A performance optimization method and system for a heavy machine tool hydrostatic guideway system

Abstract

The application discloses a performance optimization method for a heavy machine tool hydrostatic guide rail system, comprising the following steps: S1. establishing a multi-objective optimization model: determining design variables of the hydrostatic guide rail system X , constructing an optimization objective function F(X) , and establishing an equality constraint based on force and torque balance H(X and an inequality constraint for preventing dry friction G(X) ; S2. constructing a multi-fidelity database: obtaining a high-fidelity data set through high-fidelity numerical model simulation, and obtaining a low-fidelity data set through low-fidelity numerical model simulation with the same calculation logic but a lower grid density; S3. constructing a data fusion prediction model: utilizing the high-fidelity data set and the low-fidelity data set, constructing a collaborative Kriging surrogate model based on an autoregressive structure, and outputting a predicted mean and a predicted variance; S4. generating a candidate solution population; S5. multi-criteria point selection; S6. simulating and updating a Pareto frontier solution set; and S7. repeating steps S3 to S6 until a preset convergence condition or a calculation budget is met, and outputting a final Pareto optimal solution set.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of computational intelligent optimization and high-end equipment manufacturing, and in particular relates to a performance optimization method and system for hydrostatic guide rail systems of heavy machine tools. Background Technology

[0002] In the field of high-end equipment manufacturing, heavy-duty machine tools are the core equipment for processing key components in national strategic sectors such as aviation, aerospace, and energy. These machine tools need to maintain micron-level machining accuracy under extremely high loads, therefore, the performance of their core load-bearing unit—the hydrostatic guideway system—including stiffness, load-bearing capacity, and positional accuracy, is of paramount importance.

[0003] Hydrostatic guideways are commonly used in heavy-duty machine tools as load-bearing systems for moving structures, including but not limited to the spindle box, worktable, and slide. A hydrostatic guideway system typically consists of multiple hydrostatic guideways. The overall performance of this system is determined by its complex structural parameters, including the individual geometric dimensions, oil cavity depth, oil supply orifice diameter, and oil supply pressure of each hydrostatic guideway. In engineering design, evaluating the merits of a set of design parameters requires high-fidelity fluid physics simulation models. While such high-fidelity models can provide accurate performance predictions, their individual calculations are very time-consuming.

[0004] However, finding the optimal design combination in the aforementioned multidimensional parameter space is a complex nonlinear optimization problem. Existing optimization algorithms, such as genetic algorithms (GA) or evolutionary algorithms (EA) like particle swarm optimization (PSO), require thousands or even tens of thousands of iterations of the objective function (i.e., the high-fidelity FDM model) to search for the global optimum.

[0005] Therefore, existing technologies face a core contradiction: population-based evolutionary algorithms require thousands or even tens of thousands of function evaluations to find the global optimum; on the other hand, high-fidelity fluid dynamics models relying on solving the Reynolds equation or the finite element method are computationally extremely expensive, and each simulation takes a long time, especially when dealing with a large number of samples. Using simplified models or empirical formulas to calculate oil film stiffness ignores stiffness variations under complex operating conditions; while computation is fast, it sacrifices physical realism, leading to unreliable optimization results. Although some studies have used surrogate models to approximate high-fidelity models in optimization, they generally lack an intelligent infill criterion, failing to effectively balance the exploration of unknown regions and the optimization of known optimal regions, resulting in low optimization efficiency. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide an efficient and accurate method for optimizing design parameters for hydrostatic guide rail systems of medium and large machine tools. It discloses a performance optimization method and system for hydrostatic guide rail systems of heavy machine tools, which effectively improves the load-bearing capacity and accuracy of hydrostatic guide rail systems.

