Step hardness wind power laser cladding journal structure design method based on proxy model optimization algorithm

Abstract

The invention discloses a stepped hardness wind power laser cladding journal structure design method based on an agent model optimization algorithm. Along with the large-scale and open-sea development of wind turbines, the requirement for the reliability of bearings is increasingly improved. A traditional journal bearing design method still faces the problems of high abrasion and failure, and the advantages of a laser cladding manufacturing technology cannot be fully exerted. By combining a laser cladding technology with an advanced computational algorithm, the invention provides a systematic method, and aims to design a journal structure with an optimized stepped hardness configuration. And by adopting an agent model optimization algorithm, various design parameters can be simulated and analyzed, so that the optimal setting is determined, and the wear resistance and the mechanical property are improved to the greatest extent. According to the invention, the problem of wear failure caused by long-time operation of the sliding bearing of the wind power gear box is effectively solved, and a new solution is provided for manufacturers for improving the performance of a wind power generation assembly.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation, and in particular to the design of a shaft journal for a wind generator gear box. Background Art

[0002] As demand for renewable energy grows, wind power has become a key player in sustainable energy generation. Wind turbine gearboxes are essential components for optimal performance, converting the turbine's rotational energy into electricity. A critical aspect of gearbox design is the journal bearing, which must withstand high loads while minimizing friction. To improve the performance and lifespan of these bearings, advanced manufacturing techniques, such as laser cladding, are increasingly being used.

[0003] Laser cladding, as a surface enhancement technology, can significantly improve the durability of journal bearings by achieving a stepped hardness profile through tailored processing parameters. However, currently available monolithic designs may not fully exploit the potential benefits of laser cladding and maximize the step hardness profile of the cladding layer, which can compromise wear resistance and load-bearing capacity. To address these limitations, advanced design methods utilizing optimization algorithms are needed to develop an optimal stepped hardness profile. This, in turn, can improve the performance and reliability of laser-clad journal bearings in wind turbine applications. Summary of the Invention

[0004] To address the high wear and failure issues of wind turbine journal bearings operating in harsh environments over long periods of time, a step-hardness design method for wind turbine laser-clad journal structures based on a surrogate model optimization algorithm is proposed, combining recently developed efficient global optimization techniques with current laser cladding processing methods. Specifically, this method first analyzes the operating characteristics of wind turbine journal bearings in harsh environments and sets the optimization objective: minimizing wear. A step-hardness distribution scheme is designed. A wear database is generated through simulation of a friction and wear model of a wind turbine gearbox journal bearing. A surrogate model is constructed using the least squares support vector regression (LSSVR) algorithm to establish a relationship between hardness distribution and wear performance. The surrogate model is iteratively optimized using an efficient global optimization algorithm (EGO) to identify the optimal hardness distribution. This method not only improves design efficiency but also provides new insights for practical applications, significantly enhancing the wear resistance and reliability of wind turbine journal bearings in harsh environments.

[0005] The technical solution adopted by the present invention to solve the technical problem includes the following steps:

[0006] Step 1: Determine the design parameters, define the hardness distribution scheme, design multiple hardness configurations for the journal structure, and define the hardness value of each configuration and its distribution position on the bearing surface;

[0007] Step 2, build a friction and wear model for the wind turbine gearbox journal bearing to accurately simulate the wear behavior of the bearing under different working conditions, calculate the wear under each hardness distribution, and establish a wear database;

[0008] Step 3, develop a proxy model, use least squares support vector regression algorithm to build a proxy model, and combine high-efficiency global optimization algorithm for optimization;

[0009] Step 4, efficient algorithm solution, for optimization problem, convert the constrained optimization problem into an unconstrained optimization problem;

[0010] Step 5, optimization algorithm, according to the pre-defined performance standard (minimum wear), iteratively adjust the parameters.

[0011] Preferably, the hardness distribution scheme needs to be set according to the material properties and specific application conditions, and the target hardness range of the journal cladding layer is divided into multiple regions, and different step hardness values are assigned;

[0012] Preferably, the friction and wear model of the wind turbine gearbox journal bearing selects the Archard wear model, and the formula for wear is

[0013]

[0014] Where V is the wear amount in a certain time, K is the dimensionless wear coefficient, H is the surface material hardness, s is the sliding distance in a certain time, F N is the contact force.

