A quick wind power sliding bearing oil film stability evaluation method and system

Abstract

This invention provides a rapid method and system for assessing the stability of oil films in wind turbine sliding bearings. Applied to the field of oil film stability diagnosis technology, this invention first searches for the intersection points of the limit state function surface and the uncertainty domain, i.e., control points, to accurately capture the key positional information of the limit state surface that determines the oil film instability boundary. Then, based on the control points, samples are arranged in the geometric uncertainty domain, and a response surface surrogate model of the true limit state function is constructed. This effectively improves the accuracy of constructing the quadratic polynomial response surface of the maximum amplitude limit state of the shaft center trajectory, reduces the number of times the dynamic mesh calculation model needs to be called to construct a high-precision response surface surrogate model across the entire domain, and improves computational efficiency. Finally, the response surface surrogate model is used to quickly calculate and evaluate the credibility and similarity of the oil film stability assessment, effectively promoting the application of evidence theory in the assessment of oil film stability in sliding bearings under complex operating conditions such as wind power, and alleviating the efficiency and accuracy problems in the assessment of oil film stability in wind turbine sliding bearings.

Description

Technical Field

[0001] This invention relates to the field of oil film stability diagnostic technology, and in particular to a rapid method and system for evaluating the oil film stability of wind turbine sliding bearings. Background Technology

[0002] Wind turbines are rapidly developing towards higher power and larger sizes, and sliding bearings have become the core support design solution with the greatest engineering application potential for high-power wind turbine gearbox transmission shaft systems.

[0003] However, due to the random fluctuations in natural wind speed, the wind turbine drive shaft system often suffers from frequent alternating sudden heavy loads, causing the sliding bearings to be in harsh working conditions of low speed, heavy load and drastic changes in wind load for a long time. As a result, the oil film thickness and pressure field exhibit strong transient and high gradient nonlinear time-varying characteristics.

[0004] The maximum amplitude of the shaft center trajectory is a core indicator for quantitatively assessing the stability of the oil film. In engineering, the maximum amplitude of the shaft center trajectory exceeding 30% of the bearing radius clearance is usually taken as the critical criterion for oil film instability.

[0005] When the maximum amplitude increases by more than 25%-30% relative to the stable baseline, it is considered a significant early warning signal of oil film instability; the larger the amplitude, the smaller the corresponding minimum oil film thickness. When the maximum amplitude approaches or exceeds the design allowable value, the system damping will be severely weakened, easily inducing low-frequency whirling of the journal and accelerating the dynamic instability of the oil film. Once the minimum oil film thickness falls below the combined roughness of the bearing bush and journal, it will cause direct contact between the metal surfaces and localized instantaneous high temperatures, completely destroying the oil film stability and even leading to serious safety accidents such as unplanned shutdowns of wind turbine units. Therefore, accurate and efficient assessment and full-lifecycle trend monitoring of the maximum amplitude of the shaft trajectory are key technical means to achieve early warning of lubrication failures in sliding bearings and ensure the long-term safe and stable operation of wind power equipment.

[0006] Currently, the industry mainly uses the finite volume method based on dynamic mesh technology to solve the transient pressure field of the oil film and calculate the shaft center trajectory. However, this method requires continuous updates to the computational domain and mesh model during the solution process, resulting in high computational time and cost per simulation. At the same time, due to the strong randomness of the wind field environment, key parameters such as the external load of the bearing, the operating speed, and the dynamic viscosity of the lubricating oil have significant cognitive uncertainties.

[0007] Although evidence theory can effectively measure such cognitive uncertainty, it suffers from a severe combinatorial explosion problem when dealing with joint focal element extremum analysis of multidimensional variables. It requires repeated calls to the time-consuming dynamic mesh model for each joint focal element, and the computational cost increases exponentially with the number of focal elements. The overall computational efficiency decreases by an order of magnitude, making it completely unsuitable for the rapid assessment needs of wind power sites and difficult to implement in actual engineering.

[0008] To alleviate the aforementioned computing power bottleneck, existing technologies often introduce proxy models such as global response surfaces to fit the mapping relationship between parameters, thereby improving computational efficiency through dimensionality reduction mapping. However, under the condition of strong sudden changes and heavy loads in wind power, the limit state surface that determines the instability boundary of the oil film exhibits strong nonlinear characteristics. Uniform sampling not only cannot completely get rid of the high computational cost, but also easily produces fitting distortion at the key instability boundary, directly leading to a significant reduction in the reliability and accuracy of the oil film stability assessment results, and even misjudgment or omission of instability risk.

[0009] Therefore, it is necessary to develop a rapid evaluation method and system for the oil film stability of wind turbine sliding bearings that is suitable for multi-source cognitive uncertainty conditions and takes into account both extremely high computational efficiency and evaluation accuracy. Summary of the Invention

[0010] This invention provides a rapid method and system for evaluating the stability of oil film in wind turbine sliding bearings. It solves the technical problems in existing wind turbine sliding bearing oil film stability evaluation technologies, such as the high number of calls to the dynamic mesh transient calculation model, the extremely high computing cost, the combinatorial explosion problem when the evidence theory deals with cognitive uncertainty, and the inability of existing global response surface proxy models to balance boundary fitting accuracy and computational efficiency.

[0011] According to a first aspect of the present invention, a rapid method for evaluating the oil film stability of wind turbine sliding bearings is provided. The method includes: S1. Within the framework of evidence theory, the core factors influencing the axis trajectory are taken as cognitive uncertainty variables, and the intervals of each cognitive uncertainty variable are used to construct the geometric uncertainty domain of an n-dimensional polyhedron; n is a natural number greater than 0. S2. Traverse all edges of the geometric uncertainty domain of the n-dimensional polyhedron to obtain control points, where the control points are the intersections of each edge with the limit state surface of the maximum amplitude of the oil film axis trajectory. S3. Calculate the average value of all control points as the initial sample center point, expand the sample points based on the initial sample center point, use sample genetic management technology to determine whether to substitute the expanded sample points into the wind power sliding bearing dynamic mesh calculation model to calculate the maximum amplitude of the shaft center trajectory, and construct a response surface proxy model based on the maximum amplitude of the shaft center trajectory and core factors. The response surface proxy model is constructed based on a preset oil film safety threshold. S4. Solve for the maximum possible failure point based on the response surface surrogate model, and update the maximum possible failure point as the new initial sample center point. Repeat S3 until the initial sample center points on both sides meet the convergence condition, and obtain the converged response surface surrogate model. S5. Replace the dynamic grid calculation model with a convergent response surface surrogate model, and perform extreme value analysis on the joint focal element formed by the interval intersection of each cognitive uncertainty variable to obtain the maximum and minimum values ​​corresponding to each joint focal element. S6. Based on the maximum and minimum values ​​of the combined coke element, calculate the reliability function value and the similarity function value of the oil film stability of the wind turbine sliding bearing. S7. Change the preset oil film safety threshold and repeat S2 to S6 to plot the cumulative confidence and cumulative fidelity curves for oil film stability assessment.

[0012] Furthermore, in S2, traversing all edges of the geometric uncertainty domain of the n-dimensional polyhedron to obtain control points includes the following steps: On the edges of the geometric uncertainty domain of the n-dimensional polyhedron, only the cognitive uncertainty variable corresponding to that edge is allowed to change continuously within the interval, while other cognitive uncertainty variables are fixed to the endpoint values ​​of their respective edge intervals; The center point of the interval corresponding to the continuously changing cognitive uncertainty variable is used as the initial iteration value for iterative calculation until the preset convergence criterion is met, and the control point is obtained. Among them, it is determined whether the iterative process has converged or whether the obtained cognitive uncertainty variable value is within the corresponding interval. If not, the intersection of the corresponding edge and the maximum amplitude limit state surface of the oil film axis trajectory is removed.

[0013] Furthermore, in S3, the collection of samples, the screening of samples using genetic management technology, and the input of these samples into the dynamic mesh calculation model of the wind turbine sliding bearing to calculate the maximum amplitude of the shaft center trajectory specifically include the following steps: Determine if the collected sample exists in the historical sample database. If it does, directly obtain the maximum amplitude of the axis trajectory corresponding to the sample point in the historical sample database. Otherwise, the sample is input into the dynamic mesh computing model for calculation, and the calculation result is stored in the historical sample database.

