Multi-stage fatigue full-life intelligent prediction method based on data driving

Abstract

The invention belongs to the field of measurement and testing, and relates to a multi-stage fatigue full-life intelligent prediction method based on data driving, which comprises the following steps: carrying out a fatigue test, a small crack propagation test and a long crack propagation test of a material to obtain corresponding test data; establishing a small crack propagation model and a long crack propagation model based on the crack closing effect; and obtaining the multi-stage fatigue prediction life of the material based on the fatigue test data of the material, the small crack propagation model and the long crack propagation model. According to the method, three stages of crack initiation, small crack propagation and long crack propagation are fused, full-life dynamic prediction from initial defects to fractures is realized, and stage differences of material fatigue behaviors are comprehensively reflected; the problems that in the prior art, the whole process from crack initiation to final fracture of the material is difficult to comprehensively describe, and prediction deviation and prediction precision are poor due to stage splitting are solved.

Description

Technical Field

[0001] The present invention relates to the field of measurement and testing technology, and in particular to a data-driven multi-stage fatigue life intelligent prediction method. Background Art

[0002] Traditional fatigue test life prediction methods are mainly divided into two categories: fatigue test life prediction methods based on stress-life (SN curve) and crack growth life prediction methods based on fracture mechanics. However, these two methods have their own limitations and cannot fully describe the entire process of the material from crack initiation to final fracture:

[0003] The SN curve method ignores the impact of the crack propagation stage and assumes that the fatigue life of a material under cyclic loading is primarily determined by the crack initiation stage. This method performs well under high-cycle fatigue conditions (i.e., low stress levels and high cycle numbers), but often suffers from significant deviations in predictions under low-cycle fatigue (i.e., high stress levels and low cycle numbers) or complex loading conditions. The limitations of the SN curve method primarily lie in its neglect of the crack propagation stage, its inability to reflect the impact of initial defects, and its limited applicability.

[0004] Fracture mechanics methods are primarily based on crack growth theory, describing the crack propagation behavior under cyclic loading to predict the full fatigue life. The most commonly used fracture mechanics model is the Paris law. The limitations of the Paris law lie in its neglect of the crack initiation stage, its failure to consider crack closure effects, and its dependence on initial crack size.

[0005] Although traditional fatigue life prediction methods and fracture mechanics methods have achieved certain success within their respective scopes of application, they are unable to fully describe the entire process of materials from crack initiation to final fracture. Summary of the Invention

[0006] In view of the above analysis and in response to the shortcomings of the existing technology, the present invention aims to provide a data-driven multi-stage fatigue full-life intelligent prediction method to solve at least one of the problems existing in the existing technology, namely, the difficulty in comprehensively describing the entire process of the material from crack initiation to final fracture, the prediction deviation caused by stage separation, and the poor prediction accuracy.

[0007] The purpose of the present invention is mainly achieved through the following technical solutions:

[0008] The present invention discloses a data-driven multi-stage fatigue life intelligent prediction method, comprising:

[0009] Carry out fatigue tests, small crack growth tests and long crack growth tests on materials to obtain corresponding test data;

[0010] Establish small crack growth model and long crack growth model based on crack closure effect;

[0011] The multi-stage fatigue predicted life of the material is obtained based on the fatigue test data of the material, the small crack growth model and the long crack growth model.

[0012] Preferably, the data-driven multi-stage fatigue lifecycle intelligent prediction method includes:

[0013] S1: Obtain the fatigue crack initiation model of the material based on the full life of fatigue test, small crack growth model and long crack growth model;

[0014] S2: Construct a multi-stage fatigue life intelligent prediction model for materials based on the small crack growth model, long crack growth model and fatigue crack initiation model of the material;

[0015] S3: Based on the full life of fatigue test, the multi-stage fatigue full life intelligent prediction model is adjusted to obtain the multi-stage fatigue full life intelligent prediction model with the best prediction accuracy;

[0016] S4: Obtain the material multi-stage fatigue predicted life based on the adjusted multi-stage fatigue full life intelligent prediction model.

[0017] Preferably, in step S1, the small crack growth model N short include:

[0018]

[0019] Where Δσ is the stress range, N is the stress, a is the crack length, a0 is the characteristic crack length of the small crack growth model modified by EI-Haddad, and C S is the material constant for small crack growth rate, m S is the index of small crack growth rate, ΔK eff is the stress intensity factor considering the crack closure effect, a short is the starting size of small crack extension, a t It is the termination size of small crack extension.

