters and load parameters of the turbine blade to form a multi-dimensional input variable, and generate a sample pool with a capacity of N according to the probability density function of each input variable and the corresponding parameters;

[0008] Randomly select part of the samples from the sample pool to form an initial training set, learn the mapping relationship between the input variables and the fatigue life of the turbine blade through the initial training set, and complete the initial training of the BP neural network;

[0009] calculating fatigue life prediction values corresponding to samples not included in the initial training set in the sample pool by using the BP neural network trained initially; adding a sample closest to a failure surface among the samples not included in the initial training set to the initial training set according to the fatigue life prediction values corresponding to the samples not included in the initial training set and a life threshold, and continuing to train the BP neural network;

[0010] repeating the steps of supplementing samples and updating the network multiple times until a convergence condition is reached, and obtaining a converged BP neural network;

[0011] inputting all samples in the sample pool into the converged BP neural network, and outputting fatigue life prediction values corresponding to each sample; and determining the influence degree of each input variable on the fatigue performance of the turbine blade through the fatigue life prediction values corresponding to each sample.

[0012] Preferably, the step of determining the influence degree of each input variable on the fatigue performance of the turbine blade through the fatigue life prediction values corresponding to each sample specifically comprises the following steps:

[0013] comparing the fatigue life prediction values corresponding to each sample with the life threshold, screening out failure samples with fatigue life prediction values lower than the life threshold, and counting the number of failure samples;

[0014] calculating a fatigue failure probability estimate of the turbine blade according to the total number of samples in the sample pool and the number of failure samples;

[0015] based on the failure sample information, estimating a conditional failure probability estimate at a given sample point by using a specified likelihood function and a Bayesian inference theory; the given sample point is a group of samples randomly generated by a probability density function of input variables;

[0016] calculating fatigue reliability sensitivity estimates of each input variable according to the fatigue failure probability estimate of the turbine blade and the conditional failure probability estimates at each failure sample point, and quantifying the influence degree of each input variable on the fatigue performance of the turbine blade through the fatigue reliability sensitivity estimates.

[0017] Preferably, the step of repeating the steps of supplementing samples and updating the network multiple times until a convergence condition is reached, and obtaining a converged BP neural network specifically comprises:

[0018] after supplementing samples to the training set and updating the BP neural network each time, calculating fatigue life prediction values corresponding to each sample by using the updated network once, and calculating a fatigue failure probability estimate of the turbine blade once again, and continuously performing the updating and calculating at least 5 times to obtain at least 5 fatigue failure probability estimates of the turbine blade;

[0019] Calculate the mean and standard deviation of the fatigue failure probability estimates for at least five turbine blades, and obtain the final coefficient of variation based on the ratio of the mean and standard deviation.

[0020] The converged BP neural network is obtained when the coefficient of variation of the estimated failure probability of five consecutive failures is less than a given threshold.

[0021] Preferably, the formula for calculating the fatigue failure probability estimate of the turbine blade is:

[0022] ;

[0023] In the formula, This is an estimate of the fatigue failure probability of the turbine blade. This represents the number of failed samples. This represents the total number of samples in the sample pool.

[0024] Preferably, the formula for calculating the conditional failure probability estimate is as follows:

[0025] ;

[0026] In the formula, This represents the estimated probability of conditional failure. This represents the number of failed samples. For the sample The One portion, For the first One failure sample The One component; Represents the likelihood function. Indicates input variables The probability density function.

[0027] Preferably, the formula for calculating the fatigue reliability sensitivity estimate is as follows:

[0028] ;

[0029] In the formula, This is the fatigue reliability sensitivity estimate.

[0030] Preferably, the failure surface is specifically defined as: when the fatigue life prediction value obtained by mapping the input variable through a BP neural network is exactly equal to the set fatigue life threshold, all sample points that satisfy the correspondence between the input variable and the life threshold together constitute a high-dimensional spatial boundary, and the high-dimensional spatial boundary is the failure surface.

[0031] Preferably, the structure parameters include the position and radius of the air hole, the material performance parameters include the transverse Young's modulus, the longitudinal Young's modulus, the transverse Poisson's ratio, the longitudinal Poisson's ratio, the transverse shear modulus, the longitudinal shear modulus and the density, and the load parameters include the maximum speed of the primary cycle, the maximum speed of the secondary cycle and the minimum speed of the secondary cycle.