[0014] Furthermore, in S3, a response surface proxy model is constructed based on the maximum amplitude of the axis trajectory and the core factors, as shown in the following formula: ; in, The dimension of variables representing uncertainties in the evidence. , and All represent response surface coefficients. This represents the i-th cognitive uncertainty parameter. Represents the limit state function of the maximum amplitude of the axis trajectory The response surface, This indicates the preset oil film safety threshold. This represents the maximum amplitude of the axis trajectory obtained by calling the dynamic mesh calculation model.

[0015] Furthermore, in S4, the most likely failure point is updated to a new initial sample center point, and S3 is repeated until the initial sample center points on both sides satisfy the convergence condition, resulting in a converged response surface surrogate model, including: S41. Generate axially expanded sample points again using the new initial sample center point, and supplement them into the initial sample space after screening by sample genetic management technology, and update the response surface surrogate model coefficients simultaneously. S42. Repeat S41 until the initial sample center points of two adjacent iterations satisfy the iteration convergence criterion, and obtain the converged response surface surrogate model.

[0016] Furthermore, in S6, based on the maximum and minimum values ​​of the combined coke element, the reliability function value and the similarity function value of the oil film stability of the wind turbine sliding bearing are calculated, including the following steps: The joint focal element is input into the basic confidence assignment function to obtain the basic confidence assignment value corresponding to each joint focal element; Determine whether to include the basic confidence assignment value corresponding to the joint focal element in the confidence function value and the similarity function value based on the relationship between the joint focal element and the constructed stability region. Specifically, it is determined whether the maximum and minimum values ​​of the joint focal element are both within the stable region. If they are both within the stable region, the basic confidence assignment value corresponding to the joint focal element is included in the confidence function value and the similarity function value. Otherwise, determine whether the maximum value of the joint focal element is within the stable region and the minimum value is not within the stable region. If the maximum value of the joint focal element is within the stable region and the minimum value is not within the stable region, then include the basic credibility allocation value corresponding to the joint focal element in the similarity function value. Otherwise, determine whether the maximum and minimum values ​​of the joint focal element are both outside the stable region. If so, the basic confidence assignment value corresponding to the joint focal element is not included in the confidence function value and the similarity function value.

[0017] According to a second aspect of the present invention, the present invention provides a rapid wind turbine sliding bearing oil film stability assessment system, comprising: The uncertainty domain construction module is used to construct the geometric uncertainty domain of an n-dimensional polyhedron by taking the core factors affecting the axis trajectory as cognitive uncertainty variables within the identification framework of evidence theory, and using the intervals of each cognitive uncertainty variable together. The control point search module, connected to the uncertainty domain construction module, is used to traverse all edges of the geometric uncertainty domain of the n-dimensional polyhedron to obtain control points. The control points are the intersections of each edge with the limit state surface of the maximum amplitude of the oil film axis trajectory; n is a natural number greater than 0. The proxy model construction module is connected to the control point search module. It is used to calculate the average value of all control points as the initial sample center point, expand the sample points based on the initial sample center point, and use sample genetic management technology to determine whether to substitute the expanded sample points into the wind turbine sliding bearing dynamic mesh calculation model to calculate the maximum amplitude of the shaft center trajectory. Based on the maximum amplitude of the shaft center trajectory and the core factors, a response surface proxy model is constructed. The response surface proxy model is constructed based on a preset oil film safety threshold. The sequence iterative optimization module is connected to the surrogate model construction module. It is used to solve the maximum possible failure point based on the response surface surrogate model, update the maximum possible failure point as a new initial sample center point, re-enter the surrogate model construction module, and continue until the initial sample center points on both sides meet the convergence condition to obtain a converged response surface surrogate model. The extreme value analysis module, connected to the sequence iterative optimization module, is used to replace the dynamic grid calculation model with a converged response surface surrogate model to perform extreme value analysis on the joint focal elements formed by the interval intersection of various cognitive uncertainty variables, and obtain the maximum and minimum values ​​corresponding to each joint focal element. The stability assessment module, connected to the control point search module and the extreme value analysis module, is used to calculate the confidence function value and the similarity function value of the oil film stability of the wind turbine sliding bearing based on the maximum and minimum values ​​of the joint coke element. It changes the preset oil film safety threshold, repeatedly enters the control point search module, and plots the cumulative confidence and cumulative similarity curves to assess the oil film stability.

[0018] According to a third aspect of the present invention, an electronic device is provided, including a memory, a processor, and a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor to implement the method as described in the first aspect.

[0019] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, characterized in that a computer program is stored thereon; the computer program is executed by a processor to implement the method as described in the first aspect.

[0020] The beneficial effects of this invention are: To address the problem of focal explosion caused by partitioning in the measurement of cognitive uncertainty in evidence theory, i.e., the more partitions there are, the more accurate the results, but the computational cost increases exponentially, this invention searches for the intersection points of the limit state function surface and the uncertainty domain, i.e., control points, and constructs a surrogate model based on the control points to accurately capture the key position information of the limit state surface that determines the instability boundary of the oil film. Specifically, by arranging samples in the geometrically uncertain domain based on control points and constructing a response surface surrogate model of the true limit state function, the accuracy of constructing the quadratic polynomial response surface of the maximum amplitude limit state of the axisymmetric trajectory is effectively improved. This reduces the number of times the dynamic mesh calculation model needs to be called to construct a high-precision response surface surrogate model across the entire domain, thereby improving computational efficiency and solving the problems of insufficient fitting accuracy and excessive sample call volume of existing global response surface models. Moreover, the surrogate model constructed based on control points is independent of the number of foci elements in the evidence theory, thus solving the computational efficiency problem caused by foci element explosion in the oil film stability assessment of the evidence theory.

[0021] Meanwhile, this invention introduces sample genetic management technology to compare and screen new and old samples during the iterative sampling process, avoiding repeated calls to the dynamic mesh computing model to calculate the axis trajectory, and further significantly reducing the computational load; it also introduces a sampling center sequence iteration mechanism to make the sample collection area approach the maximum possible unstable region of the oil film, thereby improving the accuracy of the surrogate model constructed with a small number of samples.

[0022] Finally, the reliability and similarity of the evaluation of oil film stability are quickly calculated using the response surface surrogate model, which effectively promotes the application of evidence theory in the evaluation of oil film stability of sliding bearings under complex working conditions such as wind power, and alleviates the efficiency and accuracy problems in the evaluation of oil film stability of wind power sliding bearings.

[0023] This invention significantly improves the evaluation calculation efficiency while ensuring that the accuracy of oil film stability assessment meets engineering requirements. It effectively broadens the engineering application of evidence theory in the evaluation of oil film stability of sliding bearings under complex working conditions of strong sudden changes and heavy loads, enhances the practicality of intelligent monitoring technology for sliding bearing stability, and provides reliable technical support for condition monitoring, fault early warning and operation and maintenance optimization of wind power equipment.

[0024] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of the present invention, nor is it intended to restrict the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0025] The above and other features, advantages, and aspects of the various embodiments of the present invention will become more apparent from the accompanying drawings and the following detailed description. The drawings are provided for a better understanding of the invention and are not intended to limit the invention. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein: Figure 1 A block diagram of a rapid method for evaluating the oil film stability of wind turbine sliding bearings provided by an embodiment of the present invention is shown. Figure 2 This diagram illustrates the computational domain grid partitioning provided in an embodiment of the present invention. Figure 3A schematic diagram of grid layering provided in an embodiment of the present invention is shown; Figure 4 This diagram illustrates the positional relationship between the combined focal element and the maximum amplitude limit state function of the axis trajectory provided in an embodiment of the present invention. Figure 5 This diagram illustrates the spatial geometric relationship between the uncertainty domain, the limit state surface, and the control point provided in an embodiment of the present invention. Figure 6 This invention provides a sample layout distribution diagram based on control points. Figure 7 A schematic diagram of the experimental platform layout provided in an embodiment of the present invention is shown; Figure 8 This diagram illustrates the axisymmetric trajectory under fine grid density provided in an embodiment of the present invention. Figure 9 This diagram illustrates the axisymmetric trajectory under medium grid density according to an embodiment of the present invention. Figure 10 This diagram illustrates the axis trajectory under coarse grid density according to an embodiment of the present invention. Figure 11 The oil film stability evaluation diagrams for different BPA structures provided in the embodiments of the present invention are shown. Figure 12 This diagram illustrates the spatial distribution of control points and sequence iteration samples provided in an embodiment of the present invention. Figure 13 A comparison chart of the computational efficiency of different stability evaluation methods provided in embodiments of the present invention is shown; Figure 14 A flowchart of an oil film stability assessment method provided according to an embodiment of the present invention is shown; Figure 15 A block diagram of a rapid wind turbine sliding bearing oil film stability assessment system provided according to an embodiment of the present invention is shown. Figure 16 A block diagram of an electronic device according to an embodiment of the present invention is shown. Detailed Implementation

[0026] 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 embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0028] First, the following explanation is provided: The axis trajectory is a direct visualization of the dynamic behavior of the oil film and is the primary tool for judging the stability of the oil film.