[0020] Preferably, in step S1, the long crack growth model N large include:

[0021]

[0022] Among them, the initial value of the termination size of the long crack extension a t , the termination size of long crack extension is a c , f is the crack closure number, R is the stress ratio, ΔK th is the threshold stress intensity factor range, K c is the fracture toughness, CL is the material constant for the long crack growth rate, m L is the index of the long crack growth rate, p is the parameter describing the effect of crack length on the growth rate, and q is the parameter describing the effect of crack length on the growth rate when it approaches the critical length; K max is the maximum stress intensity factor.

[0023] Preferably, the fatigue crack initiation model in step S1 satisfies: N initial =N f -N short -N large .

[0024] Preferably, the data-driven multi-stage fatigue life intelligent prediction method further includes: initial =N f -N short -N large The crack initiation life expression is obtained by transforming the traditional stress-life curve:

[0025]

[0026] Among them, σ f ' and b are the material parameters of the SN curve equation, N initial becomes is a function of the independent variable.

[0027] Preferably, in step S2, a multi-stage fatigue life intelligent prediction model of the material is constructed. satisfy:

[0028]

[0029] Preferably, the adjustment of the multi-stage fatigue full life intelligent prediction model based on the fatigue test full life in step S3 includes optimizing the multi-stage fatigue full life intelligent prediction model by adopting the least square method.

[0030] Preferably, in step S3, adjusting the multi-stage fatigue full life intelligent prediction model based on the fatigue test full life includes:

[0031] S301: The starting size of small crack growth in the material a sho , small crack extension termination size a t , the starting size of long crack extension a t and the end size of the long crack extension a c Given an initial value, the initial value of the predicted life in the small crack growth stage is obtained by the small crack growth model, and the initial value of the predicted life in the long crack growth stage is obtained by the long crack growth model;

[0032] S302: Based on Obtain σ in fatigue crack initiation model f ' and the initial value of b, obtain the initial value of the predicted life in the fatigue crack initiation stage based on the fatigue crack initiation model;

[0033] S303: Obtaining an initial value of the material's multi-stage fatigue predicted life from the initial value of the predicted life in the small crack growth stage, the initial value of the predicted life in the long crack growth stage obtained by the long crack growth model, and the initial value of the predicted life in the fatigue crack initiation stage;

[0034] S304: Based on the above steps, different parameters are assigned values to obtain the predicted values of the multi-stage fatigue predicted life of multiple groups of materials. The relative error between the multi-stage fatigue predicted life of the material and the full life of the fatigue test is used as the overall optimization goal. The relative error is set as tolerance, which satisfies:

[0035]

[0036] in, It is a multi-stage fatigue life intelligent prediction model. Fatigue test model.

[0037] Preferably, tolerance=10%.

[0038] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0039] (1) The multi-stage fatigue full-life intelligent prediction method of the present invention integrates the three stages of crack initiation, small crack propagation, and long crack propagation, achieving full-life dynamic prediction from initial defects to fracture, comprehensively reflecting the stage differences in material fatigue behavior, solving the prediction deviation problem caused by stage separation in traditional methods, and significantly improving the accuracy of life prediction. The deviation between the predicted life at 700°C and the measured value mean at a maximum stress of 300MPa is only about 4%; the deviation between the predicted life at 700°C and the measured value mean at a maximum stress of 350MPa is only about 8%; the deviation between the predicted life at 700°C and the measured value mean at a maximum stress of 400MPa is only about 10%; the deviation between the predicted life at 700°C and the measured value mean at a maximum stress of 450MPa is only about 2.4%; and the deviation between the predicted life at 700°C and the measured value mean at a maximum stress of 500MPa is about 37.5%.

[0040] (2) The present invention comprehensively considers the crack closure effect and introduces the crack closure effect in the small crack and long crack propagation stages, which truly reflects the crack propagation behavior under complex loads and significantly improves the prediction accuracy under high stress ratio and variable amplitude loads, making up for the deficiency of the traditional method that ignores the crack closure effect and causes prediction deviation.

[0041] (3) The present invention obtains the fatigue crack initiation model and fatigue crack initiation life by inversely calculating the life of small crack and long crack propagation stages, which truly reflects the crack propagation behavior under complex loads and significantly improves the prediction accuracy.

