Modeling analysis method for turbomachinery center-pole rotor based on rough contact interface

Abstract

The application discloses a kind of based on rough contact interface turbomachinery center pull rod rotor modeling analysis method, belong to turbomachinery high temperature rotor technical field.The present application includes calling pre-trained contact stiffness prediction model, input current operating condition roughness parameter and load parameter, obtain the predicted stiffness of contact interface;The three-dimensional finite element model of rotor is established, the mesh division is carried out to contact interface area, and nonlinear spring damping unit is set between the corresponding nodes of contact interface;The predicted stiffness obtained is given to the nonlinear spring damping unit, and the reduced degrees of freedom are carried out to the finite element model after stiffness assignment using hybrid modal reduction method, and the reduced model is solved for vibration response, the fidelity of the model is ensured, the full model solving degree of freedom is greatly compressed by hybrid reduction method, while not affecting the accuracy of the model, reduce the amount of calculation, facilitate the analysis and solution of model.

Description

Technical Field

[0001] This invention relates to the field of high-temperature rotor technology for turbine machinery, and specifically to a modeling and analysis method for a central tie rod rotor of turbine machinery based on a rough contact interface. Background Technology

[0002] In the field of turbomachinery, center-tied-rod rotor systems are widely used in high-end equipment such as aero-engines and gas turbines due to their compact structure and high load-bearing capacity. These rotor systems contain numerous disk contact interfaces (such as Hirth tooth connections). The contact stiffness of these interfaces exhibits strong nonlinearity and randomness due to their surface roughness, material properties, and load conditions, significantly affecting the rotor's dynamic characteristics. Therefore, accurately simulating contact stiffness is crucial for predicting the rotor's vibration response, critical speed, and operational stability, and is a prerequisite for high-fidelity dynamic analysis.

[0003] Currently, modeling methods in this field mainly face the contradiction of balancing accuracy and efficiency, which manifests itself on two levels: First, at the level of physical modeling of contact interfaces, there is a dilemma between high precision and low efficiency versus low precision and high efficiency. On the one hand, methods based on the GW statistical model and the KE elastoplastic micro-convexity model can take surface roughness into account and theoretically obtain contact stiffness with high accuracy. However, their calculation process is complex, involving a large number of numerical integration operations, and the cost of a single stiffness calculation is high, making it difficult to apply to design optimization or parameter sensitivity analysis that requires rapid iteration, resulting in poor engineering practicality. On the other hand, in pursuit of computational efficiency, most engineering studies adopt highly simplified models, such as simplifying the complex wheel contact surface into an ideal linear spring or rigid connection. Although this simplification significantly reduces computational complexity, it completely ignores the nonlinearity of rough contact, load correlation, and possible contact separation effects, leading to serious inaccuracies in the estimation of the overall system stiffness. It is difficult to accurately analyze its dynamic behavior, especially when predicting nonlinear vibrations, higher-order modes, and complex dynamic phenomena caused by changes in contact state, resulting in large errors and low model fidelity.

[0004] Secondly, at the level of solving the dynamics of the entire system, even if relatively accurate contact stiffness is obtained, the extremely low solution efficiency is still encountered when establishing a full three-dimensional finite element model to pursue geometric fidelity due to the huge degrees of freedom of the model. Directly performing transient dynamic analysis on a full three-dimensional model containing nonlinear contact consumes huge computational resources and takes a very long time, which is almost impractical in current engineering practice.

[0005] In summary, existing technologies for modeling turbine rotors with central tie rods suffer from systemic shortcomings that are difficult to reconcile in terms of model fidelity (due to simplification and neglect of the effects of roughness and nonlinearity), computational efficiency (the time required for high-precision stiffness calculations versus the time required for solving the entire model), and engineering practicality. Therefore, developing a high-fidelity modeling and rapid analysis method that can fundamentally balance model accuracy, computational efficiency, and engineering convenience has become a long-standing and urgent key technological need in this field. Summary of the Invention

[0006] The purpose of this invention is to overcome the problems in the prior art and provide a modeling and analysis method for turbine central tie rod rotors based on rough contact interfaces.