[0029] The maximum amplitude of the shaft center trajectory usually refers to the maximum vibration displacement of the shaft center trajectory in the horizontal or vertical direction, or the maximum radial distance from the outer edge of the trajectory to the center of the bearing.

[0030] The maximum amplitude of the shaft trajectory is a key indicator for evaluating oil film stability. The essence of oil film stability lies in maintaining a sufficiently thick and stable oil film that completely separates the journal from the bearing.

[0031] The larger the maximum amplitude of the shaft trajectory, the smaller the minimum oil film thickness, and the lower the lubrication reliability. When the maximum amplitude of the shaft trajectory approaches or exceeds the allowable value, it directly indicates that the oil film is too soft and cannot provide sufficient force to constrain the journal's movement. Excessive journal wobble within the bearing, coupled with excessive play and a thinner oil film, reduces the system's damping, making the journal more prone to low-frequency whirling—the beginning of dynamic instability. If the minimum oil film thickness is less than the combined roughness of the bearing bush and journal surfaces, metal-to-metal contact will occur, leading to friction, wear, and instantaneous high temperatures. Stability is completely destroyed, and the system becomes extremely sensitive to minor disturbances (such as oil temperature fluctuations and impurities), making it more susceptible to unstable vibrations.

[0032] When the maximum amplitude of the shaft center trajectory increases by more than 25%-30% relative to the baseline, it should be considered a significant warning. Therefore, in condition monitoring, automated and trend-based monitoring of the maximum amplitude of the shaft center trajectory is a key technical means to predict sliding bearing lubrication failures and ensure the safe and stable operation of equipment.

[0033] Due to the random fluctuations in wind speed, wind turbines often bear huge and frequently changing loads. The cognitive uncertainties surrounding factors such as the external load on the sliding bearings, their rotational speed, and the viscosity of the lubricating oil make the shaft center trajectory an unpredictable and indeterminate process. Measuring cognitive uncertainty using evidence theory results in a computational cost that increases dramatically with the dimensionality of the uncertainty parameters and the number of focal elements. Furthermore, the mapping relationship between these cognitive uncertainty parameters and the maximum amplitude of the shaft center trajectory is highly nonlinear and unclear, leading to extensive use of time-consuming dynamic mesh computation models and a significant decrease in computational efficiency.

[0034] The physical model of this invention is described below, which transforms the complex problem of solving the dynamic oil film pressure into two steady subproblems: The sliding bearings in wind turbine gearboxes are often subjected to harsh operating conditions such as low speed, heavy load, and drastic changes in wind load. Under these unsteady conditions, the movement of the journal center relative to the bearing center is dynamic, resulting in significant transient characteristics in the oil film thickness and pressure field.

[0035] To describe this physical process, based on classical lubrication theory, the dynamic Reynolds equation is used as the governing equation for the flow of lubricating oil in the bearing radial clearance, as shown in the following formula: ; Where x and y represent the radial coordinates of the oil film, This indicates taking the partial derivative of the parameter in parentheses with respect to the parameter at position 'a', where 't' represents time. Indicates oil film pressure (Pa); This indicates the dynamic viscosity of the lubricating oil (Pa·s). Indicates the oil film thickness (m). Indicates the bearing radius (m); This indicates the linear velocity (m / s) on the journal surface. , Indicates the journal radius (m); This indicates the linear velocity (m / s) on the bearing surface. , and These represent the angular velocities (rad / s) of the journal and the bearing, respectively.

[0036] To improve the versatility and stability of numerical calculations, this invention introduces the following dimensionless variables: ; ; ; ; ; ; in, The bearing width is in meters (m). Indicates the bearing radius clearance (m). Indicates the eccentricity. H represents the dimensionless circumferential coordinate, and H represents the dimensionless oil film thickness. Represents dimensionless axial coordinates. Indicates the bearing's geometric characteristic parameters. This represents the dimensionless oil film pressure. This represents angular velocity. Oil film thickness can be calculated using the following formula: .

[0037] Combining the aforementioned formulas, we obtain the general form of the dimensionless dynamic Reynolds equation: ; in, The offset angle is represented by the three terms on the right-hand side of the equation, which represent the rotational effect, the rotational effect caused by the change in the offset angle, and the radial compression effect, respectively.

[0038] To clearly separate and characterize the two physical effects of rotation and compression, and to facilitate the subsequent solution of the axial motion, the Hahn method is used to revise the dimensionless dynamic Reynolds equations.

[0039] Specifically, merger Introducing effective angular velocity To comprehensively characterize the rotation effect: ; The merged result can be represented as Define dynamic parameters To characterize the intensity of the squeezing effect:

[0040] Therefore, the dimensionless dynamic Reynolds equation can be adapted to the standard form of the Hahn method:

[0041] The right side contains only two terms, the first of which represents the effective angular velocity. The driving rotational effect, the second term represents the radial velocity of the axis. The driving squeezing effect, and through the dynamic parameters Quantify it.

[0042] Since the standard form of the Hahn method is a linear partial differential equation, according to the principle of linear superposition, the overall dimensionless pressure distribution... It can be decomposed into a pure rotating pressure field With pure extrusion pressure field A linear combination of these, see the following formula: in, and These are particular solutions that satisfy the following Poisson equation:

[0043] Thus, the dynamic control equation system constructed based on the Hahn method has effectively transformed the complex dynamic load oil film pressure solution process into two steady subproblems through the above two equations. This transformation not only gives each subproblem a clear physical meaning, but also achieves complete decoupling of the solution process from the instantaneous journal velocity at the analytical level.

[0044] Based on the above, this invention provides a rapid method for evaluating the stability of oil film in wind turbine sliding bearings. (See [link to relevant documentation]). Figures 1 to 14 This includes the following steps: S1. Within the framework of evidence theory, the core factors influencing the axis trajectory are taken as cognitive uncertainty variables, and the intervals of each cognitive uncertainty variable are used to construct the geometric uncertainty domain of an n-dimensional polyhedron; n is a natural number greater than 0. In the actual operating conditions of wind turbine sliding bearings, due to the random influence of natural wind conditions and equipment status, the fluctuations of cognitively uncertain variables such as sudden heavy loads and initial eccentricity usually manifest as a small range of uncertainty around their baseline values.

[0045] The cognitive uncertainty variables constructed by evidence theory can be described using the following mathematical interval form: ; in, Let represent the i-th cognitive uncertainty variable, namely parameters such as mutation overload and initial eccentricity measured by evidence. Let I represent the uncertainty interval of the i-th cognitive uncertainty variable, and let I denote the interval. and These represent the lower and upper bounds of the uncertainty interval for cognitive uncertainty variables, respectively.

[0046] The geometric uncertainty domain of the n-dimensional polyhedron formed by these n uncertain variables See the following formula: ; Should In a polyhedron, every two orthogonal planes intersect to form an edge. The total geometric uncertainty domain includes... Edge.

[0047] S2. Traverse all edges of the geometric uncertainty domain of the n-dimensional polyhedron to obtain control points, where the control points are the intersections of each edge with the limit state surface of the maximum amplitude of the oil film axis trajectory. The key to stability assessment is finding the critical limit state surface, i.e. The value of the limit state function corresponding to the maximum amplitude of the shaft center trajectory is used to evaluate the stability of the oil film in the wind turbine sliding bearing based on the critical instability point, i.e. the point of maximum possible failure.