[0042] (4) The present invention dynamically adjusts the starting size of small crack extension and the starting size of long crack extension through iterative optimization using the least squares method, and combines this with the inverse method of obtaining the fatigue crack initiation life, thereby realizing the adaptive update of the material parameters of the SN curve equation, effectively overcoming the traditional model's reliance on the assumption of a fixed crack extension size, and improving the model's prediction accuracy and applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like parts throughout the drawings.

[0044] Figure 1 It is the overall flow chart of the present invention.

[0045] Figure 2 This is the load spectrum of the high temperature fatigue test of the present invention.

[0046] Figure 3 This is a comparison between the multi-stage fatigue full life intelligent prediction life results and the fatigue test life results in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the technical problems, technical solutions and beneficial effects solved by the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0048] About technical terms:

[0049] The crack closure effect refers to the phenomenon that the crack tip is closed relative to the matrix due to the existence of the stress field on the crack surface near the crack tip, thereby hindering the crack propagation.

[0050] Unless otherwise specified, fatigue life in the present invention refers to the full life of fatigue test including crack initiation, small crack growth and long crack growth.

[0051] Unless otherwise specified, the fatigue test of the material in the present invention refers to the fatigue test of the material, which is a test of measuring the fatigue of the material at a specific temperature under a specific stress change range, a specific change frequency, and a specific stress ratio with reference to GB / T 3075-2021 "Axial force control method for fatigue test of metallic materials" to obtain the corresponding fatigue test full life test.

[0052] The applicant's research found that traditional methods usually only focus on a single stage of crack initiation or crack propagation, and cannot fully reflect the fatigue behavior of the material: the crack closure effect has a significant impact on the crack propagation rate under high stress ratios or complex load conditions, and traditional methods usually ignore this effect; traditional methods usually assume that the initial crack size is a fixed value, while the randomness and dispersion of initial defects in actual materials will lead to uncertainty in the prediction results; traditional methods usually rely on empirical formulas or theoretical assumptions, and lack full utilization of experimental data, resulting in limited prediction accuracy.

[0053] In response to the above problems, the present invention discloses a data-driven multi-stage fatigue lifecycle intelligent prediction method, comprising:

[0054] Carry out fatigue tests, small crack growth tests and long crack growth tests on materials to obtain corresponding test data;

[0055] Establish small crack growth model and long crack growth model based on crack closure effect;

[0056] The multi-stage fatigue predicted life of the material is obtained based on the fatigue test data of the material, the small crack growth model and the long crack growth model.

[0057] When implemented, the fatigue test of the material can obtain the fatigue test full life, and the fatigue full life intelligent prediction method includes:

[0058] S1: Obtain the fatigue crack initiation model of the material based on the full life of fatigue test, small crack growth model and long crack growth model;

[0059] S2: Construct a multi-stage fatigue life intelligent prediction model for materials based on the small crack growth model, long crack growth model and fatigue crack initiation model of the material;

[0060] S3: Based on the full life of fatigue test, the multi-stage fatigue full life intelligent prediction model is adjusted to obtain the multi-stage fatigue full life intelligent prediction model with the best prediction accuracy;

[0061] S4: Obtain the material multi-stage fatigue predicted life based on the adjusted multi-stage fatigue full life intelligent prediction model.

[0062] Compared with the existing technology, the multi-stage fatigue full-life intelligent prediction method of the present invention integrates the three stages of crack initiation, small crack extension, and long crack extension, realizing full-life dynamic prediction from initial defects to fracture, comprehensively reflecting the stage differences in material fatigue behavior, solving the prediction deviation problem caused by stage separation in traditional methods, and significantly improving the life prediction accuracy.

[0063] Specifically, the fatigue test of the material is carried out in accordance with GB / T 3075-2021 "Axial force control method for fatigue test of metal materials" to measure the fatigue test of the material at a specific temperature under a specific stress change range, a specific change frequency, and a specific stress ratio to obtain the corresponding fatigue test full life.

[0064] Specifically, the small crack growth model in step S1 includes:

[0065]

[0066] Where Δσ is the stress range, N is the stress, a is the crack length, a0 is the characteristic crack length of the small crack growth model modified by EI-Haddad, which can be obtained by looking up the table in the prior art, C S is the material constant for small crack growth rate, m S is the index of small crack growth rate; ΔK eff In order to consider the stress intensity factor of crack closure effect, a0 is introduced; a short is the starting size of small crack extension; a t It is the termination size of small crack extension.

