A method, system, medium and product for predicting fatigue damage evolution of a drive shaft

Abstract

The application provides a kind of drive shaft fatigue damage evolution prediction method, system, medium and product.In the method, the prediction system is according to three-dimensional configuration data to carry out finite element grid discretization, generates global mass matrix and global stiffness matrix;Based on current local stress state parameter and the material fatigue degradation benchmark parameter of pre-set, the stiffness reduction matrix parameter is calculated, and the attenuation global stiffness matrix is obtained;Based on the attenuation global stiffness matrix and global mass matrix, the eigenvalue is solved to obtain real-time natural frequency, calculate its with pre-set working speed frequency Absolute frequency difference, resonance stress gain coefficient is obtained;The amplitude compensation of torque load of next time sequence step is carried out using resonance stress gain coefficient, and multi-round progressive iterative calculation is carried out;When real-time natural frequency is less than pre-set frequency threshold, iteration is terminated, and current cumulative total time length is extracted as predicted fatigue failure life.The method alleviates the problem that the prediction result deviates from the actual failure node, and improves the prediction accuracy.

Description

Technical Field

[0001] This application relates to the field of equipment fatigue damage prediction technology, and in particular to a method, system, medium and product for predicting the evolution of fatigue damage in a drive shaft. Background Technology

[0002] Drive shafts (such as metal or composite material drive shafts) are widely used in aerospace and heavy-duty power transmission systems due to engineering requirements such as high specific strength and lightweight design. Under long-term high-speed rotation and alternating torque service environments, the accumulation of microscopic damage within the material can lead to fatigue degradation of the drive shaft. Therefore, accurate simulation and simulation technologies are urgently needed in engineering to ensure operational safety throughout its entire life cycle.

[0003] In related technologies, fatigue prediction techniques typically employ models based on progressive damage evolution or standard SN curves. This process generally involves reading initial mechanical parameters, calculating damage accumulation over time steps under a preset static load sequence, measuring the degree of failure based on the gradual reduction of macroscopic stiffness, and finally using the time it takes for accumulated damage to reach a threshold as the life assessment result.

[0004] However, as damage accumulates and macroscopic stiffness decreases, the natural frequency of the drive shaft will spontaneously shift downwards. When the decayed natural frequency enters the basic operating speed range, it will induce severe resonance under actual operating conditions, generating high-amplitude pulse stress. Because the existing simulation process is constrained by the initially set constant load environment, the simulation calculation is still under low load stress at the end of the lifespan, resulting in an unrealistically flat extension of the prediction curve, causing the life prediction value to be much higher than the actual failure point. Summary of the Invention

[0005] This application provides a method, system, medium, and product for predicting the fatigue damage evolution of drive shafts, which alleviates the technical problem of life prediction results deviating from the actual failure point and improves the accuracy of fatigue damage evolution prediction.

[0006] In a first aspect, this application provides a method for predicting the fatigue damage evolution of a driveshaft, comprising: acquiring three-dimensional configuration data of the driveshaft, a preset operating speed frequency, and a torque load spectrum; discretizing the three-dimensional configuration data into a finite element mesh to generate a global mass matrix and a global stiffness matrix; the three-dimensional configuration data including three-dimensional geometric dimension data and material mechanical property data; extracting the current torque load sequence from the torque load spectrum according to a preset time step, and calculating the current local stress state parameters of the driveshaft under the current torque load sequence; calculating the stiffness reduction matrix parameters within the time step based on the current local stress state parameters and preset material fatigue degradation benchmark parameters; subtracting the stiffness reduction matrix parameters from the current global stiffness matrix to obtain the attenuated global stiffness matrix; and the current global stiffness matrix in the first iteration. The global stiffness matrix is ​​used as the time step. Based on the decayed global stiffness matrix and the global mass matrix, the matrix eigenvalues ​​are solved to obtain the real-time natural frequency of the drive shaft under the current decayed state. The absolute frequency difference between the real-time natural frequency and the preset operating speed frequency is calculated. The resonance stress gain coefficient is calculated based on the absolute frequency difference, and the resonance stress gain coefficient represents the resonance stress gain effect. The amplitude compensation of the torque load in the next time step is performed using the resonance stress gain coefficient, and the current global stiffness matrix is ​​updated based on the decayed global stiffness matrix to perform multi-round progressive iteration calculations of time step. When the real-time natural frequency corresponding to the current iteration is detected to be less than the preset frequency threshold, the iteration calculation is terminated, the current cumulative total time duration is extracted as the predicted fatigue failure life, and an early warning signal corresponding to the predicted fatigue failure life is generated.

[0007] By employing the above technical solution, the prediction system first generates a global mass matrix and a global stiffness matrix based on three-dimensional configuration data. In multi-round time-step progressive iterative calculations, the prediction system couples the current local stress state parameters generated by the current torque load sequence with the material fatigue degradation benchmark parameters to obtain stiffness reduction matrix parameters and update them to a decayed global stiffness matrix, realizing the state mapping from microscopic damage to macroscopic structural degradation. Based on this, the prediction system uses the absolute frequency difference between the real-time natural frequency induced by the decayed global stiffness matrix and the preset operating speed frequency to calculate the resonant stress gain coefficient. Furthermore, the prediction system uses this coefficient for load amplitude compensation in the next time step, forming a feedback loop of stiffness attenuation, frequency slip, and load amplification interaction, until a preset frequency threshold is triggered, at which point the predicted fatigue failure life is output and an early warning signal is generated. This method alleviates the technical problem of the transmission shaft's life prediction results deviating from the actual failure point and improves the accuracy of fatigue damage evolution prediction.

[0008] In conjunction with some embodiments of the first aspect, in some embodiments, finite element mesh discretization is performed based on the three-dimensional configuration data to generate a global mass matrix and a global stiffness matrix. Specifically, this includes: constructing a three-dimensional solid model of the drive shaft based on the three-dimensional geometric dimension data in the three-dimensional configuration data; performing finite element mesh discretization on the three-dimensional solid model to obtain discrete mesh elements; calculating the material stiffness properties of each discrete mesh element based on material mechanical property data, and calculating the mass distribution properties of each discrete mesh element in conjunction with the three-dimensional geometric dimension data, wherein the material mechanical property data includes at least the elastic modulus, shear modulus, and Poisson's ratio; and performing global matrix integration on the material stiffness properties and mass distribution properties of each discrete mesh element to obtain a global mass matrix and a global stiffness matrix.

[0009] By employing the aforementioned technical solution, the prediction system transforms three-dimensional geometric data into discrete mesh elements through finite element mesh discretization, and then integrates these elements with material mechanical property data to calculate the material stiffness and mass distribution properties of each element, achieving a fundamental coupling between material properties and spatial physical distribution. Furthermore, the prediction system performs global matrix integration on these discrete properties to derive a global mass matrix and a global stiffness matrix reflecting the overall mechanical characteristics of the system. This method establishes a mapping from local physical properties to macroscopic system characteristics, laying the initial mechanical benchmark for subsequent fatigue damage mechanics deductions.

[0010] In conjunction with some embodiments of the first aspect, in some embodiments, based on the current local stress state parameters and preset material fatigue degradation benchmark parameters, the stiffness reduction matrix parameters within the time step are calculated. Specifically, this includes: extracting the principal stress components and shear stress components of each discrete mesh element under the current loading state based on the current local stress state parameters; identifying discrete mesh elements whose principal stress components are greater than the fatigue tensile strength threshold in the preset material fatigue degradation benchmark parameters, or whose shear stress components are greater than the fatigue shear strength threshold in the preset material fatigue degradation benchmark parameters, as damaged mesh elements, and calculating the corresponding local damage evolution factor based on the stress amplitude exceeded by each damaged mesh element; using each local damage evolution factor to perform directional reduction on the material stiffness properties of the corresponding discrete mesh element to obtain the local stiffness reduction matrix of each damaged mesh element; mapping each local stiffness reduction matrix to the global coordinate system for matrix integration to obtain the stiffness reduction matrix parameters within the time step.

[0011] By employing the aforementioned technical solution, the prediction system compares and interacts with the principal stress components, shear stress components, and material fatigue degradation benchmark parameters to quantify the local damage evolution factor of the damaged mesh element. Subsequently, the prediction system uses the local damage evolution factor to perform directional reduction of the material stiffness properties of each discrete mesh element in the current iteration state to generate a local stiffness reduction matrix. This matrix is ​​then mapped and fused to the global coordinate system through matrix integration to obtain the macroscopic stiffness reduction matrix parameters. This method transforms the microscopic stress overload state into a stiffness attenuation parameter at the structural physics level, improving the numerical quantification accuracy of fatigue degradation simulation.