[0048] The critical limit state surface where the maximum possible failure point is located In the geometric uncertainty domain passing through the aforementioned n-dimensional polyhedron When intersecting with the edges of an n-dimensional polyhedron, intersections are inevitable. This invention defines these intersections as control points, denoted as... .

[0049] like Figure 5 As shown, these control points accurately define the critical boundary of the limit state surface falling within the geometric uncertainty domain, precisely capturing the spatial location information of the limit state surface in this region.

[0050] Extracting control points is crucial for subsequently constructing a quadratic polynomial response surface proxy model with high accuracy within local key regions and for significantly reducing the number of calls to the time-consuming physical model.

[0051] Specifically, this method traverses each edge of an n-dimensional polyhedron to obtain control points. On any given edge, there is only one cognitive uncertainty variable. In its interval , The internal changes are continuous, while the rest... Each cognitive uncertainty variable is fixed to the endpoint value of its respective interval (i.e., the upper or lower bound is taken).

[0052] At this point, the intersection problem in multidimensional space is reduced in dimension and transformed into a numerical root-finding problem of a univariate function, as shown in the following formula: ; ; in, Indicates except The remaining variables are fixed at the boundaries.

[0053] Because this nonlinear implicit function is difficult to solve, this invention uses numerical root-finding algorithms (such as Newton's iteration method or bisection method) for fast solution.

[0054] Let's take Newton's iteration method as an example: Select the center point of the edge interval corresponding to the current cognitive uncertainty variable. / 2 is used as the initial iteration value. Assume the intersection point is obtained after k iterations. Then the next intersection point Update using the following formula: ; in, The limit state function representing the maximum amplitude of the trajectory at the center point is... The partial derivatives can be approximated by the central difference method.

[0055] Repeat the above iterative process until the following preset convergence criteria are met simultaneously: Relative displacement error And the limit state function value ;in, This is the minimum tolerance set.

[0056] In real geometric space, not all edges intersect the limit state surface. Therefore, the following filtering mechanism is introduced in the algorithm implementation: If the iterative process diverges or the final result is... It landed , If the edge does not have a valid intersection point, it is determined that there is no valid intersection point on that edge and it is discarded.

[0057] For all polyhedra Repeat the root-finding and screening process described above for each edge, and finally extract m valid control points. This serves as the benchmark anchor point for subsequently constructing the response surface proxy model.

[0058] S3. Calculate the average value of all control points as the initial sample center point, expand the sample points based on the initial sample center point, use sample genetic management technology to determine whether to substitute the expanded sample points into the wind power sliding bearing dynamic mesh calculation model to calculate the maximum amplitude of the shaft center trajectory, and construct a response surface proxy model based on the maximum amplitude of the shaft center trajectory and core factors. The response surface proxy model is constructed based on a preset oil film safety threshold. The control point is the intersection of the edge of the geometric uncertainty domain and the limit state surface where the trajectory of the oil film axis reaches its maximum amplitude. Then, based on the spatial distribution characteristics of the control points, the sample space was constructed using the axial experimental design method.

[0059] Extract the mean of all control points as the initial sample center point. .

[0060] like Figure 6 As shown, with the initial sample center point Using this as a baseline, we extend the coordinate axes along the positive and negative directions of each cognitive uncertainty variable to construct an expanded sample point. , The radius of the interval representing the cognitive uncertainty variable. This represents the given axially expanded sample coefficients.

[0061] The initial sample center points and the expanded sample points together constitute the sample set used to construct the response surface proxy model.

[0062] After determining the coordinates of the sample set, the limit state function value of the maximum amplitude of its corresponding axisymmetric trajectory is obtained. To avoid the high computational cost caused by multiple subsequent calls to the global dynamic mesh calculation model, this invention introduces sample genetic management technology for data screening.

[0063] Specifically, the sample genetic management technology screening process includes the following steps: Before calling the dynamic mesh calculation model of the wind turbine sliding bearing, the historical sample database is searched. If the current extended sample point already exists in the historical sample database, its solved maximum amplitude of the shaft center trajectory is directly inherited to calculate the limit state function value of the maximum amplitude of the trajectory. If it is a brand new sample point, it is input into the dynamic mesh calculation model of the wind turbine sliding bearing for time-domain simulation calculation, and the results and coordinates are synchronously stored in the historical sample database. In the subsequent iterative point placement process, this step can intelligently identify and inherit the previously solved valid historical samples, strictly eliminate redundant calculations, and significantly reduce the computational load of the maximum amplitude of the shaft center trajectory.

[0064] This invention employs the finite difference method to numerically solve the aforementioned two steady subproblems, thereby transforming the continuous physical solution domain into a discrete computational grid.

[0065] The oil film on the cylindrical surface of the sliding bearing is spread out along the circumference and mapped to a regular rectangular calculation plane.

[0066] In a dimensionless coordinate system Next, orthogonally mesh the rectangular computational plane: along the circumferential direction. Discretized into m nodes, axial ( Discretize into k nodes to obtain Differential mesh.

[0067] Considering that wind turbine sliding bearings typically operate under heavy loads with high eccentricity, resulting in extremely thin minimum oil film thickness and a very narrow and steep pressure peak distribution, this high-gradient pressure field places stringent requirements on mesh density. If the mesh is too sparse, the discrete nodes cannot effectively capture the pressure peaks, leading to severe truncation errors and causing the calculated maximum oil film pressure and bearing capacity to be significantly lower than the actual values. Conversely, if an overly dense mesh is used, the computational cost will increase exponentially when performing long-term shaft centerline trajectory tracking and subsequent stability assessments.

[0068] To strike a balance between computational accuracy and efficiency, this invention constructs, as follows: Figure 2 The uniform structured differential grid shown is used, and the governing equations are discretized using the following five-point central difference scheme: ; in, express Dimensionless oil film pressure on the differential grid Indicates interval, express Dimensionless oil film thickness on the differential grid .

[0069] Based on the above steps, a dynamic mesh computational model was obtained. To intuitively verify the geometric adaptability of the mesh generation in the actual bearing radius clearance, the orthogonal mesh in the computational domain was mapped back to the Cartesian coordinate system in physical space, as follows: Figure 3 As shown, the above is illustrated. The projection of the differential mesh onto the bearing cross-section (two radial directions). The mesh lines in the circumferential direction correspond to the mesh lines in the computational domain. The discrete coordinates and radial layered distribution (identified by different colored blocks, with orange indicating areas with thinner oil films, blue indicating areas with thicker oil films, and gray indicating areas of mixed lubrication wear) visually reflect the coverage of the grid nodes along the circumference of the oil film.

[0070] Mesh visualization in a Cartesian coordinate system helps to confirm the distribution density of discrete nodes in the minimum oil film thickness region, ensuring that the numerical model can accurately capture the oil film pressure establishment process in the actual physical space.

[0071] Based on the aforementioned high-precision transient dynamic mesh calculation model, in order to overcome the huge computational cost of extreme value analysis in the geometric uncertainty domain, this invention introduces response surface methodology into the stability assessment of oil film in wind turbine sliding bearings. An explicit expression is used to construct the mapping relationship between the maximum amplitude of the shaft center trajectory and parameters such as sudden overload and initial eccentricity. This replaces the true limit state function of the maximum amplitude of the shaft center trajectory for extreme value analysis, which can significantly improve computational efficiency.

[0072] Polynomial response surface methodology is one of the commonly used response surface techniques. It is simple in form, easy to construct, and has a strong ability to smooth numerical noise.

[0073] This invention uses a quadratic response surface without cross terms to approximate the limit state function of the maximum amplitude of the axis-centered trajectory, i.e., a response surface surrogate model, as shown in the following formula: ; in, The dimension of variables representing uncertainties in the evidence. , and All represent response surface coefficients. Represents the limit state function of the maximum amplitude of the axis trajectory The response surface, Indicates the oil film safety threshold. This represents the maximum amplitude of the axis trajectory obtained by calling the dynamic mesh calculation model.