[0067] It should be noted that the small crack growth model includes the stress intensity factor ΔK which takes into account the crack closure effect. eff , thus being able to truly reflect the crack propagation behavior under complex loads and significantly improve the prediction accuracy under high stress ratios and variable amplitude loads.

[0068] Specifically, the small crack growth test is to obtain multiple groups of ΔK eff . ;

[0069] Specifically, the long crack growth model in step S1 includes:

[0070]

[0071] Among them, the initial value of the termination size of the long crack extension a t , the termination size of long crack extension is a c , f is the crack closure number, R is the stress ratio, ΔK th is the threshold stress intensity factor range, K c is the fracture toughness, C L is the material constant for the long crack growth rate, m L is the index of the long crack growth rate, p is the parameter describing the effect of crack length on the growth rate, and q is the parameter describing the effect of crack length on the growth rate when it approaches the critical length; K max is the maximum stress intensity factor, which can be calculated by the following formula: Kmax =ΔK / (1-R).

[0072] Specifically, the long crack growth test is to obtain multiple groups of fatigue crack growth methods according to GB / T 6398-2017 "Metallic materials fatigue test fatigue crack growth method" ΔK and K c , K max , ΔK th .

[0073] It should be noted that the small crack growth model includes the crack closure number f, and thus can truly reflect the crack growth behavior under complex loads and significantly improve the prediction accuracy under high stress ratios and variable amplitude loads.

[0074] Compared with the existing technology, the present invention comprehensively considers the crack closure effect, introduces the crack closure effect in the small crack and long crack propagation stage, truly reflects the crack propagation behavior under complex loads, significantly improves the prediction accuracy under high stress ratio and variable amplitude load, and makes up for the deficiency of traditional methods ignoring the crack closure effect and causing prediction deviation; the present invention has a deviation of only about 4% between the predicted life at 700°C and the average measured value when the maximum stress is 300MPa; the deviation of only about 8% between the predicted life at 700°C and the average measured value when the maximum stress is 350MPa; the deviation of only about 10% between the predicted life at 700°C and the average measured value when the maximum stress is 400MPa; the deviation of only about 2.4% between the predicted life at 700°C and the average measured value when the maximum stress is 450MPa; the deviation of only about 37.5% between the predicted life at 700°C and the average measured value when the maximum stress is 500MPa.

[0075] Specifically, the fatigue crack initiation model in step S1 satisfies: N initial =N f -N short -N large .

[0076] Among them, N initial is the fatigue crack initiation model, N f It is the full life of the fatigue test and a fixed value under specific measurement conditions at a specific temperature.

[0077] Furthermore, N initial =N f -N short -N large The crack initiation life expression is obtained by transforming the traditional stress-life curve (SN curve):

[0078]

[0079] Among them, σ f ' and b are the material parameters of the SN curve equation, N initial becomes is a function of the independent variable.

[0080] Specifically, in step S2, a multi-stage fatigue life intelligent prediction model of the material is constructed. satisfy:

[0081]

[0082] Specifically, in step S3, the adjustment of the multi-stage fatigue full life intelligent prediction model based on the fatigue test full life includes adopting the least squares method for optimization.

[0083] Specifically, the adjustment of the multi-stage fatigue full life intelligent prediction model based on the fatigue test full life in step S3 includes:

[0084] S301: The starting size of small crack growth in the material a short , small crack extension termination size a t , the starting size of long crack extension a t and the end size of the long crack extension a c Given an initial value, the initial value of the predicted life in the small crack growth stage is obtained by the small crack growth model, and the initial value of the predicted life in the long crack growth stage is obtained by the long crack growth model;

[0085] S302: Based on Obtain σ in fatigue crack initiation model f ' and the initial value of b, obtain the initial value of the predicted life in the fatigue crack initiation stage based on the fatigue crack initiation model;

[0086] S303: Obtaining an initial value of the material's multi-stage fatigue predicted life from the initial value of the predicted life in the small crack growth stage, the initial value of the predicted life in the long crack growth stage obtained by the long crack growth model, and the initial value of the predicted life in the fatigue crack initiation stage;

[0087] S304: Based on the above steps, different parameters are assigned values to obtain the predicted values of the multi-stage fatigue predicted life of multiple groups of materials. The relative error between the multi-stage fatigue predicted life of the material and the full life of the fatigue test is used as the overall optimization goal. The relative error is set as tolerance, which satisfies:

[0088]

[0089] in, It is a multi-stage fatigue life intelligent prediction model. Fatigue test model.