[0012] In conjunction with some embodiments of the first aspect, in some embodiments, the torque load of the next time step is compensated for using the resonant stress gain coefficient, and the current global stiffness matrix is ​​updated based on the attenuated global stiffness matrix to perform multi-round progressive iterative calculations of the time step. Specifically, this includes: extracting the initial torque load amplitude of the next time step from the torque load spectrum, multiplying the initial torque load amplitude by the resonant stress gain coefficient to obtain the compensated torque load sequence; updating the attenuated global stiffness matrix to the current global stiffness matrix of the next time step to obtain the updated global stiffness matrix; calculating the current local stress state parameters of the drive shaft based on the updated global stiffness matrix under the action of the compensated torque load sequence; and repeatedly calculating the attenuated global stiffness matrix, real-time natural frequency, resonant stress gain coefficient, and updated global stiffness matrix to perform multi-round progressive iterative calculations of the time step.

[0013] By employing the aforementioned technical solution, the prediction system integrates the resonant stress gain coefficient with the initial torque load amplitude to generate a compensated torque load sequence, while simultaneously updating the current global stiffness matrix based on the decaying global stiffness matrix. On this basis, the prediction system couples the updated compensated load input with the degraded stiffness matrix to re-obtain the current local stress state parameters and continuously drives multiple rounds of iterative calculations with progressively increasing time steps. This method facilitates the interaction between the amplification effect of the input load and the decay state of the structural stiffness, enabling the simulation to model the accelerated fatigue damage evolution under complex alternating loads.

[0014] In conjunction with some embodiments of the first aspect, in some embodiments, after repeatedly calculating the attenuation global stiffness matrix, real-time natural frequency, resonant stress gain coefficient, and updated global stiffness matrix, the method further includes: calculating the frequency drop difference between the real-time natural frequency corresponding to the current iteration and the real-time natural frequency corresponding to the previous iteration; dividing the frequency drop difference by the current time step size to obtain the current frequency drop gradient; when the absolute frequency difference is less than a preset resonance threshold and the current frequency drop gradient is greater than a preset gradient threshold, dividing the gradient threshold by the current frequency drop gradient to obtain the step size contraction coefficient; multiplying the current time step size by the step size contraction coefficient to obtain the updated next time step size, and performing multiple rounds of progressive iteration calculation of the time step size according to the updated next time step size.

[0015] By employing the above technical solution, the prediction system interactively calculates the current frequency descent gradient by comparing the real-time natural frequency descent difference with the current time step size. When the frequency descent gradient and the absolute frequency difference jointly trigger the gradient threshold and resonance threshold conditions, the prediction system uses the gradient ratio to generate a step size contraction coefficient, which is then interacted with the current time step size to obtain the reduced next time step size. This method enables automatic contraction of the prediction step size at the end of fatigue failure, mitigating prediction distortion caused by numerical leaps and improving the computational time resolution of the critical failure stage.

[0016] In conjunction with some embodiments of the first aspect, in some embodiments, after updating the current global stiffness matrix based on the attenuated global stiffness matrix to perform multi-round time-step progressive iterative calculation, the method further includes: when the real-time natural frequency corresponding to the current iteration is detected to be greater than or equal to a preset frequency threshold, determining the torque load of the next time-step after amplitude compensation as the current torque load sequence; accumulating the current time-step to update the current cumulative total time-step duration; calculating the current local stress state parameters of the drive shaft under the current torque load sequence to continue multi-round time-step progressive iterative calculation.

[0017] By employing the aforementioned technical solution, within the safe evolution range where the real-time natural frequency does not trigger the preset frequency threshold, the prediction system identifies the amplitude-compensated load as the current torque load sequence for the next cycle and simultaneously accumulates the current time step to update the total accumulated time duration. Subsequently, the prediction system recalculates new current local stress state parameters based on the updated load sequence, thus connecting to multiple rounds of progressively iterative calculations with progressively increasing time steps. This method ensures the smooth transmission and continuous action of load, time, and stress in the drive shaft's non-failure state, maintaining the consistency of the stress state and the integrity of the time-series deduction, and guaranteeing the stable accumulation of the life assessment cycle.

[0018] In conjunction with some embodiments of the first aspect, in some embodiments, the resonant stress gain coefficient is calculated based on the absolute frequency difference, specifically including: determining the structural stiffness loss ratio based on the numerical change between the global stiffness matrix and the attenuated global stiffness matrix; extracting initial structural dissipation parameters from the material mechanical property data of the three-dimensional configuration data; adjusting the initial structural dissipation parameters by amplification based on the structural stiffness loss ratio to obtain the current total structural dissipation parameters; determining the resonant amplification factor using the absolute frequency difference, and numerically reducing the resonant amplification factor using the current total structural dissipation parameters to obtain the resonant stress gain coefficient.

[0019] By employing the above technical solution, the prediction system extracts the numerical difference between the global stiffness matrix and the attenuated global stiffness matrix as the structural stiffness loss ratio, and interacts it with the initial structural dissipation parameters to obtain the current total structural dissipation parameter characterizing damage energy dissipation. Furthermore, the prediction system uses this current total structural dissipation parameter to perform damping reduction interaction on the resonance amplification factor, ultimately calculating the corrected resonance stress gain coefficient. This method alleviates the gain distortion caused by relying solely on frequency difference calculations in the later stages of fatigue, and improves the accuracy of life assessment under extreme conditions.

[0020] In a second aspect, this application provides a prediction system including one or more processors and a memory; the memory is coupled to the one or more processors and is used to store computer program code including computer instructions, which the one or more processors call to cause the prediction system to perform the methods described in the first aspect and any possible implementation thereof.

[0021] Thirdly, this application provides a computer-readable storage medium including computer instructions that, when executed on a prediction system, cause the prediction system to perform the method described in the first aspect and any possible implementation thereof.

[0022] Fourthly, this application provides a computer program product, including a computer program / instruction that, when run on a prediction system, causes the prediction system to perform the method described in the first aspect and any possible implementation thereof.

[0023] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. By adopting a state mapping based on micro-damage to macro-stiffness decay, combined with real-time natural frequency slip quantization resonance stress gain, and by dynamically compensating the load amplitude to construct a closed-loop iteration of fatigue evolution, the technical problem of life prediction results deviating from the actual failure point is effectively alleviated, and the accuracy of fatigue damage evolution prediction is improved.

[0024] 2. By adopting a joint triggering mechanism based on frequency descent gradient and resonance evolution state, and generating a step size contraction coefficient in combination with stiffness degradation degree, the technique of adaptively and dynamically reducing the time sequence step size of the deduction is used. Therefore, it effectively alleviates the problem of numerical leap evolution and loss of local stress details caused by using a fixed calculation step size due to the rapid stiffness degradation during the fatigue failure critical period, and improves the deduction accuracy of the critical failure stage.

[0025] 3. By adopting the total dissipation parameter of the system based on the dynamic derivation of structural stiffness loss, combined with the damping increase effect caused by damage accumulation, and by reducing and correcting the resonance stress amplification factor, the numerical distortion problem that easily leads to infinite amplification of resonance gain when relying solely on frequency approximation calculation in the later stage of fatigue degradation is effectively alleviated, and the numerical accuracy of fatigue life assessment under extreme working conditions is improved. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating a method for predicting the fatigue damage evolution of a drive shaft in an embodiment of this application. Figure 2 This is another flowchart illustrating the drive shaft fatigue damage evolution prediction method in the embodiments of this application; Figure 3 This is a schematic diagram of the physical device structure of the prediction system in the embodiments of this application. Detailed Implementation

[0027] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification and appended claims of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to any or all possible combinations including one or more of the listed items.

[0028] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0029] The following describes the process of the method provided in this implementation. Please refer to [link / reference]. Figure 1 This is a flowchart illustrating a method for predicting the fatigue damage evolution of a drive shaft in an embodiment of this application.

[0030] 101. Obtain the three-dimensional configuration data of the drive shaft, the preset operating speed frequency and torque load spectrum, and perform finite element mesh discretization based on the three-dimensional configuration data to generate a global mass matrix and a global stiffness matrix. The three-dimensional configuration data includes three-dimensional geometric dimension data and material mechanical property data.

[0031] The three-dimensional configuration data of the driveshaft refers to a set of digital information describing the complete physical characteristics of the driveshaft; the three-dimensional geometric dimension data represents the spatial shape and size information of the driveshaft, including parameters such as the shaft's length, diameter, and wall thickness; the material mechanical property data represents the physical performance parameters of the driveshaft material, including parameters such as elastic modulus, Poisson's ratio, density, and yield strength; the preset operating speed frequency refers to the rotational frequency of the driveshaft during actual operation; the torque load spectrum is a complete record of the torque borne by the driveshaft over time during its working cycle; finite element mesh discretization refers to the process of dividing the continuous geometry of the driveshaft into a finite number of element meshes; the global mass matrix represents the matrix of inertial characteristics of the entire driveshaft system; and the global stiffness matrix represents the matrix of the entire driveshaft system's resistance to deformation.