[0007] The present invention provides a method for modeling and analyzing the rotor of a central tie rod in turbine machinery based on a rough contact interface, including... A pre-trained contact stiffness prediction model is invoked, and the roughness parameters and load parameters of the current working condition are input to obtain the predicted stiffness of the contact interface. The pre-trained contact stiffness prediction model is a micromechanical model of a rough contact interface based on elastoplastic contact theory and statistical theory. It is obtained by generating a sample dataset by changing the roughness parameters, material properties and normal load, and using the dataset to train a machine learning model. A three-dimensional finite element model of the rotor is established, and the contact interface region is meshed. Nonlinear spring damping elements are set between corresponding nodes of the contact interface. The obtained predicted stiffness is assigned to the nonlinear spring damping elements. The degree of freedom of the finite element model after stiffness assignment is reduced by a hybrid modal order reduction method, and the vibration response of the reduced model is solved.

[0008] Preferably, when training the machine learning model, the dataset is divided into a training set, a validation set, and a test set. The model is trained using the training set and the mean squared error is used as the loss function. Hyperparameters are tuned using the validation set to suppress overfitting.

[0009] Preferably, the full three-dimensional finite element model uses hexahedral elements for meshing, and the mesh is locally refined in the contact interface area, while ensuring that the mesh nodes on both sides of the contact interface correspond one-to-one.

[0010] Preferably, the force-displacement relationship of the nonlinear spring damping unit is defined by the stiffness curve output by the contact stiffness prediction model, and the tensile strength of the nonlinear spring damping unit is set to zero or close to zero to simulate the instantaneous separation and re-contact behavior of the contact interface during vibration.

[0011] Preferably, the hybrid modal reduction method includes: reducing the degrees of freedom of the linear structural part in the rotor model using the fixed interface modal synthesis method, while retaining its low-order principal modes, wherein the linear structural part includes the wheel disk, shaft segment, and bearing support structure; retaining all physical degrees of freedom of the contact interface nodes using the nonlinear modal substructure method; and coupling the reduced linear modal coordinates with the nonlinear interface degrees of freedom to assemble them into the reduced-order system dynamic equations, wherein the nonlinear interface degrees of freedom correspond to all node pairs with spring-damped units.

[0012] The present invention also proposes an analysis system based on the above analysis method, the system comprising: The stiffness prediction module is used to call a pre-trained contact stiffness prediction model and output the predicted stiffness based on the input roughness parameters and load parameters. The modeling and element setting module is used to establish a three-dimensional finite element model of the rotor and set nonlinear spring damping elements between corresponding nodes of the contact interface. A stiffness assignment module is used to assign the predicted stiffness output by the stiffness prediction module to the nonlinear spring damping unit. The solver module is used to perform dynamic solutions on the stiffened finite element model using a hybrid modal order reduction method to obtain the vibration response.

[0013] Preferably, the solution module includes a linear substructure reduction unit, a nonlinear interface processing unit, and a solution unit. The linear substructure reduction unit is used to reduce the degrees of freedom of the disk, shaft segment, and bearing support structure in the rotor model using the fixed interface modal synthesis method, while retaining their low-order principal modes. The nonlinear interface processing unit is used to retain all physical degrees of freedom of the contact interface nodes and extract their nonlinear dynamic characteristics. The solution unit is also used to couple the reduced linear modal coordinates with the nonlinear interface degrees of freedom, assemble and solve the dynamic equations of the reduced system.

[0014] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.

[0015] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method.

[0016] Compared with the prior art, the beneficial effects of the present invention are: This invention proposes a vibration analysis model for a high-temperature rotor disk of a turbine machine with a full three-dimensional multi-scale contact interface based on elastoplastic contact theory and a contact model based on statistical theory. It eliminates the need to simplify the tie rod rotor, maximizes the consideration of the influence of micro-discontinuities of the rough interface on the motion state of the model, ensures a high degree of consistency between the model and the solid, ensures the fidelity of the model, and avoids the errors caused by traditional simplification methods that ignore roughness, nonlinearity and contact separation effects. This invention leverages machine learning to reduce the calculation time for contact stiffness from hours to seconds, and significantly compresses the degrees of freedom of the entire model through a hybrid order reduction method, making previously infeasible high-fidelity transient dynamics analysis efficient and feasible. Furthermore, this invention adds the calculated contact stiffness to the spring-damped elements set on the model's contact surface, more easily and realistically considering the influence of rough surface contact stiffness on the motion state of the central tie rod rotor model. This makes it convenient and reliable for engineering applications, while simultaneously enabling rapid prediction and calculation of contact stiffness and efficient and accurate simulation of the impact of contact stiffness on the motion state of the central tie rod rotor. Meanwhile, the prediction module of this invention enables design engineers to obtain high-precision stiffness without in-depth knowledge of mechanical theory, significantly lowering the technical threshold. This invention resolves the contradiction in the field of turbine machinery where precision, efficiency, and practicality are difficult to balance, providing a revolutionary analytical tool for the refined design, optimization, and condition assessment of turbine rotors.