[0090] It should be noted that the fatigue test process variable is Δσ, so it can also be regarded as is a function of the dependent variable.

[0091] Preferably, the tolerance can be selected as 10%.

[0092] Compared with the existing technology, the present invention obtains the fatigue crack initiation model and fatigue crack initiation life by inversely calculating the life of small crack and long crack propagation stages, which truly reflects the crack propagation behavior under complex loads and significantly improves the prediction accuracy.

[0093] Compared with the existing technology, the present invention dynamically adjusts the starting size of small crack extension and the starting size of long crack extension through iterative optimization using the least squares method, and combines it with the inverse method to obtain the fatigue crack initiation life, thereby realizing the adaptive update of the material parameters of the SN curve equation, effectively overcoming the traditional model's reliance on the assumption of a fixed crack extension size, and improving the model's prediction accuracy and applicability.

[0094] In order to better illustrate the present invention, the following embodiments are further provided:

[0095] Example

[0096] This embodiment discloses a data-driven multi-stage fatigue life intelligent prediction method, using a typical nickel-based high-temperature alloy, including:

[0097] S1: Obtain the fatigue crack initiation model of the material based on the full life of the fatigue test, the small crack growth model and the long crack growth model.

[0098] The fatigue test of the material is carried out in accordance with GB / T 3075-2021 "Method for controlling axial force in fatigue test of metallic materials" and the fatigue test of the material at 700°C is carried out at the maximum stress of 300MPa, 350MPa, 400MPa, 450MPa and 550MPa, the test frequency is 5Hz and the stress ratio R is 0.1. Figure 2 The high temperature fatigue test load spectrum is shown in Table 1 to obtain the corresponding fatigue test full life.

[0099] Table 1 Fatigue test control parameters and test results of nickel-based superalloy at 700℃

[0100]

[0101] The small crack growth model in step S1 includes:

[0102]

[0103] Where Δσ is the stress range, N is the stress, a is the crack length, a0 is the characteristic crack length of the small crack growth model modified by EI-Haddad, which can be obtained by looking up the table in the prior art, C S is the material constant for the small crack growth rate = 2.12e -13 , mS is the exponent of small crack growth rate = 1.86; ΔK eff In order to consider the stress intensity factor of crack closure effect, a0 is introduced; a sho is the starting size of small crack extension; a t It is the termination size of small crack extension.

[0104] The small crack growth test is carried out by referring to HB 7705-2001 "Test method for small crack growth rate of fatigue metal materials" to obtain multiple groups of ΔK eff , as shown in Table 2:

[0105] Table 2 Test results of small crack growth test on nickel-based superalloy at 700℃

[0106]

[0107]

[0108] The long crack growth model in step S1 includes:

[0109]

[0110] Among them, the initial value of the termination size of the long crack extension a t , the termination size of long crack extension is a c , f is the crack closure number, R is the stress ratio, ΔK th is the threshold stress intensity factor range, K c is the fracture toughness, C L is the material constant for the long crack growth rate = 3.07×10 -12 , m L =2.23, p=1.01, q=0.93, f=0.21, K max is the maximum stress intensity factor, which can be calculated by the following formula: K max =ΔK / (1-R).

[0111] The long crack growth test is based on GB / T 6398-2017 "Metallic materials fatigue test fatigue crack growth method" to obtain multiple groups of ΔK and K c =109.35MPa·m^1 / 2, K max =ΔK / (1-R),ΔK th =1.86MPa·m^1 / 2, the results are shown in Table 3:

[0112] Table 3 Test results of long crack growth test on nickel-based superalloy at 700℃

[0113]

[0114]

[0115] The fatigue crack initiation model in step S1 satisfies: N initial =N f -N shor -N large .

[0116] Among them, N initial is the fatigue crack initiation model, N f It is the full life of the fatigue test and a fixed value under specific measurement conditions at a specific temperature.