[0032] Specifically, the prediction system first obtains the three-dimensional configuration data, preset operating speed frequency, and torque load spectrum of the driveshaft from a database or input interface. The three-dimensional geometric dimensions in the configuration data define the spatial structure of the driveshaft, and the material mechanical property data defines the mechanical response characteristics of the material. The prediction system then performs finite element mesh discretization on the three-dimensional geometric model of the driveshaft, dividing the continuous driveshaft structure into several interconnected elements. Each element contains node and element type information. Based on the element node coordinates and material mechanical property data, the prediction system generates a global mass matrix and a global stiffness matrix through the assembly process of element stiffness matrices and element mass matrices. The global mass matrix reflects the mass distribution and inertial characteristics of the driveshaft, while the global stiffness matrix reflects the structural stiffness characteristics of the driveshaft.

[0033] In some embodiments, the prediction system achieves finite element mesh discretization and matrix generation in the following manner: the prediction system first identifies the geometric boundaries and feature lines of the drive shaft, then generates tetrahedral mesh elements inside the drive shaft, then assigns coordinates and degree of freedom numbers to each element node, and finally calculates the stiffness matrix and mass matrix of each element based on the element shape function and material properties and assembles them into a global matrix.

[0034] 102. Extract the current torque load sequence from the torque load spectrum according to the preset time step size, and calculate the current local stress state parameters of the drive shaft under the current torque load sequence.

[0035] The preset time step refers to the fixed time interval for dividing the torque load spectrum into time series; the torque load spectrum represents the complete data set of the torque that the drive shaft bears over the entire working cycle as a function of time; the current torque load sequence refers to the torque data segment extracted from the torque load spectrum within a time step; the current local stress state parameter refers to the set of stress components of each node or unit of the drive shaft under the current torque load sequence, including stress components such as normal stress, shear stress, and equivalent stress; the drive shaft refers to the shaft-like parts used to transmit torque and rotational motion; the action refers to the physical process of applying external torque load to the drive shaft structure and generating internal stress response.

[0036] Specifically, the prediction system segments the torque load spectrum according to a preset time step. Starting from the beginning of the torque load spectrum, it extracts a continuous segment of torque data as the current torque load sequence according to the time step length. The prediction system converts the current torque load sequence into boundary conditions applied to the finite element model of the drive shaft, and calculates the displacement response of each node of the drive shaft by solving the finite element equilibrium equations. The prediction system calculates the strain components of each element based on the node displacement and the element strain-displacement relationship matrix, and then converts the strain components into stress components through the material constitutive relation to obtain the current local stress state parameters of the drive shaft under the current torque load sequence. These parameters contain the stress tensor information of all elements or key locations of the drive shaft.

[0037] In some embodiments, the prediction system calculates the stress state parameters by first converting the torque values ​​in the current torque load sequence into nodal force vectors, then establishing global equilibrium equations and solving for nodal displacements using the direct method, then calculating the strain using the derivatives of the element shape functions, and finally calculating the stress components according to Hooke's law.

[0038] 103. Based on the current local stress state parameters and the preset material fatigue degradation benchmark parameters, calculate the stiffness reduction matrix parameters within the time step. Subtract the stiffness reduction matrix parameters from the current global stiffness matrix to obtain the attenuated global stiffness matrix. The current global stiffness matrix is ​​the global stiffness matrix in the first iteration.

[0039] The preset material fatigue degradation benchmark parameters refer to the characteristic parameters that describe the stiffness degradation law of materials under cyclic loading, including parameters such as fatigue damage threshold, damage accumulation index, and stiffness degradation coefficient; the stiffness reduction matrix parameter represents the matrix form of the amount of structural stiffness reduction caused by fatigue damage; the current global stiffness matrix represents the overall stiffness characteristic matrix of the drive shaft under the current damage state; the decay global stiffness matrix represents the updated stiffness matrix of the drive shaft after considering fatigue damage accumulation; the first iteration refers to the first time step of the prediction system to start fatigue damage accumulation calculation.

[0040] Specifically, the prediction system first calculates the stress amplitude and stress cycle characteristics of each element or node based on the current local stress state parameters. Then, it establishes a mapping relationship between stress level and stiffness degradation based on preset material fatigue degradation benchmark parameters. The prediction system calculates the stiffness degradation caused by fatigue damage in each element within a time step using a fatigue damage accumulation model, and assembles the stiffness degradation of each element into a stiffness reduction matrix parameter. The prediction system subtracts the stiffness reduction matrix parameter from the current global stiffness matrix, completing the element-wise subtraction operation to obtain the attenuated global stiffness matrix. When the prediction system is in its first iteration, the current global stiffness matrix is ​​the initial global stiffness matrix generated in step 101; in subsequent iterations, the current global stiffness matrix is ​​the attenuated global stiffness matrix calculated in the previous time step.

[0041] In some embodiments, the prediction system calculates the stiffness reduction matrix parameters in the following manner: First, the prediction system counts the stress cycle number and amplitude of the current torque load sequence according to the rainflow counting method. Then, it applies Miner's linear cumulative damage theory to calculate the damage increment. Next, it converts the damage increment into stiffness reduction based on the stiffness degradation coefficient in the material fatigue degradation reference parameters. Finally, it constructs the corresponding stiffness reduction matrix parameters.

[0042] 104. Based on the decay global stiffness matrix and global mass matrix, solve the matrix eigenvalues ​​to obtain the real-time natural frequency of the drive shaft under the current decay state.

[0043] The current attenuation state of the drive shaft represents the structural state of the drive shaft after experiencing accumulated fatigue damage; the real-time natural frequency refers to the free vibration frequency of the drive shaft in the current damage state, reflecting the changes in dynamic characteristics caused by the degradation of structural stiffness; matrix eigenvalue solving refers to the calculation process of solving the generalized eigenvalue problem to obtain the system's natural frequencies and mode shapes. The generalized eigenvalue problem is represented by the matrix equation K*Φ=ω²*M*Φ, where K is the attenuation global stiffness matrix, M is the global mass matrix, ω is the circular frequency, and Φ is the mode shape vector.

[0044] Specifically, the prediction system substitutes the decayed global stiffness matrix and global mass matrix into the generalized eigenvalue equation, constructing a matrix eigenvalue problem of the form K*Φ=λ*M*Φ, where the eigenvalue λ equals the square of the angular frequency ω. The prediction system uses a numerical algorithm to solve this generalized eigenvalue problem, obtaining a series of eigenvalues ​​and corresponding eigenvectors. The eigenvectors represent mode shapes, and the square root of the eigenvalues ​​represents the angular frequency. The prediction system divides the angular frequency by 2π to convert it to a frequency in Hertz units, obtaining the real-time natural frequency of the drive shaft under the current decay state. This real-time natural frequency decreases with the accumulation of fatigue damage, reflecting the degree of structural stiffness degradation and providing dynamic characteristic parameters for subsequent resonance risk assessment.

[0045] In some embodiments, the prediction system solves for matrix eigenvalues ​​in the following manner: The prediction system uses the Lanczos method to first transform the generalized eigenvalue problem into a standard eigenvalue problem, then generates a tridiagonal matrix through the Lanczos tridiagonalization process, then solves for the eigenvalues ​​of the tridiagonal matrix to obtain approximate eigenvalues ​​of the original problem, and finally obtains the real-time natural frequency through inverse transformation and unit transformation.

[0046] 105. Calculate the absolute frequency difference between the real-time natural frequency and the preset operating speed frequency. Calculate the resonance stress gain coefficient based on the absolute frequency difference. The resonance stress gain coefficient represents the resonance stress gain effect.

[0047] The preset operating speed frequency refers to the rotational frequency of the drive shaft during actual operation; the absolute frequency difference refers to the absolute value of the difference between the real-time natural frequency and the preset operating speed frequency, reflecting the degree of distance of the system from the resonance state; the resonance stress gain coefficient refers to the stress amplification factor caused by the excitation frequency being close to the natural frequency, quantifying the enhancement effect of the resonance effect on the stress level; the resonance stress gain effect represents the physical phenomenon that the dynamic response amplitude increases significantly when the excitation frequency is close to the natural frequency of the structure.

[0048] Specifically, the prediction system first calculates the difference between the real-time natural frequency and the preset operating speed frequency. Then, it takes the absolute value of this difference to obtain the absolute frequency difference; the smaller the absolute frequency difference, the closer the system is to the resonance state. Based on the absolute frequency difference, the prediction system calculates the resonance stress gain coefficient, taking into account the influence of frequency proximity on stress amplification. The prediction system quantifies the intensity of the resonance stress gain effect through the resonance stress gain coefficient. When the absolute frequency difference is small, the resonance stress gain coefficient is large, indicating that the stress will be significantly amplified; when the absolute frequency difference is large, the resonance stress gain coefficient is close to 1, indicating a weak resonance effect. The prediction system uses the calculated resonance stress gain coefficient for subsequent stress correction to accurately reflect the true stress state under dynamic excitation conditions.