[0074] When the maximum amplitude of the shaft center trajectory exceeds 30% of the bearing radius clearance, the shaft center trajectory is highly likely to deviate from the normal distribution. The journal displacement may have entered the nonlinear softening zone of the oil film, and the minimum oil film thickness is critically low. The oil film transforms from stable unidirectional extrusion to unstable circumferential vortexing. To avoid unstable vortexing of the oil film, it is generally required that... ,Right now The oil film was determined to be unstable.

[0075] In summary, by utilizing a set of effective sample points selected through sample genetic management technology, and based on the maximum amplitude limit state function of the trajectory, the coefficients of the undetermined quadratic polynomial response surface are solved by fitting using the least squares method. This allows for the construction of a quadratic response surface surrogate model that locally approximates the true limit state function without cross terms. .

[0076] S4. Solve for the maximum possible failure point based on the response surface surrogate model, and update the maximum possible failure point as the new initial sample center point. Repeat S3 until the initial sample center points on both sides meet the convergence condition, and obtain the converged response surface surrogate model.

[0077] Under the condition of strong sudden change and heavy load in wind power, the limit state surface that determines the instability boundary of the oil film exhibits strong nonlinear characteristics, and it is difficult to achieve ideal accuracy at the instability boundary with a single global static fitting.

[0078] This invention proposes a sequence iteration mechanism, which specifically includes the following steps: The maximum probable failure point is determined based on the current response surface surrogate model. This point is the one with the maximum joint probability density value on the limit state surface g(X)=0 in traditional probabilistic reliability analysis theory. This maximum probable failure point is then used as the new initial sample center point. .

[0079] Axially expanded sample points are generated again from the initial sample center points. After sample genetic management and screening, these points are added to the initial sample space, and the response surface surrogate model coefficients are updated simultaneously. Through this sequential iteration of continuously updating the initial sample center points, the sampling region is driven to dynamically approach the region with the greatest possible instability, until the initial sample center points of two adjacent iterations satisfy the iterative convergence criterion, i.e. .

[0080] The final output is a high-precision, convergent response surface surrogate model that closely matches the actual oil film instability boundary. .

[0081] Clearly, in the stability assessment of the evidence theory, the maximum amplitude limit state function of the trajectory of the axis is... The computational cost and effectiveness of complex and time-consuming models directly affect the efficiency and accuracy of the oil film stability assessment for wind turbine sliding bearings.

[0082] Replacing the time-consuming dynamic mesh calculation model with a response surface surrogate model can improve the computational efficiency of the limit state function of the maximum amplitude of the axis trajectory.

[0083] However, with the increasing complexity of wind turbine sliding bearing operating conditions, it is difficult to guarantee the accuracy of the response surface surrogate model constructed across the entire domain. This results in the maximum amplitude limit state function of the shaft center trajectory often failing to achieve the same accuracy as the actual model. This invention makes the following settings: S5. Replace the dynamic grid calculation model with a convergent response surface surrogate model, and perform extreme value analysis on the joint focal element formed by the interval intersection of each cognitive uncertainty variable to obtain the maximum and minimum values ​​corresponding to each joint focal element.

[0084] After iterative convergence, this convergent response surface surrogate model, possessing explicit mathematical expression characteristics, It can replace time-consuming physical time-domain simulation models.

[0085] For each joint focal element formed by the interval intersections of various uncertain variables within the identification framework of evidence theory. Extreme value analysis is performed using a convergent response surface surrogate model to quickly solve for each joint focal element. Minimum value on 'and maximum value ':

[0086] S6. Based on the maximum and minimum values ​​of the combined coke element, calculate the reliability function value and the similarity function value of the oil film stability of the wind turbine sliding bearing. Due to the lack of cognitive uncertainty, evidence theory uses a flexible basic confidence assignment function (BPA) to describe the degree of trustworthiness of uncertain information such as sudden heavy load and initial eccentricity in wind turbine sliding bearings. The confidence function Bel and the plausibility function PI are the upper and lower probability values, respectively, to jointly describe the reliability of the oil film stability of wind turbine sliding bearings, which can better measure and handle various uncertain information.

[0087] The joint focal element is input into the basic confidence assignment function to obtain the basic confidence assignment value corresponding to each joint focal element; Determine whether to include the basic confidence assignment value corresponding to the joint focal element in the confidence function value and the similarity function value based on the relationship between the joint focal element and the stability region. Specifically, it is determined whether the maximum and minimum values ​​of the joint focal element are both within the stable region. If they are both within the stable region, the basic confidence assignment value corresponding to the joint focal element is included in the confidence function value and the similarity function value. Otherwise, determine whether the maximum value of the joint focal element is within the stable region and the minimum value is not within the stable region. If the maximum value of the joint focal element is within the stable region and the minimum value is not within the stable region, then include the basic credibility allocation value corresponding to the joint focal element in the similarity function value. Otherwise, determine whether the maximum and minimum values ​​of the joint focal element are both outside the stable region. If so, the basic confidence assignment value corresponding to the joint focal element is not included in the confidence function value and the similarity function value.

[0088] Specifically, the joint focal elements are input into the basic confidence assignment function to obtain each joint focal element. Corresponding basic credibility assignment value Subsequently, according to the joint coke element With stability region The relationship between them is used to calculate the confidence function value of the oil film stability of the wind turbine sliding bearing according to the following formula. and similarity function value : ; in, This indicates that the joint focal element must be completely within the constructed stability region, and This indicates that the focal element is completely or partially located within the stable region. The stable region is constructed by the boundary of the cognitive uncertainty parameter and the limiting state function.

[0089] To accurately determine the combined coke element With stability region The relationship between them requires the limit state function of the maximum amplitude of the axis trajectory. In each joint focal element Extreme value analysis was performed on all of them to obtain the maximum value of the limiting state function of the trajectory's maximum amplitude. and minimum value .

[0090] like Figure 4 As shown, the blue box represents the joint focal element. When the maximum and minimum values ​​corresponding to the joint focal element satisfy... and At that time, the combined coke element falls completely within the oil film stability region, that is... The basic confidence assignment value corresponding to this joint focal element Simultaneously, the confidence function value Bel(G) and the similarity function value Pl(G) are taken into account; When the maximum and minimum values ​​corresponding to the combined coke element satisfy and When the combined coke element falls entirely within the oil film instability region, the corresponding basic confidence assignment value is... It is neither included in the confidence function value Bel(G) nor in the similarity function value Pl(G); When the maximum and minimum values ​​corresponding to the combined coke element satisfy and When the combined coke element partially falls within the oil film stability region, the corresponding basic confidence assignment value is... Only the similarity function value Pl(G) is included.

[0091] In summary, if If the focal element is completely located within the stable region, its BPA value (i.e., basic confidence assignment value) is simultaneously included in Bel(G) and Pl(G). like and If the focal element partially crosses the instability boundary, its BPA value is only included in Pl(G); like , If the focal element is completely trapped in the instability domain, its BPA value will not be taken into account.

[0092] S7. Change the preset oil film safety threshold and repeat S2 to S6 to plot the cumulative confidence and cumulative fidelity curves for oil film stability assessment.

[0093] The oil film safety threshold can be set according to the actual project. When the oil film safety threshold... Oil film stability assessment during changes: Change the preset oil film safety threshold Repeat the aforementioned steps to connect the confidence function values ​​under different oil film safety thresholds to obtain the cumulative confidence function curve. The cumulative plausibility function curve is plotted similarly, i.e., connecting all points where the horizontal axis represents the oil film safety threshold and the vertical axis represents the confidence function value. The cumulative confidence function curve is obtained, and all points with the oil film safety threshold on the horizontal axis and the similarity function value on the vertical axis are connected. The cumulative likelihood function curve is obtained, thus enabling a rapid quantitative assessment of oil film stability. A gap exists between the cumulative confidence function curve and the cumulative likelihood function curve; the size of this gap reflects the degree of uncertainty in the problem.

[0094] In this invention, the stability assessment method includes: determining whether the cumulative confidence level and cumulative similarity function values ​​corresponding to the oil film safety threshold are close to the expected indicators required by the engineering. If so, it indicates that the oil film is stable under the oil film safety threshold; otherwise, the oil film is prone to instability. If the distance between the two curves is small, it indicates that the degree of uncertainty is small under the current environment.