[0017] This invention builds upon traditional numerical calculation models to construct a rapid contact stiffness prediction model based on machine learning. Compared to the enormous time consumption caused by the repeated calls to complex calculation processes when dealing with complex model parameters in traditional numerical calculation models, the machine learning model used in this invention can achieve contact stiffness prediction in seconds, greatly accelerating design iteration and parameter scanning analysis. At the same time, design engineers do not need to have an in-depth understanding of complex contact mechanics theory; they only need to input key parameters to obtain high-precision stiffness prediction values, lowering the technical threshold and making it highly easy to use in engineering.

[0018] This invention addresses the difficulties posed by the high degrees of freedom and massive computational demands of traditional 3D model solving. Specifically, it employs the efficient fixed-interface modal synthesis method and the nonlinear modal substructure method to reduce the degrees of freedom of the linear and nonlinear parts of the model, respectively. This reduces computational load without compromising model accuracy, facilitating model analysis and solution. In conclusion, this invention has significant engineering implications and broad application prospects. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the overall process of a modeling and analysis method for a turbine central tie rod rotor based on a rough contact interface, according to the present invention.

[0020] Figure 2 A schematic diagram of the rough surface contact model of the contact interface of the central tie rod rotor.

[0021] Figure 3 This is a flowchart of the machine learning process.

[0022] Figure 4 This is a schematic diagram of the contact surface model; Figure 5 Finite element network partitioning for the contact surface.

[0023] Figure 6 A schematic diagram of adding a spring damping unit between the contact surfaces. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0025] Unless otherwise defined, the technical or scientific terms used herein shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0026] Reference Figure 1 The present invention provides a method for modeling and analyzing the rotor of a central tie rod in a turbine based on a rough contact interface, comprising the following steps: Based on elastoplastic contact theory and a contact mechanics model of a rough consolidated interface based on statistical theory, a high-precision calculation model for the contact interface stiffness of a central tie rod rotor is established. The contact stiffness is calculated and solved according to the rough surface parameters and load parameters. The rough surface parameters include the number of micro-protrusions per unit area, the average radius of the micro-protrusions, and the standard deviation of the micro-protrusion height.

[0027] At the microscopic scale, a rough surface can be considered as a surface composed of a series of micro-protrusions with different radii of curvature at their apex on a smooth surface. The distribution of the radii of curvature at the apex of these micro-protrusions can be assumed to follow a Gaussian distribution. Consider random, uniform, isotropic rough surfaces, such as... Figure 2 As shown, the Hirth tooth contact interface is used as an example for illustration. The tip of the micro-protrusion is approximately a radius of curvature of... A sphere, height This represents the distance between the vertex of the micro-protrusion and the average height line of the micro-protrusion. Under a normal load, the contact deformation of a single micro-protrusion on a rough surface is:

[0028] ; Where h represents the distance from the rigid plane to the average height line of the rough surface, and ys represents the distance from the average height line of the micro-convexity to the average height line of the rough surface; Based on the elastoplastic contact model, considering the deformation of a single micro-convex body, the formula for its elastoplastic normal contact load is as follows: Critical deformation amount ; For the elastic-dominant phase : ; For the elastic-plastic to fully plastic stage : ; Geometric hardness is calculated using the following formula: ; In the formula, Poisson's ratio, The yield strength of the material; It is the equivalent elastic modulus; The radius of the hemisphere; This is a dimensionless deformation quantity; The load is dimensionless. The area is dimensionless; This is a transition point.

[0029] By employing a contact mechanics model of a rough, consolidated interface based on statistical theory, the functional relationship between the total contact load and the dimensionless deformation can be obtained by integrating the normal contact loads of all micro-convexities involved in the contact: ; In the formula It is a dimensionless Gaussian distribution. , This is a parameter for surface roughness, expressed as the number of micro-protrusions per unit area, which can be measured by the contact surface topography. And it is calculated according to the surface roughness calculation formula. Here, is a surface roughness parameter, expressed as the standard deviation of the height of micro-protrusions per unit area. The standard deviation of the original surface profile height distribution.