[0032] The application further provides a turbine blade fatigue performance influencing factor analysis system.

[0033] A sample generation module is configured to acquire a plurality of structure parameters, material performance parameters and load parameters of the turbine blade, form a plurality of input variables, and generate a sample pool with a capacity of N according to the probability density functions of the input variables and the corresponding parameters.

[0034] An initial model training module is configured to randomly select part of the samples from the sample pool to form an initial training set, learn the mapping relationship between the input variables and the fatigue life of the turbine blade through the initial training set, and complete the initial training of the BP neural network.

[0035] A model iterative training module is configured to calculate the fatigue life prediction value corresponding to the samples not included in the initial training set in the sample pool by using the BP neural network trained initially, add the sample closest to the failure surface among the samples not included in the initial training set to the initial training set according to the fatigue life prediction value corresponding to the samples not included in the initial training set and the life threshold, and continue to train the BP neural network.

[0036] The steps of repeating iteration, supplementing samples and updating the network are repeated multiple times until a convergence condition is reached, and a converged BP neural network is obtained.

[0037] An analysis module is configured to input all the samples in the sample pool into the converged BP neural network, output the fatigue life prediction value corresponding to each sample, and determine the influence degree of each input variable on the fatigue performance of the turbine blade through the fatigue life prediction value corresponding to each sample.

[0038] The application further provides a computer readable storage medium, wherein the storage medium stores a computer program, and the computer program can execute the steps of any one of the turbine blade fatigue performance influencing factor analysis methods when loaded by a processor.

[0039] The turbine blade fatigue performance influencing factor analysis method provided by the application has the following beneficial effects:

[0040] The present application obtains multiple structure parameters, material performance parameters and load parameters of a turbine blade, forms multiple input variables, generates a sample pool with a capacity of N according to the probability density function of each input variable and the corresponding parameters, expands the initial parameters, can reduce the number of initial samples, and does not need a large number of samples. Further, by preferentially supplementing critical samples that are most critical to failure probability calculation, the introduction of redundant samples is reduced, thereby quickly improving the fitting accuracy of the mapping relationship near the failure surface of the BP neural network based on a small number of samples, constructing an effective model through small sample initial training and targeted sample supplement, significantly improving the calculation efficiency in the high-dimensional input variable scene, meeting the demand for efficiency of engineering application, reducing the sample demand and calculation cost, and accurately fitting the strong nonlinear relationship between the input variable and the fatigue life. Subsequently, all samples in the sample pool are input into the network, and a high-precision fatigue life prediction value can be output, providing a reliable data basis for determining the influence degree of each input variable on fatigue performance. Based on the fatigue life prediction value of each sample output by the converged BP neural network, the influence of each input variable on the fatigue performance of the turbine blade can be further quantified, the problem that the influence degree determination is unreliable due to the insufficient fitting accuracy of the traditional model is solved, and effective basis is provided for turbine blade design optimization. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present application and the design scheme thereof, the drawings required by the present embodiments will be briefly introduced as follows. The drawings in the following description are only part of the embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0042] Figure 1 The flow chart of the turbine blade fatigue performance influence factor analysis method proposed by the present application;

[0043] Figure 2 The turbine blade simulation structure schematic diagram for the embodiments of the present application;

[0044] Figure 3 The stress distribution nephogram of the turbine blade structure for the embodiments of the present application under the maximum rotating speed; wherein, Figure 3 (a) is a stress distribution nephogram, Figure 3 (b) is a partial enlarged view of the framed part in (a);

[0045] Figure 4 The reliability sensitivity analysis result of the turbine blade for the embodiments of the present application. DETAILED DESCRIPTION

[0046] To enable those skilled in the art to better understand and implement the technical solutions of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0047] Example 1

[0048] Bayesian inference methods can quantify uncertainty through prior and posterior distributions, enabling the extraction of conditional probability features. Therefore, this invention proposes a novel method that integrates the nonlinear fitting capabilities of neural networks with the probabilistic advantages of Bayesian inference to achieve high-precision and high-efficiency fatigue reliability sensitivity prediction of turbine blades, providing effective support for the reliability design and optimization of turbine blades.

