Composite material fatigue temperature rise prediction method and device combined with machine learning algorithm

Abstract

The invention belongs to the technical field of helicopter composite material fatigue loading temperature rise prediction, and relates to a composite material fatigue temperature rise prediction method and device combined with a machine learning algorithm. The method comprises the following steps: taking a fiber direction composite material laminated plate as a research object, and utilizing a random forest regression algorithm to realize accurate prediction of the surface temperature rise of the composite material under different loading frequencies and different stress levels.

Description

Technical Field

[0001] The present invention belongs to the technical field of fatigue life prediction of helicopter composite materials, and relates to a composite material fatigue temperature rise prediction method and device combined with a machine learning algorithm. Background Art

[0002] Bearingless rotor technology is one of the most advanced core rotor technologies currently. Fiber-reinforced composite structures have higher specific strength and specific stiffness, making them more advantageous when applied to bearingless rotors. However, due to the helicopter's unique hovering, side flight and other flight modes, the flexible beam structure is exposed to a complex service environment of vibration and alternating loads. At the same time, due to its relatively low thermal conductivity, the composite flexible beam structure will experience a significant temperature rise when subjected to fatigue or vibration loads. The thermal stress caused by the temperature increase of the structure, the degradation of the composite material's stiffness / strength, and the mismatch of the thermal expansion coefficients of adjacent structures will accelerate the damage and degradation of the structure, seriously endangering the safety of the structure. Summary of the Invention

[0003] Purpose of the invention: To provide a method and device for predicting fatigue temperature rise of composite materials combined with a machine learning algorithm, integrate test data and simulation results, accurately predict the temperature rise phenomenon of composite laminates at different frequencies and stress levels, and improve the safety of the rotor flexible beam structure.

[0004] Technical solution:

[0005] In a first aspect, a composite material fatigue temperature rise prediction method combined with a machine learning algorithm is provided, comprising:

[0006] Taking composite laminates as the research object, the random forest algorithm is used to predict the temperature rise of composite laminates under different frequencies and stress levels.

[0007] Furthermore, taking composite laminates as the research object, the random forest algorithm is used to predict the temperature rise of composite laminates at different frequencies and stress levels, including:

[0008] Use experiments to obtain fatigue temperature rise curves at different frequencies and stress levels;

[0009] The fatigue temperature rise data is used as the training set for the random forest algorithm; different frequencies, stress levels, and fatigue cycle numbers are used as training set inputs, and the fatigue temperature rise is used as the training set output;

[0010] The random forest algorithm was trained using the training set to obtain an algorithm for predicting fatigue temperature rise of composite materials;

[0011] Input the target frequency, target stress level and different fatigue cycle numbers into the algorithm for predicting fatigue temperature rise of composite materials, and obtain the fatigue temperature rise corresponding to different fatigue cycle numbers at the target stress level and target frequency;

[0012] According to the fatigue temperature rise corresponding to different fatigue cycle numbers at the target stress level and target frequency, the parameters of the fatigue temperature rise prediction model at the target stress level and target frequency are calculated, thereby obtaining the fatigue temperature rise prediction model at the target stress level and target frequency.

[0013] Furthermore, the fatigue temperature rise prediction model within a single fatigue cycle is:

[0014] Energy generated during one loading cycle:

[0015]

[0016] Among them, E sc is the generation energy due to the viscoelastic properties of the composite material, ε0 is the amplitude of the structural response strain, σ0 is the amplitude of the structural response stress, and δ is the hysteresis phase.

[0017] Dissipated energy in one loading cycle:

[0018]

[0019] Among them, E hs is the dissipated energy, the energy dissipated by heat conduction E cd and the energy dissipated by heat convection E dl composition, k is the thermal conductivity of each layer of the composite laminate, A is the surface area of ​​the test piece in contact with the air, represents the temperature gradient in the longitudinal and transverse directions of the test specimen structure, h is the heat convection coefficient between the test specimen and the air, T s is the surface temperature of the test piece, T ∞ is the ambient temperature.

[0020] The temperature rise of the composite material specimen during one loading cycle is:

[0021]

[0022] where ΔT is the fatigue temperature rise during one loading cycle, ρ is the density of the material, V is the volume of the composite structure, and c is the specific heat capacity of the material.