[0117] Furthermore, N initial =N f -N short -N large The crack initiation life expression is obtained by transforming the traditional stress-life curve (SN curve):

[0118]

[0119] Among them, σ f ' = 352 MPa and b = -0.1, N initial becomes is a function of the independent variable.

[0120] S2: Based on the small crack growth model, long crack growth model and fatigue crack initiation model of the material, a multi-stage fatigue life intelligent prediction model is constructed. satisfy:

[0121]

[0122] S3: Based on the full life of fatigue test, the multi-stage fatigue full life intelligent prediction model is adjusted to obtain the multi-stage fatigue full life intelligent prediction model with the best prediction accuracy.

[0123] In step S3, the adjustment of the multi-stage fatigue full life intelligent prediction model based on the fatigue test full life includes:

[0124] S301: The starting size of small crack growth in the material a short Given an initial value of 0.02 mm, the small crack extension termination size a t Given an initial value of 1 mm, the initial size of the long crack extension is a t Given an initial value of 1 mm, the long crack extension termination size a c Given an initial value of 2.89 mm, the initial value of the predicted life in the small crack growth stage is obtained by the small crack growth model;

[0125] S302: Based on Obtain σ in fatigue crack initiation model f 'With the initial value of b, the initial σ f ' = 352 MPa, b = -0.1, the initial value of the predicted life in the fatigue crack initiation stage is obtained based on the fatigue crack initiation model;

[0126] S303: Obtaining an initial value of the material's multi-stage fatigue predicted life from the initial value of the predicted life in the small crack growth stage, the initial value of the predicted life in the long crack growth stage obtained by the long crack growth model, and the initial value of the predicted life in the fatigue crack initiation stage;

[0127] S304: The relative error between the material's multi-stage fatigue prediction life and the fatigue test life is used as the overall optimization goal. The relative error is set to tolerance = 10%, and the least squares method is used for optimization.

[0128]

[0129] The results are shown in Table 4:

[0130] Table 4 Parameters of the adjusted multi-stage fatigue life intelligent prediction model

[0131]

[0132] S4: Based on the adjusted multi-stage fatigue full life intelligent prediction model, the multi-stage fatigue predicted life of the material is obtained. The results are shown in Table 5:

[0133] Table 5 Fatigue life test results and multi-stage fatigue life prediction results under different stress amplitudes

[0134]

[0135] From the above, it can be seen that when the maximum stress is 300MPa, the deviation between the predicted life at 700℃ and the average measured value is only about 4%; when the maximum stress is 350MPa, the deviation between the predicted life at 700℃ and the average measured value is only about 8%; when the maximum stress is 400MPa, the deviation between the predicted life at 700℃ and the average measured value is only about 10%; when the maximum stress is 450MPa, the deviation between the predicted life at 700℃ and the average measured value is only about 2.4%; when the maximum stress is 500MPa, the deviation between the predicted life at 700℃ and the average measured value is about 37.5%.

[0136] also, Figure 3A comparison chart comparing the predicted lifespan obtained using the data-driven, multi-stage fatigue lifespan intelligent prediction method with the fatigue lifespan of the fatigue test is presented. The chart shows that the predicted lifespans obtained using the multi-stage fatigue lifespan intelligent prediction method all fall within a dispersion band of ±1.4 times the fatigue lifespan results of the fatigue test. This demonstrates that the proposed data-driven, multi-stage fatigue lifespan intelligent prediction method has high prediction accuracy and can effectively meet the needs of fatigue lifespan prediction and assessment for high-temperature structural materials used in aircraft engines.

[0137] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with this technical field within the technical scope disclosed by the present invention should be covered by the scope of protection of the present invention.

Claims

1. A data-driven multi-stage fatigue life intelligent prediction method, characterized in that: include: Carry out fatigue tests, small crack growth tests and long crack growth tests on materials to obtain corresponding test data; Establish small crack growth model and long crack growth model based on crack closure effect; The multi-stage fatigue predicted life of the material is obtained based on the fatigue test data of the material, the small crack growth model and the long crack growth model.