[0030] Contact surface The contact morphology can be obtained by selecting three regions (upper, middle, and lower) on the tooth contact surface using a surface profilometer, and the overall roughness parameter can be calculated by taking the average value of multiple measurements.

[0031] On the element contact surface, it is assumed that the local roughness parameters and the global roughness parameters of the contact surface follow the same statistical distribution law. The element normal contact stiffness can be expressed as:

[0032] ; Tangential contact stiffness is calculated by the following formula: ; To achieve rapid prediction and calculation of contact stiffness in engineering applications, this invention establishes a rapid contact stiffness prediction model based on machine learning, such as... Figure 3 As shown, based on the above formula, a large amount of data was collected and calculated for the physical model. By changing input variables such as roughness parameters, material properties, and normal loads, the corresponding normal contact stiffness and tangential contact stiffness were calculated, thus obtaining tens of thousands of labeled data. This data was divided into training, validation, and test sets in a 7:1.5:1.5 ratio. The training set was used to train the model, which is based on a surrogate model of an artificial neural network; a multilayer perceptron architecture, including an input layer for the aforementioned input parameters, an output layer for dimensionless contact stiffness, and a hidden layer for complex nonlinear mapping relationships; and an optimization algorithm (such as Adam) was used to minimize the loss function (e.g., mean squared error, MSE) between the predicted value and the true value (calculated by the physical model). The validation set was used for hyperparameter tuning and to prevent overfitting. The mean squared error (MSE) of the contact stiffness is calculated as follows:

[0033] ; in, To verify the model-predicted normal contact stiffness in the i-th data set, To verify the normal contact stiffness in the i-th data set; To validate the model's prediction of tangential contact stiffness in the i-th data set, To verify the tangential contact stiffness in the i-th data set.

[0034] The trained, high-performance machine learning model is packaged into an independent, callable module. This module can be directly called when performing rotor dynamics analysis of the central tie rod of a turbine with a rough interface. The contact stiffness under the specified working condition can be obtained instantly, which greatly reduces the calculation time for complex models with multiple parameter changes and greatly improves engineering efficiency. At the same time, design engineers do not need to have an in-depth understanding of complex contact mechanics theory. They only need to input key parameters to obtain high-precision stiffness prediction values, which lowers the technical threshold and has high engineering usability.

[0035] Considering factors such as rotational speed and tie rod preload, a three-dimensional finite element model for analyzing and solving the rotor with the central tie rod is established.

[0036] Compared to the simplified operation of the wheel contact surface in traditional research center tie rod rotors, this invention employs a full three-dimensional finite element mesh to model the center tie rod rotor. A full hexahedral mesh is used, with special refinement of the contact area, ensuring that all network nodes on the contact surfaces correspond one-to-one, eliminating the need for simplification of the contact surfaces. Taking the Hirth tooth as an example... Figure 4 As shown, a full 3D, high-fidelity model of the rotor model of the central tie rod of a turbine with a rough contact interface is achieved.

[0037] A full three-dimensional finite element model was created using SOLIDWORK and APDL. Based on the established full three-dimensional finite element model, spring damping elements were added between the contact surfaces, and the calculated contact stiffness was averaged and added to each spring damping element.

[0038] Spring-damped elements, which provide nonlinear force-displacement relationships, are placed at the contact surface. These elements allow users to directly define function tables of force versus relative displacement or relative velocity, matching the nonlinear characteristics of contact stiffness. Figure 5 As shown, the spring damping elements should be connected one-to-one to the corresponding node pairs on the contact surface. This requires that during the preprocessing mesh generation, the mesh nodes on the contact surface must correspond one-to-one, such as... Figure 4 As shown.

[0039] A high-fidelity model needs to be able to simulate the instantaneous separation of the contact surface that may occur during vibration. This can be achieved automatically by setting the tensile strength of the spring element to zero or a minimum value: when the spring element is under tension, its internal force immediately becomes zero, naturally simulating the separation state of the contact surface. When the displacement recovers, the spring is compressed again, and the contact is re-established.