[0023] Under a certain number of cycles of loading, the fatigue temperature rise of the composite material test piece is:

[0024]

[0025] Where T is the fatigue temperature rise and n is the number of cyclic loading.

[0026] Furthermore, the stress level includes stress ratio and maximum fatigue stress.

[0027] In a second aspect, a composite material fatigue temperature rise prediction device combined with a machine learning algorithm is provided, characterized in that it includes:

[0028] The prediction module is used to predict the temperature rise of composite laminates under different frequencies and stress levels using the random forest algorithm.

[0029] Furthermore, the prediction module is specifically used to:

[0030] Use experiments to obtain fatigue temperature rise curves at different frequencies and stress levels;

[0031] The fatigue temperature rise data is used as the training set for the random forest algorithm; different frequencies, stress levels, and fatigue cycle numbers are used as training set inputs, and the fatigue temperature rise is used as the training set output;

[0032] The random forest algorithm was trained using the training set to obtain an algorithm for predicting fatigue temperature rise of composite materials;

[0033] Input the target frequency, target stress level and different fatigue cycle numbers into the algorithm for predicting fatigue temperature rise of composite materials, and obtain the fatigue temperature rise corresponding to different fatigue cycle numbers at the target stress level and target frequency;

[0034] According to the fatigue temperature rise corresponding to different fatigue cycle numbers at the target stress level and target frequency, the parameters of the fatigue temperature rise prediction model at the target stress level and target frequency are calculated, thereby obtaining the fatigue temperature rise prediction model at the target stress level and target frequency.

[0035] Furthermore, the fatigue temperature rise prediction model within a single fatigue cycle is:

[0036] Energy generated during one loading cycle:

[0037]

[0038] Among them, E sc is the generation energy due to the viscoelastic properties of the composite material, ε0 is the amplitude of the structural response strain, σ0 is the amplitude of the structural response stress, and δ is the hysteresis phase.

[0039] Dissipated energy in one loading cycle:

[0040]

[0041] Among them, E hs is the dissipated energy, the energy dissipated by heat conduction Ecd and the energy dissipated by heat convection E dl composition, k is the thermal conductivity of each layer of the composite laminate, A is the surface area of ​​the test piece in contact with the air, represents the temperature gradient in the longitudinal and transverse directions of the test specimen structure, h is the heat convection coefficient between the test specimen and the air, T s is the surface temperature of the test piece, T ∞ is the ambient temperature.

[0042] The temperature rise of the composite material specimen during one loading cycle is:

[0043]

[0044] where ΔT is the fatigue temperature rise during one loading cycle, ρ is the density of the material, V is the volume of the composite structure, and c is the specific heat capacity of the material.

[0045] Under a certain number of cycles of loading, the fatigue temperature rise of the composite material test piece is:

[0046]

[0047] Where T is the fatigue temperature rise and n is the number of cyclic loading.

[0048] Furthermore, the stress level includes stress ratio and maximum fatigue stress.

[0049] Beneficial effects:

[0050] This patent aims to develop a composite material fatigue temperature rise prediction method and device combined with machine learning to accurately predict the surface temperature rise law of composite material structures under different loading frequencies and different stress levels. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is the fatigue temperature rise curve of the composite material under different stress levels and loading frequency of 10Hz.

[0052] Figure 2 This is the fatigue temperature rise curve of the composite material under different loading frequencies and stress level 60%.

[0053] Figure 3 is the surface temperature of the composite material that changes with the number of cyclic loadings.

[0054] Figure 4 This is the fatigue temperature rise curve of the composite material under different stress levels and loading frequency of 10Hz predicted by the generated dissipated energy model.

[0055] Figure 5 This is the fatigue temperature rise curve of the composite material under different loading frequencies and stress level 60% predicted by the generated dissipated energy model.

[0056] Figure 6 Correlation analysis diagram of fatigue temperature rise of composite materials under different stress levels predicted by random forest algorithm.

[0057] Figure 7 Correlation analysis diagram of fatigue temperature rise of composite materials under different loading frequencies predicted by random forest algorithm.

[0058] Figure 8 The envelope of fatigue temperature rise curves of composite laminates at different stress levels predicted by the random forest algorithm.