2. The data-driven multi-stage fatigue lifecycle intelligent prediction method according to claim 1 is characterized in that: The data-driven multi-stage fatigue life intelligent prediction method includes: S1: Obtain the fatigue crack initiation model of the material based on the full life of fatigue test, small crack growth model and long crack growth model; S2: Construct a multi-stage fatigue life intelligent prediction model for materials based on the small crack growth model, long crack growth model and fatigue crack initiation model of the material; S3: Based on the full life of fatigue test, the multi-stage fatigue full life intelligent prediction model is adjusted to obtain the multi-stage fatigue full life intelligent prediction model with the best prediction accuracy; S4: Obtain the material multi-stage fatigue predicted life based on the adjusted multi-stage fatigue full life intelligent prediction model.

3. The data-driven multi-stage fatigue lifecycle intelligent prediction method according to claim 2 is characterized in that: Small crack growth model N in step S1 short include: Where Δσ is the stress range, N is the stress, a is the crack length, a0 is the characteristic crack length of the small crack growth model modified by EI-Haddad, and C S is the material constant for small crack growth rate, m S is the index of small crack growth rate, ΔK eff is the stress intensity factor considering the crack closure effect, a short is the starting size of small crack extension, a t It is the termination size of small crack extension.

4. The data-driven multi-stage fatigue lifecycle intelligent prediction method according to claim 3 is characterized in that: Long crack growth model N in step S1 large include: Among them, the initial value of the termination size of the long crack extension a t , the termination size of long crack extension is a c , f is the crack closure number, R is the stress ratio, ΔK th is the threshold stress intensity factor range, K c is the fracture toughness, C L is the material constant for the long crack growth rate, m L is the index of the long crack growth rate, p is the parameter describing the effect of crack length on the growth rate, and q is the parameter describing the effect of crack length on the growth rate when it approaches the critical length; K max is the maximum stress intensity factor.

5. The data-driven multi-stage fatigue lifecycle intelligent prediction method according to claim 4 is characterized in that: The fatigue crack initiation model in step S1 satisfies: N initial =N f -N short -N iarge , where N initial is the fatigue crack initiation model, N f It is the full life of the fatigue test and a fixed value under specific measurement conditions at a specific temperature.

6. The data-driven multi-stage fatigue lifecycle intelligent prediction method according to claim 5 is characterized in that: The data-driven multi-stage fatigue life intelligent prediction method also includes initial =N f -N short -N large The crack initiation life expression is obtained by transforming the traditional stress-life curve: Among them, σ f ‘ and b are the material parameters of the SN curve equation, N initial becomes is a function of the independent variable.

7. The data-driven multi-stage fatigue lifecycle intelligent prediction method according to claim 6 is characterized in that: In step S2, a multi-stage fatigue life intelligent prediction model of the material is constructed. satisfy:

8. The data-driven multi-stage fatigue lifecycle intelligent prediction method according to claim 7 is characterized in that: In step S3, adjusting the multi-stage fatigue full life intelligent prediction model based on the fatigue test full life includes optimizing the multi-stage fatigue full life intelligent prediction model using the least squares method.

9. The data-driven multi-stage fatigue lifecycle intelligent prediction method according to claim 8 is characterized in that: In step S3, the adjustment of the multi-stage fatigue full life intelligent prediction model based on the fatigue test full life includes: S301: The starting size of small crack growth in the material a short , small crack extension termination size a t , the starting size of long crack extension a t and the end size of the long crack extension a c Given an initial value, the initial value of the predicted life in the small crack growth stage is obtained by the small crack growth model, and the initial value of the predicted life in the long crack growth stage is obtained by the long crack growth model; S302: Based on Obtain σ in fatigue crack initiation model f ‘ and the initial value of b, the initial value of the predicted life in the fatigue crack initiation stage is obtained based on the fatigue crack initiation model; S303: Obtaining an initial value of the material's multi-stage fatigue predicted life from the initial value of the predicted life in the small crack growth stage, the initial value of the predicted life in the long crack growth stage obtained by the long crack growth model, and the initial value of the predicted life in the fatigue crack initiation stage; S304: Based on the above steps, different parameters are assigned values to obtain the predicted values of the multi-stage fatigue predicted life of multiple groups of materials. The relative error between the multi-stage fatigue predicted life of the material and the full life of the fatigue test is used as the overall optimization goal. The relative error is set as tolerance, which satisfies: in, It is a multi-stage fatigue life intelligent prediction model. Fatigue test model.

10. The data-driven multi-stage fatigue lifecycle intelligent prediction method according to claim 9 is characterized in that: tolerance=10%.

Images