Non-probabilistic fatigue reliability analysis method for propeller hub mechanism of data-driven down-regulation pitch propeller

Abstract

The invention relates to a non-probabilistic fatigue reliability analysis method for a controllable-pitch propeller hub mechanism under data driving, which comprises the following steps of: considering multiple uncertain sources such as elastic modulus and friction coefficient, modeling each part of the controllable-pitch propeller hub mechanism through parameterization to obtain a finite element model under different uncertain variable values; drawing a fatigue load spectrum by using a finite element simulation load result, and performing fatigue simulation by combining an S-N curve to obtain the fatigue life of each part; and obtaining a non-probability interval model of the fatigue life of each part by using a data-driven agent model and a Monte Carlo method under a limited fatigue life sample condition, and calculating the fatigue reliability of each part by using a non-probability area method. A data driving method is adopted, so that the problems of high cost and time consumption of obtaining enough data in actual engineering are avoided; fatigue reliability analysis is carried out by adopting a non-probabilistic method, so that uncertainty in engineering can be better dealt with, and the fatigue reliability of the controllable-pitch propeller hub mechanism can be more comprehensively evaluated.

Description

Technical Field

[0001] The present invention relates to the technical field of structural fatigue reliability analysis, and particularly to a non-probabilistic fatigue reliability analysis method for a controllable pitch propeller hub mechanism driven by data. Background Art

[0002] Since the first controllable pitch propeller came out in Canada more than 80 years ago, it has always attracted people's attention with its unique excellent performance. Compared with ordinary fixed pitch propellers, the advantage of controllable pitch propellers is that they can adjust the pitch at any time according to the ship's navigation conditions, so as to achieve the purposes of reducing fuel consumption and extending the life of the main engine. Therefore, it is very important to analyze the characteristics of controllable pitch propellers.

[0003] A controllable pitch propeller (referred to as a CPP) consists of a propeller blade, a hub mechanism, an oil distributor, an in-shaft oil pipe, a hydraulic system, an electric control system, etc. The propeller blade is not fixed to the hub and can rotate around an axis perpendicular to the propeller shaft. The pitch adjustment mechanism in the hub is used to drive the propeller blade to rotate, change the pitch of the propeller, and thus change the magnitude and direction of the thrust received by the propeller blade to meet the requirements of the ship for going forward, backward, braking, and speed change.

[0004] Fatigue reliability refers to the ability of a structure or component to resist fatigue failure when subjected to alternating loads under long-term service conditions. Fatigue failure is usually caused by the gradual accumulation of damage in the material under cyclic stress or strain, eventually leading to the formation and propagation of cracks until fracture. Macroscopically, when the cyclic stress level is relatively low, elastic strain plays a dominant role, and the fatigue life is relatively long at this time, which is called stress fatigue or high-cycle fatigue; when the cyclic loading level is relatively high, plastic strain plays a dominant role, and the fatigue life is relatively short at this time, which is called strain fatigue or low-cycle fatigue. Since fatigue damage is cumulative, fatigue damage is very dangerous, and great attention must be paid during design. Fatigue failure is one of the main reasons for structural failure. Especially in a high-cycle load environment, the fatigue performance of a structure is directly related to its service life and safety. Analyzing the fatigue reliability of a controllable pitch propeller device aims to evaluate the safety of the structure.

[0005] Traditional fatigue reliability analysis methods mainly rely on probability statistical models. These methods usually evaluate the fatigue life of a structure under alternating loads based on the S-N curve of the material or fracture mechanics theory. However, traditional fatigue reliability analysis methods face some challenges in engineering applications. These methods usually rely on a large amount of data, and in actual engineering, obtaining sufficient data may be costly and time-consuming. Summary of the Invention

[0006] Aiming at the problem that the existing fatigue reliability analysis relies on a large amount of data, a non-probabilistic fatigue reliability analysis method for the pitch-changing propeller hub mechanism driven by data is proposed, and the fatigue reliability analysis of the pitch-changing propeller hub mechanism is carried out based on this method. This method introduces non-probabilistic theory and data-driven technology, and combines limited data to analyze the fatigue reliability of the pitch-changing propeller hub mechanism. Compared with the traditional fatigue reliability analysis method based on the probability statistical model, this method can better cope with the problems of data scarcity and uncertainty in engineering, and the non-probabilistic reliability analysis method can more comprehensively evaluate the fatigue reliability of the pitch-changing propeller hub mechanism through the analysis and modeling of multiple failure modes.