[0015] Preferably, the wear database is simulated by a simulation tool, and the Archard wear model is used to simulate the wear under different hardness configurations, and a sliding bearing wear database containing different hardness configurations, wear conditions and related performance is obtained;

[0016] Preferably, the specific steps of the proxy model are:

[0017] Step 3.1, divide the surface nodes: divide the journal surface into multiple nodes, each node has a unique hardness value, assign a hardness value H i = f(x i ,y i ) to each node on the bearing surface;

[0018] Step 3.2, initial sample set D = {(x1,y1),(x1,y2)...(x i ,y i )...(x N ,y N )}, x i ∈ R * ,y i∈ R;

[0019] Step 3.3, according to the principle of SVR agent model, constraint minimization is a cost function w is the weight vector of the initial space;

[0020] Step 3.4, constraint condition is a kernel function, i is a bias variable, and b is a bias term;

[0021] Step 3.5, the constructed linear regression model is:

[0022] Preferably, the constraint optimization problem is converted into an unconstrained optimization problem, and the specific steps are as follows:

[0023] Step 4.1, introduce Lagrange function: a i is the Lagrange multiplier;

[0024] Step 4.2, the expression for solving the optimization problem is as follows:

[0025] Step 4.3, nonlinear mapping Kernel function:

[0026] Step 4.4, solve the equation set, and the final obtained regression model can be represented as:

[0027] Step 4.5, LSSVR agent model, adopts Gaussian kernel function as its kernel function, and its formula definition is as follows:

[0028]

[0029] Preferably, the optimization algorithm calculates the uncertainty through the KRG model, generates a new sample point through the optimization EI(x) criterion, and adds the new sample point to the sample set for circulation, and stops calculation if the condition is met.

[0030] The beneficial effects of the present application are:

[0031] (1) By improving the hardness distribution of laser cladding journal, the journal structure has stronger fatigue resistance and grinding capacity, prolongs the service life of them, and reduces the frequency of maintenance and replacement.

[0032] (2) It can adapt to various operating conditions and specific requirements of different wind turbine models, thereby providing customized solutions.

[0033] (3) Systematic approach reduces the need for extensive experimental testing, reducing development costs and time. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 is a flow chart of a stepped hardness wind power laser cladding journal structure design method based on a proxy model optimization algorithm;

[0035] Figure 2 is a flow chart of a least squares support vector regression-high efficiency global optimization algorithm (LSSVR-EGO) algorithm; DETAILED DESCRIPTION

[0036] The preferred embodiments of the present application will be described in detail below with reference to the accompanying drawings, so that the advantages and features of the present application can be more easily understood by those skilled in the art, and the scope of protection of the present application can be more clearly defined.

[0037] Referring to Figure 1 , the embodiments of the present application include:

[0038] Figure 1 A flow chart of a stepped hardness wind power laser cladding journal structure design method based on a proxy model optimization algorithm, specifically includes: wear data set acquisition (given design parameters, define hardness distribution scheme, sliding bearing wear model construction, bearing wear database building), high efficiency global optimization algorithm (establish proxy model, calculate uncertainty, optimization criteria generate new sample points, add sample points to data set, obtain optimal hardness distribution configuration). The structure design method can be used for new type of planetary wheel sliding bearing of wind power gear box, efficiently obtain the hardness distribution characteristics and rules of laser cladding on the bearing surface, has higher wear resistance and reliability compared with uniform hardness distribution surface, and has good popularization and application value.

[0039] As Figure 1 shown, a stepped hardness wind power laser cladding journal structure design method based on a proxy model optimization algorithm includes the following steps:

[0040] Step 1, determine the design parameters, define the hardness distribution scheme, design multiple hardness configurations for the journal structure, and define the hardness value of each configuration and its distribution position on the bearing surface;

[0041] Step 2, build a friction and wear model for the journal bearing of the wind power gear box to accurately simulate the wear behavior of the bearing under different working conditions, calculate the wear under each hardness distribution, and establish a wear database;

[0042] The hardness distribution scheme needs to set the target hardness range of the journal neck cladding layer according to the material characteristics and specific application conditions, divide the journal neck into multiple regions in the axial direction, and assign different step hardness values; the friction and wear model of the wind power gear box journal bearing is selected as the Archard wear model.

[0043] Step 3, develop a proxy model, use the least squares support vector regression algorithm to build a proxy model, and combine a high-efficiency global optimization algorithm for optimization.