[0040] Furthermore, since too few spring damping elements will underestimate the gradient change of the contact state, while too many will significantly increase the computational cost without any accuracy gain, we can also perform convergence analysis on the number and distribution density of spring elements. We should ensure that there are sufficiently dense elements covering the critical contact stress area. The convergence criterion is the variable monitoring criterion, that is, select the key variable, calculate the relative error of the variable change under two different density distributions, and when the relative error is less than a convergence tolerance, it can be judged as convergent. Alternatively, the energy norm error criterion built into finite element software (such as ANSYS) can be used to determine convergence.

[0041] Finally, the previously calculated contact stiffness was added to the spring damping unit, which more easily and realistically simulated the influence of rough surface contact stiffness on the motion state of the central tie rod rotor model.

[0042] The fixed interface modal synthesis method and the nonlinear modal substructure method are used to reduce the degrees of freedom of the linear and nonlinear parts of the full three-dimensional finite element model. Then, combined with the previously established full three-dimensional finite element model, boundary conditions, and load conditions, the vibration response analysis and calculation of the central tie rod rotor are performed.

[0043] Specifically, when calculating the total nonlinear force on the contact interface, a nonlinear spring element represents a tiny pair of contact points on the contact surface. At each calculation time step, the normal relative displacement (δ) between the two corresponding points on the contact interface at the current moment is obtained from the reduced-order system dynamics solver. This displacement reflects the change in the contact surface gap due to overall deformation. Based on the material properties of the contact surface (such as elastic modulus E, Poisson's ratio v, yield strength σ_y) and surface topography parameters (such as the radius of curvature R of the micro-protrusions, the standard deviation of roughness σ), a complete nonlinear F-δ curve can be pre-calculated based on the previous physical model. After calculating the force of each nonlinear spring element on the contact interface, the forces of all spring elements are vector-synthesized (mainly algebraic summation in the normal direction) to obtain the total nonlinear force and nonlinear torque acting on the entire contact interface.

[0044] Modeling the system using a three-dimensional model results in a huge number of degrees of freedom, making it difficult to directly solve the dynamic equations. Therefore, degree-of-freedom reduction is necessary to simplify the calculation. Furthermore, since the model has both linear and nonlinear components, this invention employs the fixed-interface modal synthesis method and the nonlinear modal substructure method to reduce the degrees of freedom of the linear and nonlinear components respectively, while retaining some degrees of freedom for the contact surfaces, bearings, and wheel disks. The specific solution process is as follows:

[0045] Consider the system's equations of motion: ; In the formula It is a nonlinear force vector, mainly originating from the rough interface.

[0046] physical coordinates of the system Divided into internal linear coordinates and interface nonlinear coordinates (i.e., the contact interface node).

[0047] ; For nodes such as disks and bearings that are not directly connected to the contact surface, the influence of nonlinear terms is minimal. Therefore, we adopt the fixed-interface modal synthesis method. We fix the nonlinear coordinates of all interfaces (let...). ), calculate its fixed interface principal mode That is, solving the eigenvalue problem:

[0048] ; Preserve the first k lower-order modes The order k is determined based on the required level of refinement for the vibration response analysis of the actual model, typically ranging from 5 to 20. Too much k leads to unnecessary resource waste, while too little k may fail to achieve the desired level of refinement. Simultaneously, the constraint modes caused by unit displacement at the interface are calculated. :

[0049] ; The physical displacement of the linear substructure can be expressed in modal coordinates. and interface physical coordinates Approximately expressed as: ; The coordinate transformation relationship of the entire system is as follows: in, It is the generalized coordinate vector after condensation.

[0050] For the nodes at the contact interface, due to the strong nonlinear stiffness characteristics of the rough contact interface between the disks in the turbine's central tie rod rotor system, the nonlinear effects cannot be ignored, and nonlinear forces... Retained in interface coordinates Above. Transform the coordinates. Substitute into the original equation of motion and multiply on the left. Thus, the system equations after polycondensation are obtained;

[0051] ; In the formula: It is the mass, damping, and stiffness matrix after condensation.

[0052] Nonlinear terms Now it's a generalized coordinate system The function. Since only the bottom right corner block in the Φ matrix corresponds to... The identity matrix, therefore In effect, the nonlinear force is mapped to the corresponding location in the contracted space. At this point, the nonlinear degrees of freedom are only the interface degrees of freedom. The number of degrees of freedom is far less than the total number of degrees of freedom of the original system, resulting in a significant improvement in solution efficiency.