[0059] Figure 9 The envelope of fatigue temperature rise curves of composite laminates at different loading frequencies predicted by the random forest algorithm. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solutions and advantages of the implementation of this application clearer, the technical solutions in the implementation of this application will be described in more detail below in conjunction with the drawings in the implementation of this application. In the drawings, the same or similar numbers throughout represent the same or similar elements or elements with the same or similar functions. The described implementation is a part of the implementation of this application, not all of the implementations. The implementation described below with reference to the drawings is exemplary and is intended to be used to explain this application, and should not be understood as a limitation on this application. Based on the implementation in this application, all other implementations obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. The implementation of this application is described in detail below in conjunction with the drawings.

[0061] In the description of the present invention, it should be understood that the terms "center", "axial", "vertical", "up", "down", "upper end", "bottom end", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the scope of protection of the present invention.

[0062] This paper provides a method for predicting fatigue temperature rise of composite materials combined with machine learning. Taking composite laminates as the research object, the random forest algorithm is used to accurately predict the surface temperature rise of composite materials under different stress levels and loading frequencies. The specific steps are as follows:

[0063] [1] proposed a fatigue temperature rise prediction model

[0064] Taking composite laminates as the research object and tension-tension fatigue tests of composite materials under different stress levels and loading frequencies as the model object, a surface fatigue temperature rise model of composite materials is proposed.

[0065] [2] Random Forest Algorithm

[0066] In order to further efficiently and accurately predict the temperature changes on the surface of composite laminates, combined with machine learning, the random forest algorithm was used to achieve accurate prediction of the surface temperature rise of composite fiber direction tension-tension fatigue under different stress levels and different loading frequencies.

[0067] [3] Conduct fatigue tests

[0068] Conduct tension-tension fatigue tests on the fiber direction of composite materials at different stress levels and loading frequencies, obtain surface temperature data of the test pieces, and record the surface temperature of the test pieces.

[0069] [4] Fatigue temperature rise prediction combined with fatigue temperature rise prediction model

[0070] Based on the fatigue temperature rise prediction model and test data, more surface fatigue temperature rise curves of test pieces under different stress levels and different loading frequencies are predicted.

[0071] [5] Fatigue temperature rise prediction based on machine learning

[0072] The predicted fatigue temperature rise curve is input into the fatigue temperature rise model combined with machine learning to achieve accurate prediction of the fatigue temperature rise of composite structures under different stress levels and different loading frequencies.

[0073] The differences between [3][4][5] are: [3] only conducted a small number of experiments, such as 10Hz and 27Hz; [4] predicted a part of the data based on the model, such as 15Hz, 20Hz, 25Hz, and 30Hz; [5] combined with machine learning, further learned and predicted more curves, such as 10.1Hz, 10.2Hz, 10.3Hz, etc.

[0074] The present invention is further described in detail below with reference to examples.

[0075] [1] proposed a fatigue temperature rise prediction model

[0076] Taking composite laminates as the research object and composite tension-tension fatigue tests under different stress levels and loading frequencies as the model object, a composite surface fatigue temperature rise model is proposed. The temperature rise of the composite test piece within one loading cycle is as follows:

[0077] Energy generated during one loading cycle:

[0078]

[0079] Among them, E sc is the generation energy due to the viscoelastic properties of the composite material, ε0 is the amplitude of the structural response strain, σ0 is the amplitude of the structural response stress, and δ is the hysteresis phase.

[0080] Dissipated energy in one loading cycle:

[0081]

[0082] Among them, E hs is the dissipated energy, the energy dissipated by heat conduction E cd and the energy dissipated by heat convection E dl composition, k is the thermal conductivity of each layer of the composite laminate, A is the surface area of ​​the test piece in contact with the air, represents the temperature gradient in the longitudinal and transverse directions of the test specimen structure, h is the heat convection coefficient between the test specimen and the air, T s is the surface temperature of the test piece, T ∞ is the ambient temperature.

[0083] The temperature rise of the composite material specimen during one loading cycle is:

[0084]

[0085] where ΔT is the fatigue temperature rise during one loading cycle, ρ is the density of the material, V is the volume of the composite structure, and c is the specific heat capacity of the material.

[0086] Under a certain number of cycles of loading, the fatigue temperature rise of the composite material test piece is as follows:

[0087]

[0088] Where T is the fatigue temperature rise and n is the number of cyclic loading.

[0089] [2] Random Forest Algorithm

[0090] In order to further efficiently and accurately predict the temperature changes on the surface of composite laminates, machine learning is combined with the random forest algorithm to achieve accurate prediction of the surface temperature rise of composite materials under tension-tension fatigue at different stress levels and loading frequencies.