[0095] In one specific embodiment, to verify the time consumption of the dynamic mesh computation model in calculating the trajectory model of the sliding bearing shaft center, a [system / mechanism] was built. Figure 7The experimental platform for the sliding bearing-rotor system shown was used for comparative experiments. The experimental platform mainly consists of an ABB motor 1, a frequency converter 2, a flexible coupling 3, a sliding bearing 4, a rotor shaft section 5, and a multi-stage data acquisition system 6, and is equipped with an eddy current displacement sensor 7 to collect the radial displacement signal of the journal in real time.

[0096] The specific structural parameters of the sliding bearing selected in the experiment were as follows: bearing inner diameter of 19.1 mm, bearing width of 14.5 mm, and bearing radius clearance of 0.025 mm. The experimental conditions were set as follows: lubricating oil dynamic viscosity of 0.007 Pa·s, and motor speed constant at 1000 r / min. The load transmitted from the shaft system to the sliding bearing was 15 N.

[0097] Based on the experimental parameters mentioned above, the Reynolds equations were numerically solved using a dynamic mesh computational model and the finite difference method. Since the computational accuracy of the finite difference method is highly dependent on the discreteness of the solution domain, the selection of mesh density becomes the primary step in the numerical simulation. If the mesh is too dense, the small increase in computation time per iteration will be amplified cumulatively in the iterative process, leading to excessively high total computational cost; if the mesh is too sparse, it will cause distortion in pressure gradient capture, resulting in a large truncation error. To find the optimal balance between computational accuracy and operational efficiency, three different mesh density schemes (coarse mesh, medium mesh, and fine mesh) were set up in the simulation and axis trajectory simulation calculations were performed under the same working conditions.

[0098] Table 1 Comparison of calculation results with different grid schemes and experimental data

[0099] To ensure computational accuracy and capture dynamic details, the computational step size is set to... A total of 581 steps were calculated to obtain a complete stable periodic trajectory.

[0100] The axisymmetric trajectories calculated under three different grid density schemes are compared, and the results are as follows: Figures 8 to 10As shown in Table 1, the relative error was quantified based on experimental data. The results show that the coarse mesh, due to its sparse nodes, struggles to accurately capture the high-gradient pressure field, leading to a significant deviation between the axis trajectory shape and the experimental results. The relative errors in the X and Y amplitudes are as high as 12.58% and 7.14%, respectively. In contrast, the trajectories of the medium and fine meshes both match the experimental curves well. The relative errors of the medium mesh are controlled at 1.90% and 1.60%, respectively, meeting the 5% engineering accuracy requirement; while the fine mesh further reduces the error to below 1% (0.35% / 0.59%). Although the fine mesh slightly improves accuracy, it significantly increases the computation time per run. The errors mainly originate from nonlinear factors such as misalignment of the experimental platform and small changes in actual viscosity due to lubricating oil temperature rise. In summary, the medium mesh scheme not only achieves good numerical convergence but also significantly reduces computational costs while maintaining the accuracy of the physical model.

[0101] To fully verify the efficiency and high accuracy of the wind turbine sliding bearing oil film stability evaluation method based on control points and sequential iterative response surfaces proposed in this invention, this invention uses the operating state of a typical wind turbine sliding bearing under harsh conditions as an example for verification.

[0102] Maximum amplitude of axis trajectory It is a core indicator for measuring the dynamic behavior and stability of oil film. It is affected by uncertain factors such as random gusts in the wind field. When the actual maximum amplitude is... Approaching or even exceeding the set oil film safety threshold (critical maximum amplitude) When this happens, the lubrication reliability of the system will drop sharply, which will induce dynamic instability.

[0103] Therefore, in order to comprehensively quantify the stability boundary of the system under different stringency levels and achieve trend-based instability monitoring, an oil film safety threshold is set. Dynamic evolution was performed within the range of 0.01–0.5 μm. Abrupt overload and initial eccentricity were selected as cognitive uncertainty parameters to measure oil film stability, with the abrupt overload range set to [3, 50] KN and the initial eccentricity range to [0.05, 0.2]. Evidence theory was used to measure the aforementioned two-dimensional cognitive uncertainty, and it was discretized into 4, 6, and 8 sub-intervals, respectively. The Basic Confidence Assignment (BPA) structures for abrupt overload and initial eccentricity, constructed for different partition densities, are shown in Table 2.

[0104] Table 2. BPA structures with mutation overload and initial eccentricity parameters of 4, 6, and 8 subintervals, respectively.

[0105] Under this BPA structure, the cross-combination of sub-intervals with different densities will respectively form indivual, Individual and A joint coke element.

[0106] To verify the computational accuracy of the method of the present invention under different interval divisions, for three BPA structures of 4 intervals, 6 intervals, and 8 intervals, taking the 6-interval structure as an example, the search was performed to find its correlation with... The intersection of the edges, i.e. the control points, determines the key positional information of the true axis trajectory maximum amplitude limit state surface.

[0107] Based on these important control points, sample points are collected, and a secondary response surface surrogate model is constructed. Based on the constructed response surface surrogate model, the traditional first-order quadratic moment method is used to solve for the most likely failure point. The response surface surrogate model is iteratively updated until the true failure point under the current threshold is found, thereby obtaining a high-precision limit state function for the maximum amplitude of the axis trajectory. The confidence function value and the similarity function value are then calculated.

[0108] The computational cost of the method in this invention is mainly due to the search process for control points. The number of edges in the uncertainty domain is entirely determined by the dimension of the problem, so the computational cost of this algorithm is not affected by the number of focal elements. The maximum amplitude limit state function of the axisymmetric trajectory constructed in 6 partitions is directly used to calculate the confidence and similarity of 4-part and 8-part partitions.

[0109] from Figure 11 As can be seen, comparing the proposed method with the traditional evidence theory method for partitions 4, 6, and 8, within the entire oil film thickness safety threshold variation range of 0.01-0.5 μm, regardless of whether the cognitive uncertainty parameter is divided into a coarser 4 sub-intervals or a finer 6 or 8 sub-intervals, the cumulative confidence curve and cumulative plausibility curve obtained by the proposed method almost completely overlap with those of the traditional evidence theory method. This result strongly demonstrates that, thanks to the precise capture of the boundary by the control points and the driving force of the sequence iteration mechanism, the response surface surrogate model constructed by the proposed method accurately fits the true maximum amplitude limit state surface of the axis trajectory, and the computational cost of this method is not affected by the number of focal elements.

[0110] To visually demonstrate the advantages of the sample genetic management technique and sequence iteration mechanism in this invention—avoiding repeated calls to the dynamic mesh model and reducing computational costs—sample points were extracted for three conditions under six intervals: oil film safety thresholds of 0.01µm, 0.02µm, and 0.03µm. The results were plotted as follows: Figure 12The sample collection distribution diagram is shown. The black circles, red crosses, and green plus signs represent sample points when the oil film safety thresholds are 0.01µm, 0.02µm, and 0.03µm, respectively. These samples do not exhibit a traditional globally dense distribution, but rather are highly clustered in the high-load, high-eccentricity local region in the upper right corner, intuitively demonstrating that the sequence iteration mechanism of this invention has the ability to adaptively track real failure points. Furthermore, the sample markers under different thresholds in the diagram show extremely obvious spatial overlap characteristics. This highly overlapping visual distribution constitutes a direct physical representation of the sample genetic management technology of this invention—when the oil film safety threshold changes, the method of this invention successfully identifies and directly inherits the previously solved identical sample point data, avoiding blind global repeated sampling and redundant calls to complex dynamic mesh models, thus supporting the improvement of the computational efficiency of the method of this invention.