[0049] It should be noted that, considering the inherent dissipative properties of drive shaft materials in the real physical world, to prevent the resonance stress gain coefficient from amplifying infinitely when the absolute frequency difference approaches zero, thus causing the finite element numerical calculation to collapse, the prediction system incorporates a default basic damping constant to prevent numerical divergence in the basic calculation model for calculating the resonance stress gain coefficient based on this absolute frequency difference, or sets a preset physical upper limit threshold for the resonance gain coefficient. This ensures that even when the real-time natural frequency of the drive shaft completely drops to the extreme resonance point equivalent to the preset operating speed frequency, the calculated resonance stress gain coefficient remains a bounded convergent value conforming to the physical laws of finite deformation, guaranteeing the stable implementation of the system's multi-round time-step progressive iterative calculations.

[0050] In some embodiments, the prediction system first determines the structural stiffness loss ratio based on the numerical change between the global stiffness matrix and the attenuated global stiffness matrix. It then calculates the norm of the difference between the corresponding elements of the attenuated global stiffness matrix and the global stiffness matrix, and divides this norm by the norm of the global stiffness matrix to obtain the structural stiffness loss ratio. This ratio reflects the degree of stiffness degradation caused by fatigue damage.

[0051] Then, the prediction system extracts initial structural dissipation parameters from the material mechanical property data of the 3D configuration data. These initial structural dissipation parameters include parameters characterizing energy dissipation capacity, such as material damping ratio and viscous damping coefficient. The prediction system adjusts the initial structural dissipation parameters based on the structural stiffness loss ratio. The current total structural dissipation parameters are obtained by multiplying the initial structural dissipation parameters by an adjustment factor. The adjustment factor is determined by the structural stiffness loss ratio through empirical formulas or theoretical models. The greater the stiffness loss, the greater the increase in dissipation parameters, reflecting the increased energy dissipation effect caused by damage.

[0052] Finally, the prediction system determines the resonance amplification factor using the absolute frequency difference and calculates it using the resonance amplification formula, which states that the amplification factor equals 1 divided by the square root of the ratio of the absolute frequency difference to the natural frequency. As the absolute frequency difference approaches zero, the amplification factor approaches infinity. The prediction system then numerically reduces the resonance amplification factor using the current total structural dissipation parameter. By dividing the resonance amplification factor by a reduction factor that includes the current total structural dissipation parameter, the resonance stress gain coefficient is obtained. This reduction process reflects the suppression effect of structural dissipation on the resonance peak; the larger the dissipation parameter, the smaller the resonance stress gain coefficient.

[0053] 106. The amplitude of the torque load in the next time step is compensated by the resonant stress gain coefficient, and the current global stiffness matrix is ​​updated based on the decay global stiffness matrix to perform multi-round time step progressive iterative calculation.

[0054] The next time step refers to the next time interval immediately following the current time step; the torque load represents the torsional moment applied to the drive shaft; amplitude compensation refers to the process of correcting and adjusting the value of the torque load according to the resonant stress gain coefficient; the current global stiffness matrix represents the overall stiffness characteristic matrix of the drive shaft under the current damage state; multi-round time step progressive iterative calculation refers to the cyclic process of progressively advancing according to the time sequence and repeatedly performing damage accumulation and stiffness update calculations.

[0055] Specifically, the prediction system uses the resonant stress gain coefficient to compensate for the amplitude of the torque load in the next time step. By multiplying the original torque load amplitude by the resonant stress gain coefficient, the corrected torque load is obtained. This correction process reflects the actual enhancement effect of resonance on the stress level. The prediction system updates the current global stiffness matrix based on the currently calculated attenuated global stiffness matrix, using this matrix as the initial stiffness matrix for the next time step calculation, thus achieving continuous transmission of stiffness degradation. Based on the updated stiffness matrix and the compensated torque load, the prediction system performs stress calculation, damage accumulation, and stiffness reduction calculations for the next time step, forming a complete iterative cycle. Through multiple rounds of progressively iterative calculations in each time step, the prediction system gradually simulates the cumulative evolution of fatigue damage and the gradual degradation of structural stiffness in the drive shaft during long-term service.

[0056] In some embodiments, the prediction system achieves amplitude compensation and iterative calculation in the following manner: The prediction system first extracts the initial torque load amplitude for the next time step from the torque load spectrum. This amplitude corresponds to the torque value sequence for the next time period in the torque load spectrum. The prediction system multiplies the initial torque load amplitude with the resonant stress gain coefficient, completes the multiplication operation for each data point, and obtains the compensated torque load sequence. The amplitude of the compensated torque load sequence is increased compared to the original load amplitude, and the degree of increase is determined by the resonant stress gain coefficient.

[0057] Then, the prediction system updates the decayed global stiffness matrix to the current global stiffness matrix for the next time step. Through matrix assignment, all element values ​​of the decayed global stiffness matrix are transferred to the current global stiffness matrix, resulting in an updated global stiffness matrix that reflects the actual stiffness state of the driveshaft at the current moment. Under the action of the compensated torque load sequence, the prediction system calculates the current local stress state parameters of the driveshaft based on the updated global stiffness matrix. By establishing finite element equilibrium equations and solving for nodal displacements, the system calculates the strain and stress distribution. These stress state parameters already include the influence of resonance effects. The prediction system repeatedly calculates the decayed global stiffness matrix, real-time natural frequency, resonance stress gain coefficient, and updated global stiffness matrix, iteratively following steps 103 to 106. In each iteration, the time is advanced by one time step to perform multiple rounds of progressively increasing time step iterations, simulating the damage accumulation process of the driveshaft throughout its service life.

[0058] Next, the prediction system calculates the frequency drop difference between the real-time intrinsic frequency of the current iteration and the real-time intrinsic frequency of the previous iteration. This difference is obtained by subtracting the real-time intrinsic frequency of the previous iteration from the current iteration's real-time intrinsic frequency and taking the absolute value. This difference reflects the magnitude of the intrinsic frequency decrease within a time step. The prediction system then divides the frequency drop difference by the current time step to obtain the current frequency drop gradient. This gradient represents the rate of decrease of the intrinsic frequency per unit time, reflecting the speed of damage accumulation.

[0059] Finally, the prediction system determines that when the absolute frequency difference is less than the preset resonance threshold and the current frequency descent gradient is greater than the preset gradient threshold, it indicates that the system is rapidly approaching resonance and the damage accumulation rate is high. The prediction system divides the gradient threshold by the current frequency descent gradient to obtain a step size contraction coefficient. This step size contraction coefficient is less than 1 and is used to reduce the time step size to improve the calculation accuracy of the critical region. The prediction system multiplies the current time step size by the step size contraction coefficient to obtain the updated next time step size. This updated next time step size is less than the original time step size. The prediction system performs multiple rounds of iterative calculations based on the updated next time step size. By adaptively adjusting the time step size, it achieves refined analysis in the resonance critical region, ensuring accurate capture of the dynamic process of rapid stiffness degradation and rapid frequency descent. Meanwhile, proportionally reducing the time step size of the physical analysis allows the cumulative torque cycle load number within each step of the analysis to be reduced proportionally. This prevents abrupt changes in the "stress-stiffness" calculation caused by excessively long single-step time when the drive shaft approaches the resonance point and triggers a high pulse gain section. This eliminates the problems of matrix numerical divergence and distortion of the predicted evolution curve.

[0060] 107. When the real-time inherent frequency corresponding to the current iteration is detected to be less than the preset frequency threshold, the iteration calculation is terminated, the current cumulative total time series duration is extracted as the predicted fatigue failure life, and an early warning signal corresponding to the predicted fatigue failure life is generated.

[0061] The real-time natural frequency corresponding to the current iteration refers to the natural frequency value of the drive shaft calculated in the current iteration loop; the preset frequency threshold refers to the pre-set critical value of the natural frequency, which represents the frequency criterion for structural failure; terminating the iteration calculation refers to stopping the cyclic process of multi-round progressive iteration calculation of time steps; the cumulative total time duration refers to the sum of all time steps experienced from the initial moment to the current moment; the predicted fatigue failure life refers to the working time of the drive shaft to reach the failure state predicted by cumulative damage analysis; the warning signal indicates the notification information issued to the user or system regarding the risk of structural failure.