[0091] The training set is constructed using the composite material tension-tension fatigue surface temperature rise data obtained from the experiment, in which the loading frequency, stress level and current fatigue cycle number are used as the training set input, and the surface temperature rise is used as the training set output.

[0092] [3] Conduct fatigue tests

[0093] Conduct tension-tension fatigue tests on composite materials at different stress levels and loading frequencies, obtain the tension-tension fatigue surface temperature rise curve of the composite materials, and record the normalized fatigue cycle number n / N f , stress level, loading frequency, and surface temperature rise.

[0094] Taking 3232A / S4C10-800 glass fiber composite laminate as an example, a fiber direction tension-tension fatigue temperature rise test was carried out, with a stress ratio of R = 0.1, fatigue stress levels of 75%, 70%, 65%, 60%, and 55% of the maximum load level, and loading frequencies of 10Hz and 27Hz. The scatter plot of surface temperature rise versus cycle number at different stress levels is shown in the figure below. Figure 1 The scatter plot of surface temperature rise versus cycle number under different loading frequencies is shown in Figure 2. Figure 2 .

[0095] [4] Fatigue temperature rise prediction combined with fatigue temperature rise prediction model

[0096] Based on the fatigue temperature rise prediction model and test data, more surface fatigue temperature rise curves of test specimens at different stress levels and loading frequencies were predicted. Based on the generated dissipated energy model, the surface temperature rise of glass fiber composite laminates at a loading frequency of 27Hz and a stress level of 65% was achieved. The surface temperature of the composite material that changes with the number of cyclic loading cycles is shown in the figure below. Figure 3 .

[0097] The developed generation and dissipation energy model is further used to predict the surface temperature of the test piece under different stress levels and different loading frequencies. The surface temperature diagram of the test piece under different stress levels predicted by the generation and dissipation energy model is as follows: Figure 4 The surface temperature diagram of the test piece under different loading frequencies predicted by the generated dissipated energy model is as follows: Figure 5 .

[0098] [5] Fatigue temperature rise prediction based on machine learning

[0099] The predicted fatigue temperature rise curve is input into a fatigue temperature rise model combined with machine learning to accurately predict the fatigue temperature rise of composite structures at different stress levels and loading frequencies. The loading frequency, stress level, and current number of fatigue cycles are used as input variables, and the surface temperature rise is used as the output.

[0100] The correlation between the prediction results and the test results was analyzed. The Pearson correlation coefficients at different stress levels and loading frequencies were 0.99 and 1.00 respectively. The trained random forest model can accurately predict the real-time temperature of the test piece surface. The correlation analysis diagram of fatigue temperature rise of the test piece at different stress levels predicted by the random forest algorithm is shown in the figure below. Figure 6The correlation analysis diagram of fatigue temperature rise of test pieces under different loading frequencies predicted by random forest algorithm is shown in the figure below. Figure 7 .

[0101] At the same time, the envelope of fatigue temperature rise curve of composite laminates under different stress levels and loading frequencies predicted by random forest algorithm is given when stress level is in the range of 55% to 80% and loading frequency is in the range of 10Hz to 27Hz. The envelope of fatigue temperature rise curve of composite laminates under different stress levels predicted by random forest algorithm is obtained as follows Figure 8 The envelope of fatigue temperature rise curves of composite laminates under different loading frequencies predicted by random forest algorithm is as follows: Figure 9 .

[0102] Other embodiments of the present disclosure will readily occur to those skilled in the art after considering the specification and practicing the disclosure herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the present disclosure being indicated by the following claims.

[0103] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A composite material fatigue temperature rise prediction method combined with a machine learning algorithm, characterized in that: include: Taking composite laminates as the research object, the random forest algorithm is used to predict the temperature rise of composite laminates under different frequencies and stress levels.