[0111] While ensuring the same extremely high accuracy, we further compare and analyze the ability of each method to overcome the computational bottleneck caused by the combined explosion of evidence theory methods. Figure 13 The diagram illustrates the number of dynamic mesh computation model calls required for each method under interval partitioning of 4, 6, and 8. Clearly, the computational cost of the traditional evidence theory method suffers from a combinatorial explosion problem as the number of interval partitions increases, with the number of calls increasing exponentially. Specifically, for the two cognitively uncertain variables of mutational overloading and initial eccentricity parameters in this invention, each with a BPA structure containing 4, 6, and 8 sub-intervals, the traditional method (calling the dynamic mesh model only twice per focal element) requires computation to complete the stability assessment. , , The proposed method significantly improves computational efficiency by leveraging the dual core advantages of sample genetic management computation and the independence of control points from focal element partitioning. Taking a 6-interval evaluation framework as an example, with an oil film safety threshold of 0.01 μm, 72 dynamic mesh simulations and a historical database need to be performed. When the oil film safety threshold evolves to 0.02 μm, through sample genetic management technology, historical samples with overlapping spatial locations are directly inherited, requiring only 13 additional simulations, and the sample database expands to 85 samples (72 + 13 = 85).

[0112] As the oil film safety threshold continues to evolve towards 0.03 μm, 0.04 μm, and higher levels, the method of this invention will continuously perform intelligent retrieval and reuse in the ever-expanding historical sample database. For example, at 0.03 μm and 0.04 μm, only 13 new simulations are required. Even though the number of new simulations per instance may dynamically change due to the nonlinear shift of the limit state surface in subsequent threshold changes, the vast majority of historical samples are effectively reused.

[0113] This sample genetic management technique based on overlapping elimination and rolling inheritance reduces the number of calls to the dynamic mesh model to an extremely low level throughout the entire threshold dynamic change process, significantly alleviating the computing power bottleneck.

[0114] Furthermore, within the 6-interval evaluation framework, the implicit complex fluid physics relationships have been successfully transformed into a response surface surrogate model with a high-precision explicit quadratic polynomial mathematical expression covering the entire oil film instability hazard region. This surrogate model, constructed based on control points, is independent of focal element partitioning. When the stability evaluation switches to a coarser 4-interval or a finer 8-interval, the surrogate model of the maximum amplitude limit state function of the axisymmetric trajectory under the 6-interval framework is directly invoked, without needing to call any dynamic mesh calculation model, thus enabling… Figure 13 The number of physical simulation calls for the Zhongben method in the 4-interval and 8-interval ranges is reduced to zero.

[0115] Based on the above technical solution, the following beneficial effects are achieved: To address the problem of focal explosion caused by partitioning in the measurement of cognitive uncertainty in evidence theory, i.e., the more partitions there are, the more accurate the results, but the computational cost increases exponentially, this invention searches for the intersection points of the limit state function surface and the uncertainty domain, i.e., control points, and constructs a surrogate model based on the control points to accurately capture the key position information of the limit state surface that determines the instability boundary of the oil film. Specifically, by arranging samples in the geometrically uncertain domain based on control points and constructing a response surface surrogate model of the true limit state function, the accuracy of constructing the quadratic polynomial response surface of the maximum amplitude limit state of the axisymmetric trajectory is effectively improved. This reduces the number of times the dynamic mesh calculation model needs to be called to construct a high-precision response surface surrogate model across the entire domain, thereby improving computational efficiency and solving the problems of insufficient fitting accuracy and excessive sample call volume of existing global response surface models. Moreover, the surrogate model constructed based on control points is independent of the number of foci elements in the evidence theory, thus solving the computational efficiency problem caused by foci element explosion in the oil film stability assessment of the evidence theory.

[0116] Meanwhile, this invention introduces sample genetic management technology to compare and screen new and old samples during the iterative sampling process, avoiding repeated calls to the dynamic mesh computing model to calculate the axis trajectory, and further significantly reducing the computational load; it also introduces a sampling center sequence iteration mechanism to make the sample collection area approximate the oil film's maximum possible instability area, thereby improving the accuracy of the surrogate model built with a small number of samples; Finally, the reliability and similarity of the evaluation of oil film stability are quickly calculated using the response surface surrogate model, which effectively promotes the application of evidence theory in the evaluation of oil film stability of sliding bearings under complex working conditions such as wind power, and alleviates the efficiency and accuracy problems in the evaluation of oil film stability of wind power sliding bearings.

[0117] This invention significantly improves the evaluation calculation efficiency while ensuring that the accuracy of oil film stability assessment meets engineering requirements. It effectively broadens the engineering application of evidence theory in the evaluation of oil film stability of sliding bearings under complex working conditions of strong sudden changes and heavy loads, enhances the practicality of intelligent monitoring technology for sliding bearing stability, and provides reliable technical support for condition monitoring, fault early warning and operation and maintenance optimization of wind power equipment.

[0118] This invention also provides a rapid wind turbine sliding bearing oil film stability assessment system 1500, see [link to documentation]. Figure 15 ,include: Uncertainty domain construction module 1510 is used to construct the geometric uncertainty domain of an n-dimensional polyhedron by taking the core factors affecting the axis trajectory as cognitive uncertainty variables within the identification framework of evidence theory, and using the intervals of each cognitive uncertainty variable together. The control point search module 1520, connected to the uncertainty domain construction module 1510, is used to traverse all edges of the geometric uncertainty domain of the n-dimensional polyhedron to obtain control points. The control points are the intersections of each edge with the limit state surface of the maximum amplitude of the oil film axis trajectory; n is a natural number greater than 0. The proxy model construction module 1530 is connected to the control point search module 1520. It is used to calculate the average value of all control points as the initial sample center point, collect samples, screen them through sample genetic management technology, and substitute them into the wind power sliding bearing dynamic mesh calculation model to calculate the maximum amplitude of the shaft center trajectory. Based on the maximum amplitude of the shaft center trajectory and the core factors, a response surface proxy model is constructed. The sequence iteration optimization module 1540 is connected to the surrogate model construction module 1530. It is used to solve the maximum possible failure point based on the response surface surrogate model, update the maximum possible failure point to a new initial sample center point, re-enter the surrogate model construction module, and continue until the initial sample center points on both sides meet the convergence condition to obtain a converged response surface surrogate model. The extreme value analysis module 1550, connected to the sequence iterative optimization module 1540, is used to replace the dynamic grid calculation model with a converged response surface surrogate model to perform extreme value analysis on the joint focal element formed by the interval intersection of various cognitive uncertainty variables, and obtain the maximum and minimum values ​​corresponding to each joint focal element. The stability assessment module 1560, connected to the control point search module 1520 and the extreme value analysis module 1550, is used to calculate the confidence function value and the similarity function value of the oil film stability of the wind turbine sliding bearing based on the maximum and minimum values ​​of the joint coke element, change the preset oil film safety threshold, repeatedly enter the control point search module, and draw the cumulative confidence and cumulative similarity curves to conduct oil film stability assessment.

[0119] For other details, please refer to the methods described above; they will not be repeated here.

[0120] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0121] The acquisition, storage, and application of user personal information involved in the technical solution of this invention all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0122] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the described module can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0123] Figure 16 A schematic block diagram of an electronic device 1600 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0124] Electronic device 1600 includes a computing unit 1601, which can perform various appropriate actions and processes according to a computer program stored in ROM 1602 or a computer program loaded into RAM 1603 from storage unit 1608. RAM 1603 may also store various programs and data required for the operation of electronic device 1600. The computing unit 1601, ROM 1602, and RAM 1603 are interconnected via bus 1604. I / O interface 1605 is also connected to bus 1604.

[0125] Multiple components in electronic device 1600 are connected to I / O interface 1605, including: input unit 1606, such as keyboard, mouse, etc.; output unit 1607, such as various types of displays, speakers, etc.; storage unit 1608, such as disk, optical disk, etc.; and communication unit 1609, such as network card, modem, wireless transceiver, etc. Communication unit 1609 allows electronic device 1600 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0126] The computing unit 1601 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 1601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1601 performs the various methods and processes described above. For example, in some embodiments, the methods may be implemented as computer software programs tangibly contained in a machine-readable medium, such as storage unit 1608. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 1600 via ROM 1602 and / or communication unit 1609. When the computer program is loaded into RAM 1603 and executed by the computing unit 1601, one or more steps of the rapid wind turbine sliding bearing oil film stability assessment method described above can be performed. Alternatively, in other embodiments, the computing unit 1601 may be configured to perform the rapid wind turbine sliding bearing oil film stability assessment method by any other suitable means (e.g., by means of firmware).