[0062] Specifically, the prediction system compares the real-time natural frequency corresponding to the current iteration with a preset frequency threshold to determine if the real-time natural frequency is less than the preset frequency threshold. When the real-time natural frequency is less than the preset frequency threshold, it indicates that the drive shaft has suffered severe stiffness degradation due to accumulated fatigue damage, and the structure has reached the failure criteria. The prediction system immediately terminates the iterative calculation and stops the analysis process for subsequent time steps. The prediction system extracts the sum of all time steps experienced from the initial moment to the current moment. This cumulative total time duration represents the working time that the drive shaft can withstand from a damaged state to a failure state. The prediction system outputs the cumulative total time duration as the predicted fatigue failure life.

[0063] The prediction system will output the generated early warning signal to the physical field control terminal (such as a PLC controller) where the drive shaft is located in real time via the communication bus, triggering the control terminal to issue a frequency conversion adjustment command to the generator signal under the running drive system, so that the physical drive motor can automatically adjust its actual working speed to avoid the current real-time inherent frequency; or when the system detects that the lifespan is about to end, it will trigger the hardware to power off, stop, lock, and mechanically brake through the early warning signal, thereby realizing intervention in the operating status of the equipment and ensuring its safety.

[0064] In some embodiments, the prediction system achieves failure determination and early warning generation in the following manner: The prediction system first monitors three indicators simultaneously: real-time natural frequency, maximum stress level, and stiffness loss ratio, and then compares them with their respective thresholds. When the real-time natural frequency is less than a preset frequency threshold or other indicators exceed the corresponding threshold, failure determination is triggered. Finally, the cumulative total time duration is extracted as the predicted fatigue failure life and a comprehensive early warning signal containing multi-indicator failure information is generated.

[0065] The driveshaft fatigue damage evolution prediction method described in this application first discretizes the finite element mesh based on the three-dimensional configuration data of the driveshaft, generating a global mass matrix and a global stiffness matrix. Then, the prediction system calculates local stress state parameters based on the current torque load sequence, coupling them with material fatigue degradation benchmark parameters to obtain stiffness reduction matrix parameters, thus achieving a quantitative conversion from microscopic damage to macroscopic stiffness attenuation. Next, the prediction system solves for the eigenvalues ​​to obtain the real-time natural frequency, and calculates the resonance stress gain coefficient based on the operating speed frequency, completing the quantitative characterization of the resonance effect. Finally, the prediction system uses this coefficient to compensate for the torque load amplitude at the next time step, conducting multiple rounds of progressive iteration until the real-time natural frequency reaches a threshold, at which point the iteration terminates, and the cumulative duration is extracted as the failure lifetime. This method alleviates the problem of prediction results deviating from the actual failure point and improves prediction accuracy.

[0066] The following provides a more detailed description of the process of the method provided in this implementation. Please refer to [link / reference]. Figure 2This is another flowchart illustrating the method for predicting the fatigue damage evolution of the drive shaft in this application.

[0067] 201. Obtain the three-dimensional configuration data, preset operating speed frequency, and torque load spectrum of the drive shaft. The three-dimensional configuration data includes three-dimensional geometric dimension data and material mechanical property data. (Refer to the execution process of step 101) 202. Construct a three-dimensional solid model of the drive shaft based on the three-dimensional geometric dimension data in the three-dimensional configuration data, and discretize the three-dimensional solid model into a finite element mesh to obtain discrete mesh elements.

[0068] A three-dimensional solid model refers to a complete geometric model of a drive shaft generated in a computer environment based on three-dimensional geometric dimension data. This model contains all the spatial morphological features and boundary information of the drive shaft. Discrete mesh elements represent the basic computational units generated after the finite element mesh is discretized. Each discrete mesh element is composed of several connected nodes and has a definite shape function and interpolation characteristics.

[0069] Specifically, the prediction system first extracts 3D geometric dimension data from the 3D configuration data, reading key dimensional parameters such as the length, outer diameter, inner diameter, and wall thickness of the drive shaft. Based on the extracted 3D geometric dimension data, the prediction system constructs a 3D solid model of the drive shaft in a 3D modeling environment. This 3D solid model uses solid geometry representation to fully describe the geometric features of the drive shaft, including its cylindrical main body, conical transition section, and end connection structure. The prediction system performs finite element mesh discretization on the constructed 3D solid model, generating meshes in the internal space and outer surface of the drive shaft's 3D solid model, dividing the continuous solid model into multiple interconnected discrete mesh elements. The prediction system determines the mesh density and element type based on the geometric complexity of the drive shaft and the required computational accuracy, using a denser mesh in areas with large curvature changes and a sparser mesh in areas with small stress gradients. The prediction system uses a mesh generation algorithm to arrange node positions in the 3D solid model and connects adjacent nodes to form discrete mesh elements. Each discrete mesh element contains information such as node coordinates, element number, and element type, resulting in a complete finite element mesh model.

[0070] 203. Calculate the material stiffness properties of each discrete mesh element based on the material mechanical property data, and calculate the mass distribution properties of each discrete mesh element in combination with the three-dimensional geometric dimension data. The material mechanical property data shall include at least the elastic modulus, shear modulus and Poisson's ratio.

[0071] The elastic modulus is the ratio of stress to strain in a material during the elastic deformation stage, characterizing the material's ability to resist elastic deformation; the shear modulus is the ratio of shear stress to shear strain in a material under shear load, characterizing the material's ability to resist shear deformation; Poisson's ratio is the ratio of lateral strain to axial strain in a material under axial tension or compression, characterizing the material's lateral deformation characteristics; the material stiffness property represents the ability of discrete mesh elements to resist deformation, expressed mathematically through the element stiffness matrix; the mass distribution property represents the mass of discrete mesh elements and its spatial distribution characteristics, expressed mathematically through the element mass matrix.

[0072] Specifically, the prediction system extracts material mechanical property data from the 3D configuration data, reading material parameters such as elastic modulus, shear modulus, Poisson's ratio, and density of the drive shaft material. Based on the material mechanical property data, the prediction system calculates the material stiffness properties of each discrete mesh element. For each discrete mesh element, the prediction system constructs an element strain-displacement matrix according to the element's geometry and node coordinates. Then, based on the material constitutive relation, it assembles the elastic modulus, shear modulus, and Poisson's ratio into a material elasticity matrix. By multiplying the transpose of the element strain-displacement matrix with the material elasticity matrix and the element strain-displacement matrix and integrating, the system obtains the element stiffness matrix of the discrete mesh element. This element stiffness matrix represents the material stiffness property of the discrete mesh element.

[0073] The prediction system combines three-dimensional geometric dimension data to calculate the mass distribution attributes of each discrete mesh element. For each discrete mesh element, the prediction system calculates the element volume based on the node coordinates of the element, multiplies the element volume by the material density to obtain the element mass, and then distributes the element mass to each node according to the element shape function. The element mass matrix of the discrete mesh element is obtained by multiplying and integrating the element shape functions. The element mass matrix represents the mass distribution attributes of the discrete mesh element.

[0074] In some embodiments, the prediction system calculates the material stiffness and mass distribution properties as follows: First, the prediction system selects appropriate Gaussian integration point locations and weighting coefficients based on the element type. At each Gaussian integration point, it calculates the element strain-displacement matrix and the material elasticity matrix. The stiffness contribution at each integration point is obtained through matrix multiplication. Finally, the stiffness contributions from all integration points are summed according to the weighting coefficients to obtain the element stiffness matrix. This calculation process is performed using formula K. E =∫B T *D*BdV implementation, where K E B is the element stiffness matrix, B is the strain-displacement matrix, D is the material elasticity matrix, dV is the volume element, T represents the matrix transpose operation, and B... T This represents the transpose of the strain-displacement matrix B.

[0075] 204. Perform global matrix integration on the material stiffness properties and mass distribution properties of each discrete mesh element to obtain the global mass matrix and global stiffness matrix.

[0076] Global matrix integration refers to the process of assembling the element matrices of all discrete grid cells into a global system matrix according to the node connection relationships.

[0077] Specifically, the prediction system performs global matrix integration of the material stiffness properties of each discrete mesh element. First, the global stiffness matrix is ​​initialized as a zero matrix, with the number of rows and columns equal to the total degrees of freedom (DOF) of the drive shaft finite element model. The total DDF equals the total number of nodes multiplied by the DDF of each node. The prediction system iterates through each discrete mesh element, extracting its element stiffness matrix and node number. Based on the node number, a mapping relationship is established between the local DDF numbers and the global DDF numbers. The elements of the element stiffness matrix are then superimposed onto the corresponding positions in the global stiffness matrix according to this mapping relationship. This superposition process ensures that the stiffness contribution at shared nodes is correctly accumulated. After completing the stiffness matrix integration of all discrete mesh elements, the prediction system obtains a complete global stiffness matrix. This global stiffness matrix is ​​a symmetric sparse matrix that reflects the overall stiffness characteristics of the drive shaft. The prediction system uses the same global matrix integration method to process the mass distribution attributes of each discrete grid cell. The global mass matrix is ​​initialized as a zero matrix. Each discrete grid cell is traversed, and the cell mass matrix is ​​superimposed onto the corresponding position of the global mass matrix according to the node number mapping relationship. After the mass matrix integration of all cells is completed, a complete global mass matrix is ​​obtained. This global mass matrix is ​​also a symmetric sparse matrix, reflecting the overall inertial characteristics of the drive shaft.