[0049] Based on this, the present invention provides a method for analyzing factors affecting the fatigue performance of turbine blades, specifically a method for predicting the fatigue reliability sensitivity of turbine blades based on neural networks and Bayesian inference, such as... Figure 1 As shown, the prediction method includes the following steps:

[0050] Step S1: Obtain multiple structural parameters, material property parameters, and load parameters of the turbine blade to form multi-dimensional input variables. Generate a sample pool to predict the fatigue reliability of the turbine blade based on the probability density function of each input variable and its corresponding parameters. ,in This indicates the capacity of the sample pool. Structural parameters include the location and radius of the air mold holes; material property parameters include transverse Young's modulus, longitudinal Young's modulus, transverse Poisson's ratio, longitudinal Poisson's ratio, transverse shear modulus, longitudinal shear modulus, and density; load parameters include the maximum speed of the main cycle, the maximum speed of the secondary cycle, and the minimum speed of the secondary cycle.

[0051] Step S2, from the sample pool A small number of samples are randomly selected to form the initial training set. Construct a structure to describe the relationship between turbine blade fatigue life and input variables. BP neural network The initial training of the BP neural network is completed by learning the mapping relationship between input variables and turbine blade fatigue life through the initial training set.

[0052] Step S3: Calculate the fatigue life prediction value corresponding to the sample in the sample pool that was not included in the initial training set using the initially trained BP neural network; based on the fatigue life prediction value and life threshold corresponding to the sample that was not included in the initial training set, add the sample closest to the failure surface among the samples not included in the initial training set to the initial training set, and continue training the BP neural network.

[0053] Specifically, according to the life threshold in the turbine blade fatigue reliability analysis , the sample closest to the failure surface is selected according to the following formula and added to the initial training set , and the BP neural network is reconstructed .

[0054] ;

[0055] In the formula, , the sample added to the training set .

[0056] In this embodiment, the failure surface is a judgment boundary established based on the fatigue life threshold of the turbine blade: when the fatigue life prediction value obtained by mapping the input variables (structural parameters, material performance parameters, load parameters) through the BP neural network is exactly equal to the set fatigue life threshold, the high-dimensional space boundary formed by all sample points satisfying the corresponding relationship of the input variables-life threshold is the failure surface. The failure surface is the basis for screening the sample closest to the failure surface and the failure sample. When screening the failure sample: if the fatigue life prediction value of a sample is (T0, i.e. the sample point is located on the "failure side" of the failure surface), the sample is defined as a failure sample; when screening the sample closest to the failure surface, the deviation (such as -T0|) of the fatigue life prediction value of the sample from the threshold T0 is calculated, and the sample with the smallest deviation is the sample closest to the failure surface. Such samples are most critical to improving the fitting accuracy of the BP neural network near the failure boundary.

[0057] Step S4, repeat step S3 at least 5 times, and calculate the coefficient of variation of the failure probability estimation values obtained by using the BP neural network in the last 5 times, until the coefficient of variation of the failure probability estimation values of the turbine blade in the last 5 times is less than a given threshold, and a converged BP neural network is obtained.

[0058] If the coefficient of variation is less than the given threshold, the number of failure samples satisfying is counted , specifically: input all samples in the sample pool into the converged BP neural network, and output the fatigue life prediction values corresponding to each sample. Compare the fatigue life prediction values corresponding to each sample with the life threshold, select the failure samples with fatigue life prediction values lower than the life threshold, and count the number of failure samples.

[0059] According to the total number of samples in the sample pool and the number of failure samples, the fatigue failure probability estimation value of the turbine blade is calculated using the following formula:

[0060] ;

[0061] In the formula, The predicted value representing the probability of failure. N This represents the total number of samples.

[0062] If the coefficient of variation is greater than the given threshold, return to step S3.

[0063] Specifically, the coefficient of variation is obtained through the following method:

[0064] Each time samples are added to the training set and the BP neural network is updated, the fatigue life prediction value corresponding to each sample is calculated once using the updated network. Then, the fatigue failure probability estimate of the turbine blade is calculated again. This process is repeated at least 5 times to obtain the fatigue failure probability estimate of at least 5 turbine blades.