[0053] Combining all the above, a high-fidelity full three-dimensional finite element simulation model of the turbine's central tie rod rotor, considering the rough contact interface, can finally be obtained.

[0054] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for modeling and analyzing a turbine machine's central tie rod rotor based on a rough contact interface, characterized in that, include: A pre-trained contact stiffness prediction model is invoked, and the roughness parameters and load parameters of the current working condition are input to obtain the predicted stiffness of the contact interface. The pre-trained contact stiffness prediction model is a micromechanical model of a rough contact interface based on elastoplastic contact theory and statistical theory. It is obtained by generating a sample dataset by changing the roughness parameters, material properties and normal load, and using the dataset to train a machine learning model. A three-dimensional finite element model of the rotor is established, and the contact interface region is meshed. Nonlinear spring damping elements are set between corresponding nodes of the contact interface. The obtained predicted stiffness is assigned to the nonlinear spring damping elements. The degree of freedom of the finite element model after stiffness assignment is reduced by a hybrid modal order reduction method. The vibration response of the reduced model is then solved.

2. The method for modeling and analyzing the turbine machine center tie rod rotor based on a rough contact interface as described in claim 1, characterized in that, When training the machine learning model, the dataset is divided into a training set, a validation set, and a test set. The model is trained using the training set and the mean squared error is used as the loss function. Hyperparameters are tuned using the validation set to suppress overfitting.

3. The method for modeling and analyzing the rotor of a turbine's central tie rod based on a rough contact interface as described in claim 1, characterized in that, The full three-dimensional finite element model uses hexahedral elements for meshing, and the mesh is locally refined in the contact interface area, while ensuring that the mesh nodes on both sides of the contact interface correspond one-to-one.

4. The method for modeling and analyzing the rotor of a turbine's central tie rod based on a rough contact interface as described in claim 1, characterized in that, The force-displacement relationship of the nonlinear spring damping unit is defined by the stiffness curve output by the contact stiffness prediction model, and the tensile strength of the nonlinear spring damping unit is set to zero or close to zero to simulate the instantaneous separation and re-contact behavior of the contact interface during vibration.

5. The method for modeling and analyzing the rotor of a turbine's central tie rod based on a rough contact interface as described in claim 1, characterized in that, The hybrid modal reduction method includes: using a fixed interface modal synthesis method to reduce the degrees of freedom of the linear structural parts in the rotor model while retaining their low-order principal modes, wherein the linear structural parts include the wheel disk, shaft segments, and bearing support structures; using a nonlinear modal substructure method to retain all physical degrees of freedom of the contact interface nodes; and coupling the reduced linear modal coordinates with the nonlinear interface degrees of freedom to assemble them into the reduced-order system dynamic equations, wherein the nonlinear interface degrees of freedom correspond to all node pairs with spring-damped elements.

6. The analysis system of the modeling and analysis method for turbine central tie rod rotor based on rough contact interface as described in any one of claims 1-5, characterized in that, include: The stiffness prediction module is used to call a pre-trained contact stiffness prediction model and output the predicted stiffness based on the input roughness parameters and load parameters. The modeling and element setting module is used to establish a three-dimensional finite element model of the rotor and set nonlinear spring damping elements between corresponding nodes of the contact interface. A stiffness assignment module is used to assign the predicted stiffness output by the stiffness prediction module to the nonlinear spring damping unit. The solver module is used to perform dynamic solutions on the stiffened finite element model using a hybrid modal order reduction method to obtain the vibration response.

7. The analysis system as described in claim 6, characterized in that, The solution module includes a linear substructure order reduction unit, a nonlinear interface processing unit, and a solution unit. The linear substructure order reduction unit is used to reduce the degrees of freedom of the disk, shaft segment, and bearing support structure in the rotor model using the fixed interface modal synthesis method, while retaining their low-order principal modes. The nonlinear interface processing unit is used to retain all physical degrees of freedom of the contact interface nodes and extract their nonlinear dynamic characteristics. The solving unit is also used to couple the reduced linear modal coordinates with the nonlinear interface degrees of freedom, assemble and solve the reduced-order system dynamic equations.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 5.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 5.

Images