[0078] 205. Extract the current torque load sequence from the torque load spectrum according to the preset time step size, and calculate the current local stress state parameters of the drive shaft under the current torque load sequence (refer to the execution process of step 102). 206. Based on the current local stress state parameters, extract the principal stress components and shear stress components of each discrete mesh element under the current loading state.

[0079] The current loading state refers to the stress state of the drive shaft under the action of the torque load sequence within the current time step; the principal stress components represent the normal stress values ​​in the three orthogonal directions after the stress tensor is transformed by the principal axis, namely the first principal stress, the second principal stress, and the third principal stress, and the principal stress components correspond to the principal plane where the shear stress is zero; the shear stress components represent the tangential stress values ​​acting inside the element, including the shear stress components on the three coordinate planes.

[0080] Specifically, the prediction system reads the stress tensor of each discrete mesh element from the current local stress state parameters. This stress tensor contains three normal stress components and three shear stress components, represented as a three-dimensional stress state in matrix form. The prediction system calculates the principal stresses for the stress tensor of each discrete mesh element, extracting the principal stress components by solving the eigenvalue problem of the stress tensor. This calculation process establishes the characteristic equation det(σ-λI)=0, where det represents the mathematical operator for determinant, referring to a scalar value obtained after performing a specific algebraic operation on a square matrix (here, the difference matrix between the stress tensor matrix and the eigenvalue multiplied by the identity matrix, i.e., σ-λI). σ is the stress tensor matrix, λ is the eigenvalue, and I is the identity matrix. Solving this cubic equation yields three eigenvalues, which are the three principal stress components. The prediction system sorts the principal stress components according to their numerical values, designating the largest principal stress as the first principal stress component, the middle principal stress as the second principal stress component, and the smallest principal stress as the third principal stress component. The prediction system extracts shear stress components from the stress tensor and obtains shear stress components on three coordinate planes by reading the off-diagonal elements of the stress tensor matrix, including shear stress in the xy plane, yz plane, and xz plane. The prediction system stores the extracted principal stress components and shear stress components in the stress feature data structure of each discrete mesh element, providing stress criteria for subsequent damage assessment.

[0081] 207. Discrete mesh elements whose principal stress components exceed the fatigue tensile strength threshold in the preset material fatigue degradation benchmark parameters, or whose shear stress components exceed the fatigue shear strength threshold in the material fatigue degradation benchmark parameters, are identified as damaged mesh elements. The corresponding local damage evolution factor is calculated based on the stress amplitude exceeding the limit of each damaged mesh element.

[0082] The fatigue tensile strength threshold refers to the critical stress value at which a material begins to accumulate fatigue damage under tensile stress; the fatigue shear strength threshold refers to the critical stress value at which a material begins to accumulate fatigue damage under shear stress; a damaged mesh element represents a discrete mesh element whose stress level exceeds the fatigue strength threshold, thus causing fatigue damage; the excess stress amplitude represents the difference between the actual stress value of the damaged mesh element and the corresponding fatigue strength threshold; the local damage evolution factor is a dimensionless parameter that quantifies the degree of fatigue damage accumulation of a single discrete mesh element within the current time step, and the larger the factor value, the more severe the damage accumulation.

[0083] Specifically, the prediction system reads the fatigue tensile strength threshold and fatigue shear strength threshold from the material fatigue degradation benchmark parameters. These two thresholds are determined based on fatigue test data of carbon fiber composites. The prediction system traverses all discrete mesh elements and performs damage assessment on each element. First, it compares the first principal stress component of the element with the fatigue tensile strength threshold to determine if the principal stress component is greater than the fatigue tensile strength threshold. Then, it compares the maximum shear stress component of the element with the fatigue shear strength threshold to determine if the shear stress component is greater than the fatigue shear strength threshold. The prediction system identifies discrete mesh elements that satisfy either the condition that the principal stress component is greater than the fatigue tensile strength threshold or the shear stress component is greater than the fatigue shear strength threshold as damaged mesh elements, and marks elements that do not satisfy either condition as undamaged elements. For each damaged mesh element, the prediction system calculates the excess stress amplitude. For tensile damaged elements, it calculates the difference between the principal stress component and the fatigue tensile strength threshold; for shear damaged elements, it calculates the difference between the shear stress component and the fatigue shear strength threshold. The prediction system calculates the corresponding local damage evolution factor based on the stress amplitude exceeding the limit of each damaged mesh element. It uses a fatigue damage accumulation model to convert the stress amplitude into damage increment. This local damage evolution factor represents the stiffness degradation ratio of the element within the current time step.

[0084] In some embodiments, the prediction system first obtains the stress amplitude exceeding the limit of the damaged mesh element, adds the excess stress amplitude to the fatigue strength threshold to obtain the equivalent peak stress of the element, then queries the SN curve of the material based on the equivalent peak stress to obtain the fatigue life number N corresponding to the stress level, then records the number of load cycles in the current time step as n, calculates the local damage evolution factor D=n / N using the Miner rule, which represents the proportion of the newly consumed lifetime in the current time step, and finally converts the local damage evolution factor into the local stiffness reduction factor α=D for the current step, which is used for subsequent material stiffness orientation reduction.

[0085] 208. The material stiffness properties of the corresponding discrete mesh elements are directionally reduced by each local damage evolution factor to obtain the local stiffness reduction matrix of each damaged mesh element.

[0086] Material stiffness properties represent the ability of discrete mesh elements to resist deformation. They are expressed mathematically as element stiffness matrices. During multi-round iterative calculations with progressively increasing time steps, the material stiffness properties of each discrete mesh element are dynamically updated with the iteration process. In the first iteration, it is the original material stiffness property obtained from the initial calculation based on the material mechanical property data in step 203. In subsequent iterations, it is the current material stiffness property after cumulative reduction and updating from previous iterations. That is, the directional reduction operation always acts on the material stiffness property that reflects the current damage state. Directional reduction refers to the process of selectively reducing different directional components of the material stiffness property according to the damage type and degree. The local stiffness reduction matrix represents the amount of stiffness reduction caused by fatigue damage in the damaged mesh element. The dimension of this matrix is ​​the same as that of the element stiffness matrix.

[0087] Specifically, the prediction system first extracts the material stiffness properties (i.e., the current element stiffness matrix) and the corresponding local damage evolution factor of the damaged mesh elements in the current iteration state. The current material stiffness properties are the material stiffness properties initially calculated based on the material mechanical property data in step 203 (i.e., the initial element stiffness matrix) during the first iteration. In subsequent iterations, they are the updated material stiffness properties (i.e., the updated element stiffness matrix) that have been reduced and retained in memory after previous iterations. The prediction system determines the stiffness retention coefficient based on the local damage evolution factor (the formula is 1 minus the local damage evolution factor).

[0088] Since this embodiment uses the overall reduction of isotropic materials as an example, in some embodiments, the prediction system directly converts the local damage evolution factor D into a stiffness reduction coefficient α=D, and extracts the element stiffness matrix K corresponding to the current material stiffness property. E Multiply all relevant elements by the coefficient α to directly calculate the local stiffness reduction matrix ΔK resulting from the current time step. E =α×K E Simultaneously, the material stiffness properties of the element are updated to the reduced material stiffness properties (i.e., the reduced element stiffness matrix K). E = (1-α)×K E This will be reserved for use in the next iteration. The prediction system repeats the above extraction and product reduction process for all damaged mesh elements to obtain a set of local stiffness reduction matrices for each damaged element, ensuring the engineering fact that stiffness is continuously stripped away and attenuated.

[0089] In some optional embodiments, the drive shaft can be made of metallic materials (such as isotropic materials like high-strength alloy steel, aluminum alloy, and titanium alloy) or anisotropic materials such as carbon fiber composites. This embodiment mainly uses the reduction of the overall stiffness matrix of isotropic materials as an example for illustration; when using anisotropic materials such as carbon fiber composites, it can be achieved by independently reducing the stiffness tensors in specific directions of the matrix (such as the fiber direction or the matrix direction), and the iterative calculation logic for the resonant frequency slip is consistent with that of this embodiment.

[0090] 209. Map each local stiffness reduction matrix to the global coordinate system and perform matrix integration to obtain the stiffness reduction matrix parameters within the time step.