[0007] The technical solution of the present invention is: a non-probabilistic fatigue reliability analysis method for the pitch-changing propeller hub mechanism driven by data, specifically including the following steps:

[0008] 1) Finite element model establishment: Considering the uncertainties of the material elastic modulus, Poisson's ratio, friction coefficient and component fitting clearance of each component of the pitch-changing propeller hub mechanism, parametric modeling is carried out on each component of the pitch-changing propeller hub mechanism to obtain a high-precision finite element model of each component of the pitch-changing propeller hub mechanism;

[0009] 2) Fatigue simulation analysis: Apply the hydrodynamic fluctuation load during the operation and rotation of the simulated blade to the pitch-changing propeller hub mechanism after modeling and assembly. Under the hydrodynamic conditions of the blade, compile the fatigue load spectrum according to the stress amplitude of each component, and calculate the fatigue life of each component of the pitch-changing propeller hub mechanism based on the original S-N curve of the component material in the fatigue load spectrum;

[0010] 3) Data-driven surrogate model: According to the calculated finite fatigue life samples of each component, use the data-driven surrogate model to establish the relationship between the basic uncertainty variables and the fatigue life. On the basis of this relationship, introduce the interval non-probabilistic reliability model, so as to establish the fatigue life interval of each component affected by the basic uncertainty variables;

[0011] 4) Fatigue reliability calculation: Use the fatigue life intervals of each component obtained in step three to evaluate the fatigue reliability of each component by the non-probabilistic area method.

[0012] Furthermore, the surrogate model in step 3) selects the Kriging surrogate model. As a common and efficient response interval calculation method, it is used for uncertainty propagation. The Kriging surrogate model consists of two main parts: the fitting term and the correlation term. The fitting term constructs the fatigue life curve through the function values and derivative values of the sample points, and the correlation term describes the correlation between the sample points through the covariance function, so that the Kriging surrogate model not only realizes the unbiased estimation at the sample points, but also gives the corresponding prediction error when predicting the fatigue life.

[0013] Furthermore, the interval non-probabilistic reliability model combines the Kriging surrogate model and the Monte Carlo method to obtain the fatigue life boundary information of each component.

[0014] Furthermore, in the interval non-probabilistic reliability model, the reliability index for evaluating structural safety is defined by the boundary characteristics of the actual response of the structure and the allowable interval of the response, φ s is the allowable value of the fatigue life of the components of the controllable pitch propeller mechanism, φ a is the actual value of the fatigue life of the components of the controllable pitch propeller mechanism. Then, the safety state of the mechanism is judged using the performance function in reliability theory:

[0015] M(φ a , φ s ) = φ s - φ a ,

[0016] When the performance function M > 0, it is considered that the controllable pitch propeller mechanism is in a safe state. When M < 0, it is considered that the controllable pitch propeller mechanism is in a failure state. Thus, the structural failure plane can be defined

[0017] M(φ a , φ s ) = φ s - φ a = 0,

[0018] The actual value and the allowable value of the structural response function are normalized:

[0019]

[0020] where and are the radius of the intervals to which φ s and φ a belong, respectively, and are the average values of the intervals to which φ s and φ a belong, respectively; substituting δφ s and δφ a into M(φ a , φ s ) = 0, the normalized failure plane can be obtained:

[0021]

[0022] Taking δφ a as the abscissa and δφ s as the ordinate, the ranges of δφ a and δφ s and the failure plane described by the above equation are represented in the Cartesian coordinate system. Then, δφ a and δφs The enclosed area is divided into a safe area and a failure area by the failure plane, corresponding to M>0 and M<0 respectively, and is used to evaluate the fatigue reliability of each part.