[0065] Calculate the mean and standard deviation of the fatigue failure probability estimates for at least five turbine blades, and obtain the final coefficient of variation based on the ratio of the mean and standard deviation.

[0066] Step S5: Based on the failure sample information, estimate the conditional failure probability at a given sample point using a specified likelihood function and Bayesian inference theory.

[0067] Specifically, the failure sample obtained in step S4 is denoted as... Using these samples, estimate a given sample according to the following formula. The following is an estimate of the probability of failure under certain conditions:

[0068] ;

[0069] In the formula, This represents the number of failed samples. The likelihood function is defined as follows: , With a mean of 0 and a standard deviation of The probability density function of a normal distribution. This represents the estimated probability of conditional failure. Indicates input variables The probability density function.

[0070] Step S6: Based on the estimated fatigue failure probability of the turbine blade and the estimated conditional failure probability at each failure sample point, calculate the estimated fatigue reliability sensitivity of the turbine blade using the following formula. , The dimensions of the input variables are defined. The impact of each input variable on the fatigue performance of the turbine blade is quantified using fatigue reliability sensitivity estimates.

[0071] ;

[0072] In the formula, For the sample the first component, represents an estimate of sensitivity , represents an estimate of the fatigue failure probability of the turbine blade.

[0073] The turbine blade fatigue reliability sensitivity prediction method based on neural network and Bayesian inference provided by the embodiment of the present disclosure will be described in detail below with reference to the accompanying drawings:

[0074] In step S1, consider the turbine blade simulation structure as shown in Figure 2 , including the blade body, the rim plate 3 and the tenon tooth 2, the blade body includes the blade basin 5, the blade back 1, the film hole 4 and other structures. The turbine blade material is nickel-based high-temperature alloy DZ125, the density is , the linear expansion coefficient, the ultimate strength, the conditional yield strength, the elastic modulus, the shear modulus, the Poisson's ratio and the specific heat capacity at different temperatures can refer to the material manual.

[0075] The load of the turbine blade mainly considers the centrifugal force generated by the high-speed rotation of the blade, which is closely related to the rotating speed. The working temperature of the blade is considered as , the aerodynamic force at the blade basin is , and the aerodynamic force at the blade back is . Table 1 is the simplified main cycle load and secondary cycle load under the condition of one take-off and landing.

[0076] Table 1 Simplified main cycle and secondary cycle load under the condition of one take-off and landing

[0077]

[0078] Load the rotating speed, aerodynamic force and other loads on the turbine blade model to obtain the stress distribution nephogram at the maximum rotating speed as shown in Figure 3 . As shown in (b) of Figure 3 , the stress around the film hole in the second column closest to the tenon tooth end of the turbine blade is the largest, so this place is taken as the dangerous position of the turbine blade for subsequent analysis.

[0079] After the examination point is determined, the Manson-Coffin model with Morrow elastic stress linear correction is used to predict the fatigue life of the turbine blade. The fatigue life model at is determined by using the least square method according to the experimental data as follows:

[0080] ;

[0081] In the formula, is the strain amplitude; is the maximum stress; It is the elastic modulus; The fatigue life is defined as the sum of the primary and secondary cycles. Miner's linear cumulative damage theory is used to address the cumulative effects of these cycles.

[0082] The main random factors affecting the combined high- and low-cycle fatigue life of turbine blades under primary and secondary cycles include, but are not limited to, structural geometry, material properties, and loads. For the turbine blade structure, the selected random variables of structural geometry include the positions of the air mold holes in column 1. The position of the air mold holes in the second column and air mold hole radius The selected random variables representing material properties include transverse Young's modulus. Longitudinal Young's modulus Horizontal Poisson's ratio Longitudinal Poisson's ratio Transverse shear modulus and longitudinal shear modulus ,density The selected load-type random variables include the maximum rotational speed of the main loop. Maximum speed of the next cycle and the minimum speed of the next cycle The distribution forms and distribution parameters of each input variable are shown in Table 2.

[0083] Let the random input variables composed of the 13 random input variables in Table 2 be denoted as _____. Then, the combined high- and low-cycle fatigue life of a turbine blade under successive cycles can be expressed as the input variable. function .according to The distribution form and distribution parameters generate A sample pool is constructed from 1 sample. .