[0091] The global coordinate system refers to the unified spatial coordinate system used in the finite element model of the drive shaft, under which all node positions and displacements are defined; the stiffness reduction matrix parameter represents the overall stiffness reduction of the entire drive shaft system due to fatigue damage accumulation within the current time step, and the dimension of this matrix is ​​the same as that of the global stiffness matrix.

[0092] Specifically, the prediction system first performs coordinate system mapping on the local stiffness reduction matrix of each damaged mesh element. Since the local stiffness reduction matrix is ​​defined in the element's local coordinate system, it needs to be transformed to the global coordinate system before assembly. The prediction system extracts the transformation matrix from the local coordinate system to the global coordinate system for the damaged mesh element. This transformation matrix is ​​determined by the element's node coordinates and element orientation, and is constructed by calculating the direction cosines of the element's local coordinate axes in the global coordinate system. The prediction system uses the transformation matrix to perform a similarity transformation on the local stiffness reduction matrix. The calculation process is that the global stiffness reduction matrix is ​​equal to the transpose of the transformation matrix multiplied by the local stiffness reduction matrix, and then multiplied by the transformation matrix again, to obtain the stiffness reduction matrix of the damaged mesh element in the global coordinate system. The prediction system initializes the stiffness reduction matrix parameter as a zero matrix, which has the same dimensions as the global stiffness matrix. The prediction system traverses all damaged mesh elements, mapping the stiffness reduction matrix of each damaged mesh element in the global coordinate system to the corresponding position of the stiffness reduction matrix parameter according to the element node number. The matrix integration operation is completed by accumulating the matrix elements one by one. After the prediction system completes the matrix integration of all damaged mesh elements, it obtains the stiffness reduction matrix parameters within the time step.

[0093] 210. Subtract the stiffness reduction matrix parameter from the current global stiffness matrix to obtain the attenuated global stiffness matrix. The current global stiffness matrix is ​​the same as the global stiffness matrix in the first iteration. (Refer to the execution process of step 103) 211. Based on the attenuation global stiffness matrix and global mass matrix, solve for the matrix eigenvalues ​​to obtain the real-time natural frequency of the drive shaft under the current attenuation state. (Refer to the execution process of step 104) 212. Calculate the absolute frequency difference between the real-time natural frequency and the preset operating speed frequency. Based on the absolute frequency difference, calculate the resonance stress gain coefficient, which represents the resonance stress gain effect. (Refer to the execution process of step 105) 213. Utilize the resonant stress gain coefficient to compensate for the amplitude of the torque load in the next time step, and update the current global stiffness matrix based on the decaying global stiffness matrix, to perform multi-round progressive iterative calculations for each time step. (Refer to the execution process of step 106) 214. When the real-time natural frequency corresponding to the current iteration is detected to be greater than or equal to the preset frequency threshold, the torque load of the next time step after amplitude compensation is determined as the current torque load sequence.

[0094] The real-time natural frequency corresponding to the current iteration represents the natural frequency value of the drive shaft obtained by solving the attenuation global stiffness matrix in the current time step calculation; the torque load of the next time step after amplitude compensation represents the torque load numerical sequence after correction by the resonant stress gain coefficient; the current torque load sequence refers to the torque load data applied to the drive shaft for stress calculation within the current time step.

[0095] Specifically, the prediction system will use the real-time inherent frequency f corresponding to the current iteration. c With preset frequency threshold f t The comparison is made, and the judgment condition is f. c ≥f t When the condition is met, it indicates that the stiffness degradation of the drive shaft has not reached the failure criterion, and it can continue to operate under load. The prediction system performs a load sequence update operation, assigning the compensated torque load of the next time step to the current torque load sequence. This operation is performed using formula T. c =T n c Implementation, where T c For the current torque load sequence, T n c This is the torque load at the next time step after amplitude compensation. This compensated load includes the amplification effect of the resonant stress gain coefficient β, as shown in the formula T. n c =β*T n o T n o This represents the original torque load amplitude. The prediction system uses this operation to ensure that the next iteration uses a corrected load that reflects the resonance effect, incorporating the dynamic stress amplification effect into the simulation.

[0096] 215. Accumulate the current time series step size to update the current cumulative total time series duration.

[0097] The current time step size represents the length of time spanned by this round of iteration calculation. This time step size may have been dynamically adjusted by an adaptive shrinkage mechanism. The current cumulative total time duration represents the total length of time experienced from the start of fatigue damage evolution deduction to the current round of iteration.

[0098] Specifically, the prediction system extracts the current time step t. c The prediction system reads the current cumulative total time series duration T. o This duration records the time from the initial time t s The sum of all time steps completed up to the current time. The prediction system performs an accumulation operation, adding the current time step to the current cumulative total time duration, calculated as T. n =T o +t c T n For the updated cumulative total time series duration, T o The cumulative total time series duration before accumulation, t c This is the current time step. The prediction system will calculate T. n The value is assigned to the cumulative total timing duration variable, completing the timing duration update operation. This updated cumulative total timing duration is used to record the cumulative fatigue damage duration experienced by the drive shaft. When the iteration terminates, the prediction system extracts the final cumulative total timing duration as the predicted fatigue failure life output. This process ensures the continuity of timing simulation.

[0099] 216. Calculate the current local stress state parameters of the drive shaft under the current torque load sequence, so as to continue to perform multi-round time step progressive iteration calculation.

[0100] The current local stress state parameter represents the set of stress tensors of each discrete mesh element of the drive shaft under the current torque load sequence; multi-round time-step progressive iterative calculation refers to the cyclic process of progressively advancing and repeatedly performing damage accumulation and stiffness update calculations in chronological order.

[0101] Specifically, the prediction system converts the current torque load sequence Tc into an equivalent nodal force vector F applied to the boundary nodes of the drive shaft end face. By mapping the torque Tc according to the polar coordinate distribution of the boundary nodes, the calculation formula is decomposed into F. i =[Tc / ∑(r i ²)]×r i , where r i Let F be the radial span vector of the i-th boundary node from the geometric center of the drive shaft. i To accurately assign the equivalent tangential force components to this node, the equivalent nodal force vector F required for structural derivation is generated. The prediction system is based on the updated global stiffness matrix K. u Establish the finite element equilibrium equations, in the form Ku *u=F, where u is the nodal displacement vector. The prediction system solves this system of equations to obtain the displacement vector u. Then, it calculates the strain vector ε based on the element strain-displacement matrix B, with the formula ε=B*v, where v is the element nodal displacement vector. The prediction system converts the strain vector into a stress vector σ through the material constitutive relation, with the formula σ=D*ε, where D is the material elastic matrix. This vector constitutes the current local stress state parameter. The prediction system uses the current local stress state parameter as the basic input for the next round of iteration calculation, returns to step 206, and continues to perform operations such as extracting principal stress components and shear stress components, damage determination, and stiffness reduction based on the current updated state, thereby forming a logically closed time-step progressive iterative loop.

[0102] 217. When the real-time natural frequency corresponding to the current iteration is detected to be less than the preset frequency threshold, the iteration calculation is terminated, the current cumulative total time series duration is extracted as the predicted fatigue failure life, and an early warning signal corresponding to the predicted fatigue failure life is generated. (Refer to the execution process of step 107) The drive shaft fatigue damage evolution prediction method described in this application first constructs a three-dimensional solid model based on three-dimensional configuration data and completes finite element mesh discretization. It then calculates the material stiffness and mass distribution properties of each discrete mesh element and integrates them into a global matrix. Next, the prediction system extracts the principal stress and shear stress components of the mesh elements, identifies damaged mesh elements, calculates the local damage evolution factor, and integrates the global stiffness reduction matrix parameters after directional stiffness reduction, achieving a precise conversion from microscopic damage to macroscopic stiffness attenuation. Then, the prediction system solves for the real-time natural frequency and calculates the resonant stress gain coefficient. After load amplitude compensation, it updates the global stiffness matrix and iterates, adaptively shrinking the time step size according to frequency changes to ensure the prediction closely matches the actual fatigue evolution law. Finally, the prediction system dynamically updates the load, stiffness, and stress during iteration until the real-time natural frequency reaches a threshold, terminating the iteration and extracting the failure life. This method alleviates the shortcomings of fixed step size and fixed load predictions, making the predicted fatigue failure life more consistent with actual working conditions.

[0103] The methods provided in the above embodiments can be executed by a prediction system. The prediction system in the embodiments of this invention is described below from a hardware processing perspective; please refer to [link to relevant documentation]. Figure 3 This is a schematic diagram of the physical device structure of the prediction system in an embodiment of this application.