[0023] A non-probabilistic fatigue reliability analysis method for an assembled structure takes each component of the assembled structure as the research object, considers the uncertainties of the material elastic modulus, Poisson's ratio, friction coefficient and component mating clearance of each component, obtains the finite element model under different values of uncertain variables through parametric part modeling, uses the finite element simulation load results to draw the fatigue load spectrum, and combines the S-N curve to conduct fatigue simulation to obtain the fatigue life of each component; under the condition of a limited fatigue life sample, uses the data-driven surrogate model and the Monte Carlo method to obtain the non-probabilistic interval model of the fatigue life of each component, and uses the non-probabilistic area method to calculate the fatigue reliability of each component of the assembled structure.

[0024] The beneficial effects of the present invention are as follows: The non-probabilistic fatigue reliability analysis method for the controllable pitch propeller hub mechanism driven by data of the present invention adopts parametric modeling to efficiently establish a finite element model under different parameter values; the present invention adopts a data-driven method to avoid the problem that it is costly and time-consuming to obtain sufficient data in actual engineering; the present invention adopts a non-probabilistic method to conduct fatigue reliability analysis, which can better cope with the uncertainty problems in engineering and more comprehensively evaluate the fatigue reliability of the controllable pitch propeller hub mechanism. Description of the Drawings

[0025] Figure 1 It is a schematic flow chart of the non-probabilistic reliability analysis method for the controllable pitch propeller hub mechanism driven by data of the present invention;

[0026] Figure 2 It is a schematic diagram of an embodiment of the present invention;

[0027] Figures 3A - 3H It is the fatigue load spectrum of each component of the controllable pitch propeller hub mechanism in the embodiment of the present invention;

[0028] Figures 4A - 4F It is the S-N curve of each component of the controllable pitch propeller hub mechanism in the embodiment of the present invention;

[0029] Figures 5A - 5B It is a schematic diagram of the data-driven Kriging surrogate model in the embodiment of the present invention;

[0030] Figure 6 It is the non-probabilistic fatigue life interval model of each component of the controllable pitch propeller hub mechanism in the embodiment of the present invention. Detailed Embodiment

[0031] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented on the premise of the technical solution of the present invention, and gives detailed implementation manners and specific operation processes, but the protection scope of the present invention is not limited to the following embodiments.

[0032] The non-probabilistic reliability analysis method of the pitch-adjusting propeller hub mechanism driven by data of the present invention takes each component of the pitch-adjusting propeller hub mechanism as the research object, considers multiple uncertain sources such as elastic modulus and friction coefficient, obtains finite element models under different values of uncertain variables through parametric part modeling, uses the finite element simulation load results to draw fatigue load spectra, and combines with the S-N curve to conduct fatigue simulation to obtain the fatigue life of each component. Under the condition of a limited number of fatigue life samples, a data-driven surrogate model is used and the Monte Carlo method is used to obtain the non-probabilistic interval model of the fatigue life of each component, and the non-probabilistic area method is used to calculate the fatigue reliability of each component of the pitch-adjusting propeller hub mechanism.

[0033] As Figure 1 shown, the method specifically includes the following steps:

[0034] Step 1: Establishment of finite element model: Considering the uncertainties of the material elastic modulus, Poisson's ratio, friction coefficient and component fitting clearance of each component of the pitch-adjusting propeller hub mechanism, parametric modeling is carried out on each component of the pitch-adjusting propeller hub mechanism to obtain high-precision finite element models of each component of the pitch-adjusting propeller hub mechanism;

[0035] Step 2: Fatigue simulation analysis: Apply the hydrodynamic fluctuation load during the simulated rotation of the propeller blade to the pitch-adjusting propeller hub mechanism after modeling and assembly. Under the hydrodynamic condition of the propeller blade, compile the fatigue load spectrum according to the stress amplitude of each component, and calculate the fatigue life of each component of the pitch-adjusting propeller hub mechanism based on the original S-N curve of the component material;

[0036] Step 3: Data-driven surrogate model: According to the calculated finite fatigue life samples of each component, use the data-driven surrogate model to establish the relationship between the basic uncertain variables and the fatigue life, and introduce the interval non-probabilistic reliability model on the basis of this relationship, so as to establish the fatigue life interval of each component affected by the basic uncertain variables;

[0037] Step 4: Calculation of fatigue reliability: Use the fatigue life intervals of each component obtained in Step 3 to evaluate the fatigue reliability of each component by the non-probabilistic area method.