[0044] Step 3.1, divide the surface nodes: divide the journal surface into multiple nodes, each node has a unique hardness value, and assign a hardness value H to each node on the bearing surface i =f(x i ,y i );

[0045] Step 3.2, initial sample set D={(x1,y1),(x1,y2)...(x i ,y i )...(x N ,y N )},x i ∈R * ,y i ∈R;

[0046] Step 3.3, according to the principle of SVR proxy model, the constraint minimization is the cost function w is the weight vector of the initial space, and

[0047] Step 3.4, constraint condition is the kernel function, i is the bias variable, and b is the bias term

[0048] Step 3.5, construct a linear regression model:

[0049] Step 4, high-efficiency algorithm solution, for optimization problems, convert the constrained optimization problem into an unconstrained optimization problem

[0050] Step 4.1, introduce the Lagrange function: a i is the Lagrange multiplier;

[0051] Step 4.2, the expression of the optimization problem is as follows:

[0052] Step 4.3, nonlinear mapping kernel function of

[0053] Step 4.4, solving the equation set, and finally the regression model obtained can be expressed as:

[0054] Step 4.5, LSSVR surrogate model, adopts Gaussian kernel function as its kernel function, and its formula definition is as follows:

[0055]

[0056] Step 5, optimization algorithm, according to the predefined performance standard (minimum wear), iteratively adjust the parameters.

[0057] Wherein, the optimization algorithm calculates the uncertainty through the KRG model, generates a new sample point through the optimization EI(x) criterion, adds the new sample point to the sample set for circulation, and stops calculation if the condition is met.

[0058] Figure 2 The least square support vector regression-efficient global optimization algorithm (LSSVR-EGO) algorithm flow chart, specifically includes: initial sample set, LSSVR surrogate model, calculating the uncertainty of the surrogate model according to the KRG method, obtaining the current optimal solution by maximizing the EI(x) criterion and updating the sample point, judging whether to reach the NFEs stop criterion, adding the updated sample point to the current sample set, and designing the step hardness distribution.

[0059] Wherein

[0060] The step hardness wind power laser cladding journal structure design method based on the proxy model optimization algorithm, the present application utilizes the friction and wear model and LSSVR-EGO optimization algorithm, provides a new design method for the step hardness laser cladding journal structure in the application of wind power generation. This method effectively solves the challenge of sliding bearing wear, promotes the progress of wind turbine technology, and improves the reliability and efficiency of renewable energy systems.

[0061] The above only describes the embodiments of the present application, and does not limit the patent range of the present application, and any equivalent structure or equivalent process transformation using the content of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection range of the present application.

Claims

1. A step-hardness wind turbine laser cladding journal structure design method based on surrogate model optimization algorithm, characterized in that: include: Step 1: Determine the design parameters, define the hardness distribution scheme, design multiple hardness configurations for the journal structure, and define the hardness value of each configuration and its distribution position on the bearing surface; Step 2: Build a friction and wear model for the journal bearing of a wind turbine gearbox to accurately simulate the wear behavior of the bearing under different operating conditions, calculate the wear conditions under each hardness distribution, and establish a wear database; Step 3: Develop a proxy model using the least squares support vector regression algorithm and optimize it with an efficient global optimization algorithm. Step 4: The algorithm efficiently solves the optimization problem and transforms the constrained optimization problem into an unconstrained optimization problem; Step 5: Optimize the algorithm and iteratively adjust the parameters according to the predefined performance criteria (minimum wear).

2. The step-hardness wind turbine laser cladding journal structure design method based on the surrogate model optimization algorithm according to claim 1 is characterized in that In step 1, the hardness distribution scheme needs to set the target hardness range of the journal cladding layer according to the material properties and specific application conditions, divide the journal axially into multiple areas, and assign different step hardness values.

3. The step-hardness wind turbine laser cladding journal structure design method based on the surrogate model optimization algorithm according to claim 1 is characterized in that The friction and wear model of the wind turbine gearbox journal bearing in step 2 uses the Archard wear model, and the wear formula is: The wear database uses simulation tools and the Archard wear model to simulate the wear conditions under different hardness configurations, and summarizes the sliding bearing wear database that includes different hardness configurations, wear conditions and related performance.

4. The step-hardness wind turbine laser cladding journal structure design method based on the surrogate model optimization algorithm according to claim 1 is characterized in that The optimization algorithm described in step 5 calculates uncertainty through the KRG model, generates a new sample point by optimizing the EI(x) criterion, adds the new sample point to the sample set and cycles, and stops the calculation if the conditions are met.

Images