[0084] Table 2. Distribution form and distribution parameters of input variables in the fatigue reliability sensitivity analysis of turbine blades.

[0085]

[0086] In step S2, from the sample pool A training set was formed by randomly selecting 130 samples. Constructing a BP neural network .

[0087] In steps S3 and S4, according to... Select new training samples to add Finally, 19 training samples were added to bring the estimated failure probability to convergence. The obtained estimated failure probability is... The number of failed samples is .

[0088] In steps S5 and S6, by using 492 failure samples, the fatigue life reliability sensitivity of the turbine blade can be obtained by using the Bayesian inference method, as shown in Table 3 and Figure 4 Figure 4 For the reliability sensitivity of each input variable of the turbine blade, the ordinate is the sensitivity value, and the ranking of each sensitivity value can be directly observed.

[0089] Based on the analysis results of the fatigue reliability sensitivity of the turbine blade, the following conclusions can be drawn: there is a significant difference in the influence degree of each parameter on the fatigue reliability, wherein the maximum speed of the secondary cycle and the maximum speed of the primary cycle are ranked first and second, respectively, indicating that the speed parameter has the most critical influence on the fatigue reliability. The position of the film hole, the transverse Young's modulus and the film hole radius are ranked third to fifth, respectively, indicating that the gas film cooling structure parameters and the material transverse stiffness also have important influence on the fatigue reliability; the influence of the density, the longitudinal Poisson's ratio and other parameters is relatively small, and the influence of the longitudinal Young's modulus and the transverse Poisson's ratio is the weakest. This ranking result shows that in the fatigue reliability optimization design of the turbine blade, the optimization of the speed operating parameter and the gas film cooling structure parameter should be prioritized, and the improvement of the material transverse mechanical property can be considered secondly, and the influence of the longitudinal mechanical property parameter can be appropriately relaxed. The analysis result can provide clear guidance for the reliability design and parameter optimization of the turbine blade, and help to improve the fatigue life of the blade under the premise of ensuring performance.

[0090] Table 3 Analysis results of fatigue life reliability sensitivity of turbine blade

[0091]

[0092] The present application directly quantifies the influence degree of each input sample on the failure probability through the fatigue reliability sensitivity, can clearly distinguish the key influence parameters and the secondary parameters, provides a clear direction for the reliability design and optimization of the turbine blade, helps to prioritize the adjustment of the high sensitivity parameters under the premise of ensuring performance, improves the reliability of the blade fatigue life, and solves the defects that the existing proxy model method has low efficiency or cannot be applied to high-dimensional problems when solving the reliability sensitivity of the turbine blade.

[0093] Based on the same inventive concept, the present application also proposes a turbine blade fatigue performance influence factor analysis system, comprising:

[0094] A sample generation module is configured to obtain a plurality of structure parameters, material performance parameters and load parameters of a turbine blade, to constitute a plurality of input variables, and to generate a sample pool with a capacity of N according to the probability density function of each input variable and the corresponding parameters.

[0095] ​The model initial training module is configured to randomly select part of samples from the sample pool to form an initial training set, learn a mapping relationship between the input variables and the turbine blade fatigue life through the initial training set, and complete initial training of the BP neural network.

[0096] The model iterative training module is configured to calculate fatigue life prediction values corresponding to samples not included in the initial training set in the sample pool by using the BP neural network trained initially, add samples closest to a failure surface from the samples not included in the initial training set to the initial training set according to the fatigue life prediction values corresponding to the samples not included in the initial training set and the life threshold, and continue training the BP neural network.

[0097] The steps of repeating iteration and supplement of samples and updating the network are repeated multiple times until a convergence condition is reached, and a converged BP neural network is obtained.

[0098] The analysis module is configured to input all samples in the sample pool into the converged BP neural network, output fatigue life prediction values corresponding to each sample, and determine the influence degree of each input variable on the turbine blade fatigue performance through the fatigue life prediction values corresponding to each sample.