[0104] It should be noted that, Figure 3 The structure of the prediction system shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0105] like Figure 3As shown, the prediction system includes a Central Processing Unit (CPU) 301, which can perform various appropriate actions and processes based on a program stored in Read-Only Memory (ROM) 302 or a program loaded from storage portion 308 into Random Access Memory (RAM) 303, such as performing the methods described in the above embodiments. Various programs and data required for system operation are also stored in RAM 303. The CPU 301, ROM 302, and RAM 303 are interconnected via bus 304. An Input / Output (I / O) interface 305 is also connected to bus 304.

[0106] The following components are connected to I / O interface 305: input section 306 including audio input devices, push-button switches, etc.; output section 307 including a liquid crystal display (LCD) and audio output devices, indicator lights, etc.; storage section 308 including a hard disk, etc.; and communication section 309 including a network interface card such as a LAN (Local Area Network) card, modem, etc. Communication section 309 performs communication processing via a network such as the Internet. Drive 310 is also connected to I / O interface 305 as needed. Removable media 311, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 310 as needed so that computer programs read from them can be installed into storage section 308 as needed.

[0107] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing computer programs for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 309, and / or installed from removable medium 311. When the computer program is executed by central processing unit (CPU) 301, it performs the various functions defined in the present invention.

[0108] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0109] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. Each block in a flowchart or block diagram may represent a module, program segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those shown in the drawings.

[0110] Specifically, the prediction system in this embodiment includes a processor and a memory. The memory stores a computer program, and when the computer program is executed by the processor, it implements the drive shaft fatigue damage evolution prediction method provided in the above embodiment.

[0111] In another aspect, the present invention also provides a computer-readable storage medium, which may be included in the prediction system described in the above embodiments; or it may exist independently and not assembled into the prediction system. The storage medium carries one or more computer programs, which, when executed by a processor of the prediction system, cause the prediction system to implement the drive shaft fatigue damage evolution prediction method provided in the above embodiments.

[0112] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0113] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as meaning "if...", "after...", "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if (the stated condition or event) is interpreted as meaning "if determining...", "in response to determining...", "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

[0114] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0115] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for predicting the fatigue damage evolution of a drive shaft, characterized in that, include: The three-dimensional configuration data, preset operating speed frequency, and torque load spectrum of the drive shaft are obtained. Finite element mesh discretization is performed based on the three-dimensional configuration data to generate a global mass matrix and a global stiffness matrix. The three-dimensional configuration data includes three-dimensional geometric dimension data and material mechanical property data. The current torque load sequence is extracted from the torque load spectrum according to the preset time step size, and the current local stress state parameter of the drive shaft under the action of the current torque load sequence is calculated. Based on the current local stress state parameters and the preset material fatigue degradation benchmark parameters, the stiffness reduction matrix parameters within the time step are calculated. The stiffness reduction matrix parameters are subtracted from the current global stiffness matrix to obtain the attenuated global stiffness matrix. The current global stiffness matrix is ​​the global stiffness matrix in the first iteration. The real-time natural frequency of the drive shaft in the current attenuation state is obtained by solving the matrix eigenvalues ​​based on the attenuation global stiffness matrix and the global mass matrix. Calculate the absolute frequency difference between the real-time natural frequency and the preset operating speed frequency, and calculate the resonance stress gain coefficient based on the absolute frequency difference. The resonance stress gain coefficient represents the resonance stress gain effect. The amplitude compensation of the torque load in the next time step is performed using the resonant stress gain coefficient, and the current global stiffness matrix is ​​updated based on the attenuated global stiffness matrix to perform multi-round time step progressive iterative calculation. When the real-time inherent frequency corresponding to the current iteration is detected to be less than the preset frequency threshold, the iteration calculation is terminated, the current cumulative total time series duration is extracted as the predicted fatigue failure life, and an early warning signal corresponding to the predicted fatigue failure life is generated.

2. The method according to claim 1, characterized in that, Based on the aforementioned three-dimensional configuration data, finite element mesh discretization is performed to generate a global mass matrix and a global stiffness matrix, specifically including: A three-dimensional solid model of the drive shaft is constructed based on the three-dimensional geometric dimension data in the three-dimensional configuration data. The three-dimensional solid model is then discretized using the finite element mesh to obtain discrete mesh elements. The material stiffness properties of each discrete mesh element are calculated based on the material mechanical property data, and the mass distribution properties of each discrete mesh element are calculated in combination with the three-dimensional geometric dimension data. The material mechanical property data includes at least the elastic modulus, shear modulus, and Poisson's ratio. The material stiffness properties and mass distribution properties of each discrete mesh element are integrated globally to obtain the global mass matrix and the global stiffness matrix.

3. The method according to claim 2, characterized in that, Based on the current local stress state parameters and the preset material fatigue degradation benchmark parameters, the stiffness reduction matrix parameters within the time step are calculated, specifically including: Based on the current local stress state parameters, extract the principal stress components and shear stress components of each discrete mesh element under the current loading state; Discrete mesh elements whose principal stress component is greater than the fatigue tensile strength threshold in the preset material fatigue degradation benchmark parameters, or whose shear stress component is greater than the fatigue shear strength threshold in the material fatigue degradation benchmark parameters, are identified as damaged mesh elements, and the corresponding local damage evolution factor is calculated based on the stress amplitude exceeding the limit of each damaged mesh element. The material stiffness properties of the corresponding discrete mesh elements are directionally reduced using the local damage evolution factors to obtain the local stiffness reduction matrix of each damaged mesh element. The local stiffness reduction matrices are mapped to the global coordinate system and integrated to obtain the stiffness reduction matrix parameters within the time step.

4. The method according to claim 1, characterized in that, The amplitude of the torque load in the next time step is compensated using the resonant stress gain coefficient, and the current global stiffness matrix is ​​updated based on the attenuated global stiffness matrix to perform multi-round progressive iterative calculations in time steps, specifically including: Extract the initial torque load amplitude of the next time step from the torque load spectrum, and multiply the initial torque load amplitude by the resonant stress gain coefficient to obtain the compensated torque load sequence; The decayed global stiffness matrix is ​​updated to the current global stiffness matrix of the next time step to obtain the updated global stiffness matrix; Under the action of the compensated torque load sequence, the current local stress state parameters of the drive shaft are calculated based on the updated global stiffness matrix; The attenuation global stiffness matrix, the real-time natural frequency, the resonant stress gain coefficient, and the updated global stiffness matrix are repeatedly calculated to perform multiple rounds of time-step progressive iterative calculations.

5. The method according to claim 4, characterized in that, After the step of repeatedly calculating the attenuation global stiffness matrix, the real-time natural frequency, the resonant stress gain coefficient, and the updated global stiffness matrix, the method further includes: Calculate the frequency drop difference between the real-time intrinsic frequency corresponding to the current iteration and the real-time intrinsic frequency corresponding to the previous iteration; Divide the frequency decrease difference by the current timing step size to obtain the current frequency decrease gradient; When the absolute frequency difference is less than a preset resonance threshold and the current frequency descent gradient is greater than a preset gradient threshold, the gradient threshold is divided by the current frequency descent gradient to obtain the step size contraction coefficient. Multiply the current time step size by the step size contraction coefficient to obtain the updated next time step size, and perform the multi-round time step size progressive iteration calculation according to the updated next time step size.

6. The method according to claim 1, characterized in that, After the step of updating the current global stiffness matrix based on the attenuated global stiffness matrix to perform multi-round time-step progressive iterative calculation, the method further includes: When the real-time inherent frequency corresponding to the current iteration is detected to be greater than or equal to the preset frequency threshold, the torque load of the next time step after amplitude compensation is determined as the current torque load sequence. Accumulate the current timing step size to update the current cumulative total timing duration; Calculate the current local stress state parameters of the drive shaft under the current torque load sequence to continue the multi-round time step progressive iteration calculation.

7. The method according to claim 1, characterized in that, The resonant stress gain coefficient is calculated based on the absolute frequency difference, specifically including: The structural stiffness loss ratio is determined based on the numerical change between the global stiffness matrix and the attenuation global stiffness matrix. Extract initial structural dissipation parameters from the material mechanical property data of the three-dimensional configuration data; The initial structural dissipation parameters are adjusted by increasing the ratio of structural stiffness loss to obtain the current total structural dissipation parameters. The resonant amplification factor is determined using the absolute frequency difference, and the resonant amplification factor is numerically reduced using the current total structural dissipation parameter to obtain the resonant stress gain coefficient.

8. A prediction system, characterized in that, Includes one or more processors and memory; The memory is coupled to the one or more processors, the memory being used to store computer program code, the computer program code including computer instructions, the one or more processors invoking the computer instructions to cause the prediction system to perform the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising computer instructions, characterized in that, When the computer instructions are run on the prediction system, the prediction system performs the method as described in any one of claims 1-7.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are run on the prediction system, the prediction system performs the method as described in any one of claims 1-7.

Images