[0038] As Figure 2In the illustrated embodiment, consider a full-scale structure of a controllable pitch propeller hub mechanism, which includes: a hub body, propeller blades, blade root bolts, blade root pins, bearing rings, crank disks, piston rods, sliders, oil cylinders, oil cylinder bolts, shaft flanges, shaft flange bolts, and shaft flange pins. All materials are set with bilinear properties. All friction coefficients are set to 0.1. In engineering, usually 0.5 times the allowable tensile stress is used as the value of the allowable shear stress. Therefore, the allowable shear stress of all materials here is 0.5 times the corresponding yield limit. The main mechanical properties of the materials used in the calculation are shown in Table 1.

[0039] Table 1

[0040]

[0041]

[0042] Considering the uncertainties of parameters such as the elastic modulus, Poisson's ratio, and friction coefficient of materials, using parametric modeling technology, the rapid establishment of finite element models for different values of uncertain variables is realized. On this basis, a finite element simulation of the controllable pitch propeller hub mechanism is carried out. A node group is established on each propeller blade, and a point is taken on the central axis to capture the mpc. The mpc flexible constraint is established using the rbe3 command, and the thrust, shear force, thrust moment, shear force moment, and swivel blade moment fluctuating load spectra of the propeller blade are distributed to the node group of the propeller blade through this point to simulate the hydrodynamic fluctuating load during the operation of the propeller blade. According to the stress amplitude changes of each component caused by the working conditions of the propeller blade under 100% hydrodynamic conditions, the five-point method that consumes less computing resources is used to draw the fatigue load spectra of each part (3A blade root bolt, 3B shaft flange bolt, 3C oil cylinder bolt, 3D blade root pin, 3E shaft flange pin, 3F crank disk, 3G slider, 3H hub body) as shown Figures 3A - 3H Next, the S-N curves of each component are determined using engineering algorithms. Through years of experience accumulation, the empirical formulas between fatigue and tensile properties have been relatively mature (especially for steel). The S-N curves of components with different strengths can be plotted using the fatigue limit (i.e., the stress amplitude at 1×10 6 cycles) and (i.e., the stress amplitude at 1×10 3 cycles). The S-N curves of each component (4A crank disk, 4B slider, 4C blade root pin, 4D shaft flange pin, 4E blade root bolt, shaft flange bolt, oil cylinder bolt, 4F hub body) of the controllable pitch propeller mechanism are shown as Figures 4A - 4F shown.

[0043] Perform fatigue simulation by combining the fatigue load spectrum and the S-N curve. Since the parameters are uncertain, it is necessary to perform fatigue simulation multiple times to obtain the fatigue life samples of each component. Obtaining a large number of samples requires multiple finite element simulations, which is extremely time-consuming. Therefore, a data-driven surrogate model is used for training to obtain the relationship between the fatigue life and the basic uncertainty variables. The Kriging surrogate model, as a commonly used and efficient response interval calculation method, can be used for uncertainty propagation, such as Figures 5A - 5B shown. The Kriging surrogate model consists of two main parts: a fitting term and a correlation term. Figure 5A shows the fitting term of the Kriging surrogate model, which constructs the fatigue life curve through the function values and derivative values of the sample points. Among them, w (i) and λ (i) are the weights of the function values and derivative values of the sample points respectively. By simultaneously using the function value and derivative value information of the sample points, the fitting accuracy of the fatigue life curve is improved. The correlation term describes the correlation between sample points through the covariance function, enabling the Kriging surrogate model to not only achieve unbiased estimation at the sample points but also give the corresponding prediction error when predicting the fatigue life, as shown in Figure 5B . Based on the Kriging surrogate model, establish the relationship between the uncertain variables and the fatigue life, and give the corresponding prediction error.