2. The method according to claim 1, characterized in that Taking composite laminates as the research object, the random forest algorithm is used to predict the temperature rise of composite laminates at different frequencies and stress levels, including: Use experiments to obtain fatigue temperature rise curves at different frequencies and stress levels; The fatigue temperature rise data is used as the training set for the random forest algorithm; different frequencies, stress levels, and fatigue cycle numbers are used as training set inputs, and the fatigue temperature rise is used as the training set output; The random forest algorithm was trained using the training set to obtain an algorithm for predicting fatigue temperature rise of composite materials; Input the target frequency, target stress level and different fatigue cycle numbers into the algorithm for predicting fatigue temperature rise of composite materials, and obtain the fatigue temperature rise corresponding to different fatigue cycle numbers at the target stress level and target frequency; According to the fatigue temperature rise corresponding to different fatigue cycle numbers at the target stress level and target frequency, the parameters of the fatigue temperature rise prediction model at the target stress level and target frequency are calculated, thereby obtaining the fatigue temperature rise prediction model at the target stress level and target frequency.

3. The method according to claim 2, characterized in that The fatigue temperature rise prediction model within a single fatigue cycle is: Energy generated during one loading cycle: Among them, E sc is the generation energy due to the viscoelastic properties of the composite material, ε0 is the amplitude of the structural response strain, σ0 is the amplitude of the structural response stress, and δ is the hysteresis phase; Dissipated energy in one loading cycle: Among them, E hs is the dissipated energy, the energy dissipated by heat conduction E cd and the energy dissipated by heat convection E dl composition, k is the thermal conductivity of each layer of the composite laminate, A is the surface area of ​​the test piece in contact with the air, represents the temperature gradient in the longitudinal and transverse directions of the test specimen structure, h is the heat convection coefficient between the test specimen and the air, T s is the surface temperature of the test piece, T ∞ is the ambient temperature; The temperature rise of the composite material specimen during one loading cycle is: where ΔT is the fatigue temperature rise during one loading cycle, ρ is the density of the material, V is the volume of the composite structure, and c is the specific heat capacity of the material. Under a certain number of cycles of loading, the fatigue temperature rise of the composite material test piece is: Where T is the fatigue temperature rise and n is the number of cyclic loading.

4. The method according to claim 3, characterized in that Stress levels include stress ratio and maximum fatigue stress.

5. A composite material fatigue temperature rise prediction device combined with a machine learning algorithm, characterized in that: include: The prediction module is used to predict the temperature rise of composite laminates under different frequencies and stress levels using the random forest algorithm.

6. The device according to claim 5, characterized in that The prediction module is specifically used to: Use experiments to obtain fatigue temperature rise curves at different frequencies and stress levels; The fatigue temperature rise data is used as the training set for the random forest algorithm; different frequencies, stress levels, and fatigue cycle numbers are used as training set inputs, and the fatigue temperature rise is used as the training set output; The random forest algorithm was trained using the training set to obtain an algorithm for predicting fatigue temperature rise of composite materials; Input the target frequency, target stress level and different fatigue cycle numbers into the algorithm for predicting fatigue temperature rise of composite materials, and obtain the fatigue temperature rise corresponding to different fatigue cycle numbers at the target stress level and target frequency; According to the fatigue temperature rise corresponding to different fatigue cycle numbers at the target stress level and target frequency, the parameters of the fatigue temperature rise prediction model at the target stress level and target frequency are calculated, thereby obtaining the fatigue temperature rise prediction model at the target stress level and target frequency.

7. The device according to claim 6, characterized in that The fatigue temperature rise prediction model within a single fatigue cycle is: Energy generated during one loading cycle: Among them, E sc is the generation energy due to the viscoelastic properties of the composite material, ε0 is the amplitude of the structural response strain, σ0 is the amplitude of the structural response stress, and δ is the hysteresis phase; Dissipated energy in one loading cycle: Among them, E hs is the dissipated energy, the energy dissipated by heat conduction E cd and the energy dissipated by heat convection E dl composition, k is the thermal conductivity of each layer of the composite laminate, A is the surface area of ​​the test piece in contact with the air, represents the temperature gradient in the longitudinal and transverse directions of the test specimen structure, h is the heat convection coefficient between the test specimen and the air, T s is the surface temperature of the test piece, T ∞ is the ambient temperature; The temperature rise of the composite material specimen during one loading cycle is: where ΔT is the fatigue temperature rise during one loading cycle, ρ is the density of the material, V is the volume of the composite structure, and c is the specific heat capacity of the material. Under a certain number of cycles of loading, the fatigue temperature rise of the composite material test piece is: Where T is the fatigue temperature rise and n is the number of cyclic loading.

8. The device according to claim 7, characterized in that Stress levels include stress ratio and maximum fatigue stress.

Images