[0127] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0128] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A quick wind turbine plain bearing oil film stability evaluation method, characterized in that include: S1. Within the framework of evidence theory, the core factors influencing the axis trajectory are taken as cognitive uncertainty variables, and the intervals of each cognitive uncertainty variable are used to construct the geometric uncertainty domain of an n-dimensional polyhedron; n is a natural number greater than 0. S2. Traverse all edges of the geometric uncertainty domain of the n-dimensional polyhedron to obtain control points, where the control points are the intersections of each edge with the limit state surface of the maximum amplitude of the oil film axis trajectory. S3. Calculate the average value of all control points as the initial sample center point, expand the sample points based on the initial sample center point, use sample genetic management technology to determine whether to substitute the expanded sample points into the wind power sliding bearing dynamic mesh calculation model to calculate the maximum amplitude of the shaft center trajectory, and construct a response surface proxy model based on the maximum amplitude of the shaft center trajectory and core factors. The response surface proxy model is constructed based on a preset oil film safety threshold. S4. Solve for the maximum possible failure point based on the response surface surrogate model, and update the maximum possible failure point to a new initial sample center point. Repeat S3 until the initial sample center points on both sides meet the convergence condition, and obtain a converged response surface surrogate model. S5. Replace the dynamic grid calculation model with a convergent response surface surrogate model, and perform extreme value analysis on the joint focal element formed by the interval intersection of each cognitive uncertainty variable to obtain the maximum and minimum values ​​corresponding to each joint focal element. S6. Based on the maximum and minimum values ​​of the combined coke element, calculate the reliability function value and the similarity function value of the oil film stability of the wind turbine sliding bearing. S7. Change the preset oil film safety threshold and repeat S2 to S6 to plot the cumulative confidence and cumulative fidelity curves for oil film stability assessment.

2. The rapid wind turbine plain bearing oil film stability evaluation method according to claim 1, characterized in that In S2, traversing all edges of the geometric uncertainty domain of the n-dimensional polyhedron to obtain control points includes the following steps: On the edges of the geometric uncertainty domain of the n-dimensional polyhedron, only the cognitive uncertainty variable corresponding to that edge is allowed to change continuously within the interval, while other cognitive uncertainty variables are fixed to the endpoint values ​​of their respective edge intervals; The center point of the interval corresponding to the continuously changing cognitive uncertainty variable is used as the initial iteration value for iterative calculation until the preset convergence criterion is met, and the control point is obtained. Among them, it is determined whether the iterative process has converged or whether the obtained cognitive uncertainty variable value is within the corresponding interval. If not, the intersection of the corresponding edge and the maximum amplitude limit state surface of the oil film axis trajectory is removed.

3. The rapid method for evaluating the stability of oil film in wind turbine sliding bearings according to claim 1, characterized in that, In S3, the process of collecting samples, screening them using sample genetic management technology, and then inputting them into the dynamic mesh calculation model of the wind turbine sliding bearing to calculate the maximum amplitude of the shaft center trajectory includes the following steps: Determine if the collected sample exists in the historical sample database. If it does, directly obtain the maximum amplitude of the axis trajectory corresponding to the sample point in the historical sample database. Otherwise, the sample is input into the dynamic mesh computing model for calculation, and the calculation result is stored in the historical sample database.

4. The rapid method for evaluating the stability of oil film in wind turbine sliding bearings according to claim 1, characterized in that, In S3, a response surface proxy model is constructed based on the maximum amplitude of the axis trajectory and the core factors, as shown in the following formula: ； in, The dimension of variables representing uncertainties in the evidence. , and All represent response surface coefficients. This represents the i-th cognitive uncertainty parameter. Represents the limit state function of the maximum amplitude of the axis trajectory The response surface, This indicates the preset oil film safety threshold. This represents the maximum amplitude of the axis trajectory obtained by calling the dynamic mesh calculation model.

5. The rapid method for evaluating the stability of oil film in wind turbine sliding bearings according to claim 1, characterized in that, In S4, the most likely failure point is updated to a new initial sample center point, and S3 is repeated until the initial sample center points on both sides satisfy the convergence condition, resulting in a converged response surface surrogate model, including: S41. Generate axially expanded sample points again using the new initial sample center point, and supplement them into the initial sample space after screening by sample genetic management technology, and update the response surface surrogate model coefficients simultaneously. S42. Repeat S41 until the initial sample center points of two adjacent iterations satisfy the iteration convergence criterion, and obtain the converged response surface surrogate model.

6. The rapid method for evaluating the oil film stability of wind turbine sliding bearings according to claim 1, characterized in that, In S6, based on the maximum and minimum values ​​of the combined coke element, the reliability function value and the similarity function value of the oil film stability of the wind turbine sliding bearing are calculated, including the following steps: The joint focal element is input into the basic confidence assignment function to obtain the basic confidence assignment value corresponding to each joint focal element; Determine whether to include the basic confidence assignment value corresponding to the joint focal element in the confidence function value and the similarity function value based on the relationship between the joint focal element and the constructed stability region. Specifically, it is determined whether the maximum and minimum values ​​of the joint focal element are both within the stable region. If they are both within the stable region, the basic confidence assignment value corresponding to the joint focal element is included in the confidence function value and the similarity function value. Otherwise, determine whether the maximum value of the joint focal element is within the stable region and the minimum value is not within the stable region. If the maximum value of the joint focal element is within the stable region and the minimum value is not within the stable region, then include the basic credibility allocation value corresponding to the joint focal element in the similarity function value. Otherwise, determine whether the maximum and minimum values ​​of the joint focal element are both outside the stable region. If so, the basic confidence assignment value corresponding to the joint focal element is not included in the confidence function value and the similarity function value.

7. A rapid wind turbine sliding bearing oil film stability assessment system, used to implement the rapid wind turbine sliding bearing oil film stability assessment method as described in any one of claims 1 to 6, characterized in that, include: The uncertainty domain construction module is used to construct the geometric uncertainty domain of an n-dimensional polyhedron by taking the core factors affecting the axis trajectory as cognitive uncertainty variables within the identification framework of evidence theory, and using the intervals of each cognitive uncertainty variable together. The control point search module, connected to the uncertainty domain construction module, is used to traverse all edges of the geometric uncertainty domain of the n-dimensional polyhedron to obtain control points. The control points are the intersections of each edge with the limit state surface of the maximum amplitude of the oil film axis trajectory; n is a natural number greater than 0. The proxy model construction module is connected to the control point search module. It is used to calculate the average value of all control points as the initial sample center point, expand the sample points based on the initial sample center point, and use sample genetic management technology to determine whether to substitute the expanded sample points into the wind turbine sliding bearing dynamic mesh calculation model to calculate the maximum amplitude of the shaft center trajectory. Based on the maximum amplitude of the shaft center trajectory and the core factors, a response surface proxy model is constructed. The response surface proxy model is constructed based on a preset oil film safety threshold. The sequence iterative optimization module is connected to the surrogate model construction module. It is used to solve the maximum possible failure point based on the response surface surrogate model, update the maximum possible failure point as a new initial sample center point, re-enter the surrogate model construction module, and continue until the initial sample center points on both sides meet the convergence condition to obtain a converged response surface surrogate model. The extreme value analysis module, connected to the sequence iterative optimization module, is used to replace the dynamic grid calculation model with a converged response surface surrogate model to perform extreme value analysis on the joint focal elements formed by the interval intersection of various cognitive uncertainty variables, and obtain the maximum and minimum values ​​corresponding to each joint focal element. The stability assessment module, connected to the control point search module and the extreme value analysis module, is used to calculate the confidence function value and the similarity function value of the oil film stability of the wind turbine sliding bearing based on the maximum and minimum values ​​of the joint coke element. It changes the preset oil film safety threshold, repeatedly enters the control point search module, and plots the cumulative confidence and cumulative similarity curves to assess the oil film stability.

8. An electronic device, characterized in that, It includes a memory, a processor, and a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor to implement a rapid method for evaluating the oil film stability of wind turbine sliding bearings as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, It stores a computer program; the computer program is executed by a processor to implement a rapid method for evaluating the stability of the oil film in wind turbine sliding bearings as described in any one of claims 1 to 6.

Images