[0044] In engineering systems, the quantification of uncertainty is usually achieved by probability methods, which require a large amount of statistical information on the distributions of various uncertainty parameters. Considering the difficulty of accurately evaluating uncertainty under actual conditions, it is a more feasible method to use non-probability methods to quantify uncertainty, which only requires the boundary information of the uncertainty parameters. Combine the Kriging surrogate model with the Monte Carlo method to obtain the fatigue life boundary information of each component and establish a fatigue life interval model.

[0045] In the interval non-probabilistic reliability model, the reliability index for evaluating the structural safety is defined by the boundary characteristics of the actual response of the structure and the allowable interval of the response. Let be the allowable value of the fatigue life of the components of the controllable pitch propeller mechanism, be the allowable fatigue life interval of the structure, φ s be the lower bound of the allowable fatigue life, be the upper bound of the allowable fatigue life; be the actual value of the fatigue life of the components of the controllable pitch propeller mechanism, be the actual fatigue life interval of the structure, φ a be the lower bound of the actual fatigue life, be the upper bound of the actual fatigue life; Then the safety state of the mechanism can be judged using the performance function in reliability theory:

[0046] M(φ a ,φ s )=φs -φ a ,

[0047] When the performance function M > 0, the controllable pitch propeller mechanism is considered to be in a safe state. When M < 0, the controllable pitch propeller mechanism is considered to be in a failure state. Thus, the structural failure plane can be defined.

[0048] M(φ a , φ s ) = φ s -φ a = 0,

[0049] Normalize the actual value and the allowable value of the structural response function:

[0050]

[0051] where and are the radii of the intervals to which φ s and φ a belong, respectively. and are the average values of the intervals to which φ s and φ a belong, respectively. Substitute δφ s and δφ a into M(φ a , φ s ) = 0 to obtain the normalized failure plane:

[0052]

[0053] Taking δφ a as the abscissa and δφ s as the ordinate, represent the ranges of δφ a and δφ s and the failure plane described by the above formula in the Cartesian coordinate system. Then, the region enclosed by δφ a and δφ s is divided by the failure plane into a safe region and a failure region, corresponding to M > 0 and M < 0, respectively. The reliability index R is defined as the ratio of the area of the structural safety domain to the area of the overall region. There are six different expressions according to the position of the failure plane:

[0054]

[0055] The above expressions correspond to Figure 6 six different cases. Use the non-probabilistic area method to evaluate the fatigue reliability of each component. Tables 2 and 3 show the fatigue life intervals (upper and lower bounds) and fatigue reliabilities of the components of the controllable pitch propeller hub mechanism, respectively.

[0056] Table 2

[0057]

[0058] Table 3

[0059]

[0060] As can be seen from Table 2 and Table 3, the lower bounds of the fatigue life of the crank disk and the hub body are both lower than 10^6 times. When the fatigue life of 1e6 times is used as the allowable lower bound of the fatigue life, the reliability levels of the two are 0.9632 and 0.7322 respectively. In addition, the reliability levels of the remaining components are all 1. This indicates that at 100% hydrodynamic conditions, the screw hole connection of the crank disk and the hub body is most likely to generate fatigue cracks first and cause fatigue fracture. The main reason is that both the blade and the crank disk are fixed to the hub body through root bolts. Under this condition, the bending moment borne by the blade will cause stress concentration at the bolt connection, and fatigue cracks will be generated under the action of long-term alternating loads.

[0061] The above are only the specific steps of the present invention and do not constitute any limitation to the protection scope of the present invention; it can be extended and applied to the field of static structure uncertainty analysis based on convex polyhedron models. Any technical solutions formed by equivalent transformation or equivalent substitution fall within the scope of the protection of the rights of the present invention.