[0099] The modules in the turbine blade fatigue performance influence factor analysis system described above can be realized by software, hardware, or a combination thereof. The modules described above can be embedded in or independent of a processor in a computer device in a hardware form, or can be stored in a memory in a computer device in a software form, so as to be called and executed by a processor to perform the operations corresponding to the modules.

[0100] Further, the present application also provides a non-transitory computer readable storage medium containing instructions, and the storage medium stores a computer program. For example, a memory containing instructions, the instructions can be executed by a processor of a computer device to complete the method described above. For example, the non-transitory computer readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc. When the computer program is executed by the processor, the steps in the turbine blade fatigue performance influence factor analysis method embodiment can be realized. For specific implementation methods, please refer to the method embodiments, which will not be described here.

[0101] Those skilled in the art should understand that the embodiments of the present application can provide a method, a system or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented 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.

[0102] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 means for functionally implementing the steps listed in the flowchart block or blocks.

[0103] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 means for functionally implementing the steps listed in the flowchart block or blocks.

[0104] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 means for functionally implementing the steps listed in the flowchart block or blocks.

[0105] It should be noted that the above-mentioned detailed embodiments can make those skilled in the art more fully understand the present application, but in no way limit the present application. Therefore, although the present application has been described in detail in the present specification and examples, those skilled in the art should understand that the present application can still be modified or replaced by equivalents; and all technical solutions and improvements which do not deviate from the spirit and scope of the present application are covered in the protection scope of the present application. Any reference signs in the claims should not be considered as limiting the claims. Any simple changes or equivalent replacements which are obvious to those skilled in the art within the technical scope disclosed in the present application are all within the protection scope of the present application.

Claims

1. A method for analyzing factors affecting fatigue performance of a turbine blade, characterized by, The method comprises the following steps: obtaining a plurality of structural parameters, material performance parameters and load parameters of a turbine blade to form a plurality of input variables, generating a sample pool with a capacity of N according to the probability density functions of the input variables and the corresponding parameters; randomly selecting part of the samples in the sample pool to form an initial training set, learning the mapping relationship between the input variables and the fatigue life of the turbine blade through the initial training set, and completing the initial training of the BP neural network; calculating the fatigue life prediction value corresponding to the samples not included in the initial training set in the sample pool by using the BP neural network trained initially; adding the sample closest to the failure surface among the samples not included in the initial training set to the initial training set according to the fatigue life prediction value corresponding to the sample not included in the initial training set and the life threshold, and continuing to train the BP neural network; repeating the steps of supplementing samples and updating the network for multiple times until a convergence condition is reached, and obtaining a converged BP neural network; inputting all the samples in the sample pool into the converged BP neural network, and outputting the fatigue life prediction value corresponding to each sample; determining the influence degree of each input variable on the fatigue performance of the turbine blade through the fatigue life prediction value corresponding to each sample; the determination of the influence degree of each input variable on the fatigue performance of the turbine blade through the fatigue life prediction value corresponding to each sample comprises the following steps: comparing the fatigue life prediction value corresponding to each sample with the life threshold, screening out failure samples with fatigue life prediction values lower than the life threshold, and counting the number of failure samples; calculating the fatigue failure probability estimate value of the turbine blade according to the total number of samples in the sample pool and the number of failure samples; based on the failure sample information, estimating the conditional failure probability estimate value at a given sample point by using a specified likelihood function and a Bayesian inference theory; the given sample point is a group of samples randomly generated by the probability density function of the input variables; calculating the fatigue reliability sensitivity estimate value of each input variable according to the fatigue failure probability estimate value of the turbine blade and the conditional failure probability estimate value at each failure sample point, and quantifying the influence degree of each input variable on the fatigue performance of the turbine blade through the fatigue reliability sensitivity estimate value; the steps of repeating the iteration of supplementing samples and updating the network for multiple times until a convergence condition is reached, and obtaining a converged BP neural network, specifically comprising: after supplementing samples to the training set and updating the BP neural network each time, calculating the fatigue life prediction value corresponding to each sample by using the updated network, and then repeatedly calculating the fatigue failure probability estimate value of the turbine blade, and continuously performing the updating and calculation for at least 5 times to obtain at least 5 fatigue failure probability estimate values of the turbine blade; calculating the mean value and the standard deviation of the at least 5 fatigue failure probability estimate values respectively, and obtaining the final coefficient of variation according to the ratio of the mean value and the standard deviation; until the coefficient of variation of the failure probability estimate values of the adjacent 5 times is less than a given threshold, the converged BP neural network is obtained.

2. The method of claim 1, wherein, the calculation formula of the fatigue failure probability estimate value of the turbine blade is: ； wherein is the fatigue failure probability estimate for a turbine blade, is the number of failure samples, is the total number of samples in the sample pool.

3. The method of claim 2, wherein: the calculation formula of the conditional failure probability estimate value is: ； wherein denotes the conditional failure probability estimate, is the number of failure samples, is a sample of the first component, is the first component of the failure sample ; is the second component of the failure sample ; denotes the likelihood function, denotes the probability density function of the input variable 4. The method of claim 3, wherein the method further comprises: the calculation formula of the fatigue reliability sensitivity estimate value is: ； In the formula, is the fatigue reliability sensitivity estimate.

5. The method of claim 1, wherein: The failure surface is specifically: when the fatigue life prediction value of the input variable mapped by the BP neural network is exactly equal to the set fatigue life threshold, all sample points satisfying the input variable-life threshold correspondence jointly constitute a high-dimensional space boundary, and the high-dimensional space boundary is the failure surface.

6. The method of claim 1, wherein: The structural parameters include the positions and radii of the air die holes, the material performance parameters include the transverse Young's modulus, the longitudinal Young's modulus, the transverse Poisson's ratio, the longitudinal Poisson's ratio, the transverse shear modulus, the longitudinal shear modulus and the density, and the load parameters include the maximum speed of the primary cycle, the maximum speed of the secondary cycle and the minimum speed of the secondary cycle.

7. A system for analyzing factors affecting the fatigue performance of turbine blades, characterized in that, The method comprises the steps of: The sample generation module is configured to obtain a plurality of structural parameters, material performance parameters and load parameters of the turbine blade, to form a plurality of input variables, and to generate a sample pool with a capacity of N according to the probability density functions of the input variables and the corresponding parameters; The model initial training module is configured to randomly select part of the samples from the sample pool to form an initial training set, to learn the mapping relationship between the input variables and the fatigue life of the turbine blade through the initial training set, and to complete the initial training of the BP neural network; The model iterative training module is configured to calculate the fatigue life prediction values corresponding to the samples not included in the initial training set by using the BP neural network trained initially; The sample closest to the failure surface among the samples not included in the initial training set is added to the initial training set according to the fatigue life prediction values corresponding to the samples not included in the initial training set and the life threshold, and the BP neural network is continuously trained; The steps of supplementing the samples and updating the network are repeated multiple times until a convergence condition is reached, and a converged BP neural network is obtained; The analysis module is configured to input all the samples in the sample pool into the converged BP neural network and to output the fatigue life prediction values corresponding to the samples; The influence degree of each input variable on the fatigue performance of the turbine blade is determined according to the fatigue life prediction values corresponding to the samples. The influence degree of each input variable on the fatigue performance of the turbine blade is determined according to the fatigue life prediction values corresponding to the samples. The influence degree of each input variable on the fatigue performance of the turbine blade is determined according to the fatigue life prediction values corresponding to the samples. The influence degree of each input variable on the fatigue performance of the turbine blade is determined according to the fatigue life prediction values corresponding to the samples. The influence degree of each input variable on the fatigue performance of the turbine blade is determined according to the fatigue life prediction values corresponding to the samples. ​ ​ After each time the training set is supplemented with samples and the BP neural network is updated, the updated network is used to calculate the fatigue life prediction value corresponding to each sample, and the fatigue failure probability estimation value of the turbine blade is calculated again. This updating and calculation is repeated at least 5 times to obtain at least 5 fatigue failure probability estimation values of the turbine blade; The mean and standard deviation of the at least 5 fatigue failure probability estimation values of the turbine blade are calculated respectively, and the final coefficient of variation is obtained according to the ratio of the mean and the standard deviation; Until the coefficient of variation of the adjacent 5 failure probability estimation values is less than a given threshold value, a converged BP neural network is obtained.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program can execute the steps of the method of any one of claims 1 to 6 when loaded by a processor.

Images