Claims

1. A non-probabilistic fatigue reliability analysis method for the pitch-changing propeller hub mechanism driven by data, characterized in that, Specifically, it includes the following steps: 1) Finite element model establishment: Considering the uncertainties of the material elastic modulus, Poisson's ratio, friction coefficient and component mating clearance of each component of the controllable pitch propeller hub mechanism, parametric modeling is carried out for each component of the controllable pitch propeller hub mechanism to obtain a high-precision finite element model of each component of the controllable pitch propeller hub mechanism; 2) Fatigue simulation analysis: Apply the hydrodynamic fluctuation load during the simulated operation and rotation of the propeller blade to the assembled controllable pitch propeller hub mechanism. Under the hydrodynamic condition of the propeller blade, compile the fatigue load spectrum according to the stress amplitude of each component, and calculate the fatigue life of each component of the controllable pitch propeller hub mechanism based on the original S-N curve of the component material; 3) Data-driven surrogate model: According to the calculated finite fatigue life samples of each component, use the data-driven surrogate model to establish the relationship between the basic uncertainty variables and the fatigue life. On the basis of this relationship, introduce the interval non-probabilistic reliability model, so as to establish the fatigue life interval of each component under the influence of the basic uncertainty variables; 4) Fatigue reliability calculation: Use the fatigue life intervals of each component obtained in step three to evaluate the fatigue reliability of each component by the non-probabilistic area method.

2. The non-probabilistic fatigue reliability analysis method of the data-driven adjustable pitch propeller hub mechanism according to claim 1, characterized in that The surrogate model in step 3) selects the Kriging surrogate model. As a commonly used and efficient response interval calculation method, it is used for uncertainty propagation. The Kriging surrogate model consists of two main parts: the fitting term and the correlation term. The fitting term constructs the fatigue life curve through the function values and derivative values of the sample points, and the correlation term describes the correlation between the sample points through the covariance function, so that the Kriging surrogate model not only realizes the unbiased estimation at the sample points, but also gives the corresponding prediction error when predicting the fatigue life.

3. The non-probabilistic fatigue reliability analysis method of the data-driven adjustable pitch propeller hub mechanism according to claim 2, characterized in that, The interval non-probabilistic reliability model combines the Kriging surrogate model and the Monte Carlo method to obtain the fatigue life boundary information of each component.

4. The non-probabilistic fatigue reliability analysis method of the data-driven adjustable pitch propeller hub mechanism according to claim 3, wherein In the interval non-probabilistic reliability model, the reliability index for evaluating structural safety is defined by the boundary characteristics of the actual response of the structure and the allowable interval of the response, φ s is the allowable value of the fatigue life of the components of the controllable pitch propeller mechanism, φ a is the actual value of the fatigue life of the components of the controllable pitch propeller mechanism. Then, the safety state of the mechanism is judged using the performance function in reliability theory: M(φ a ,φ s )=φ s -φ a , When the performance function M > 0, it is considered that the controllable pitch propeller mechanism is in a safe state. When M < 0, it is considered that the controllable pitch propeller mechanism is in a failure state. Thus, the structural failure plane can be defined M(φ a ,φ s )=φ s -φ a =0, Normalize the actual value and the allowable value of the structural response function: where φ s R and φ a R are the radius of the interval to which φ s and φ a belong respectively, and φ s C and φ a C are the average values of the intervals to which φ s and φ a belong respectively; substituting δφ s and δφ a into M(φ a , φ s ) = 0 gives the standardized failure plane: With δφ a as the horizontal axis and δφ s as the vertical axis, the ranges of δφ a and δφ s and the failure plane described by the above formula are represented in the Cartesian coordinate system. Then, the region enclosed by δφ a and δφ s is divided into a safe region and a failure region by the failure plane, corresponding to M > 0 and M < 0 respectively, and is used to evaluate the fatigue reliability of each component.

5. A non-probabilistic fatigue reliability analysis method for an assembly structure, characterized in that, Taking each component of the assembled structure as the research object, considering the uncertainties of the material elastic modulus, Poisson's ratio, friction coefficient and component mating clearance of each component, obtain the finite element model under different values of uncertain variables through parametric part modeling, draw the fatigue load spectrum using the finite element simulation load results, and conduct fatigue simulation in combination with the S-N curve to obtain the fatigue life of each component; Under the condition of finite fatigue life samples, use the data-driven surrogate model and the Monte Carlo method to obtain the non-probabilistic interval model of the fatigue life of each component, and calculate the fatigue reliability of each component of the assembled structure by the non-probabilistic area method.

Images