[0007] The objective of this invention is achieved through the following technical solution: A performance optimization method for hydrostatic guideway systems in heavy-duty machine tools includes: S1. Establish a multi-objective optimization model: Determine the design variables of the hydrostatic guide rail system. X Construct an optimization objective function that includes translational stiffness, overturning stiffness, and positioning posture deviation. F(X) And establish equation constraints based on the balance of force and torque. H(X ) and inequality constraints to prevent dry friction G(X) ; S2. Constructing a multi-fidelity database: Obtaining high-fidelity datasets through high-fidelity numerical model simulations. Low-fidelity datasets are obtained through simulations using low-fidelity numerical models with the same computational logic but lower grid density. ; S3. Construct a data fusion prediction model: using the aforementioned high-fidelity dataset With low-fidelity datasets A co-kriging surrogate model based on an autoregressive structure is constructed for handling arbitrary design variables. X The objective function and constraints are used to make predictions, and the predicted mean and predicted variance are output. S4. Generating a candidate solution population: Based on a high-fidelity dataset The Pareto front solution set in the model is used to construct a probabilistic model. Combined with the sampling distribution constructed from the prediction variance, a population of candidate solutions is generated using the cross-entropy fusion method. ; S5. Multi-criteria selection: For the candidate solution population... Perform constraint filtering to remove constraints that do not satisfy the inequality. G(X) Solutions with probabilities below a threshold; calculate the hypervolume index increment resulting from incorporating the remaining candidate solutions into the current Pareto front, and select several design variables with the largest increments to form an addition set. ; S6. Simulation Update: Use the high-fidelity numerical model to update the point set. Perform simulation calculations to obtain accurate responses and update them to a high-fidelity database. At the same time, update the Pareto front solution set; S7. Iterative optimization: Repeat steps S3 to S6 until the preset convergence condition or computation budget is met, and output the final Pareto optimal solution set.

[0008] Furthermore, the design variables X It contains i sets of parameters, corresponding to i hydrostatic guide rails; each set of parameters includes: a set of outer contour dimensions of the hydrostatic guide rail. Oil cavity size parameter set Oil cavity depth Oil supply hole diameter and film thickness design value .

[0009] Furthermore, the optimized target vector F(X) Specifically, this includes: maximizing translational stiffness. The reciprocal of the maximization of overturning stiffness The reciprocal of the value and minimizing the positioning pose deviation Wherein, the positioning pose deviation The bearing capacity of each guide rail plate is obtained by solving the Reynolds equation, and then the constraint is based on the aforementioned equation. H(X) =0 is the five-dimensional attitude positioning deviation obtained by iterative solution.

[0010] Furthermore, the inequality constraint G(X) The construction method is as follows: Select the points on the outer contour of each hydrostatic guide plate that are farthest from the geometric center of the machine tool hydrostatic guide system as constraint points, and calculate the actual position normal components of each constraint point after the attitude positioning deviation occurs. Normal component of the critical position where dry friction is about to occur The difference must be greater than or equal to 0.

[0011] Furthermore, the formula for the co-kriging surrogate model based on the autoregressive structure in step S3 is as follows: ,in, For high-fidelity numerical model response, Kriging interpolation model for low-fidelity numerical model response. Scalar scaling factor The deviation function is modeled as an independent Gaussian process; the hyperparameters of the covariance matrix C in the co-Kriging proxy model are obtained by the maximum likelihood estimation method.

[0012] Furthermore, the specific steps for generating the candidate solution population in step S4 include: fitting the current Pareto front solution set with a Gaussian mixture model to obtain the Pareto front probability model. Based on the prediction variance provided by the co-kriging agent model, a sampling distribution proportional to the total prediction uncertainty is constructed. Initialize the sampling distribution, and set multiple design variables in the sampling distribution. XThe response is predicted using a co-kriging surrogate model based on an autoregressive structure, and multiple design variables are incorporated. X The corresponding predicted responses are included in the Pareto front solution set to evaluate these design variables. X And the predicted response as the hypervolume index increment brought by the Pareto front, selecting the hypervolume index increment before Design variables X The corresponding predicted responses are used as a population of high-quality candidate solutions. The sampling distribution is updated to minimize the KL divergence between it and the population of high-quality candidate solutions, and then a new round of iterations is started using the updated sampling distribution. Finally, samples are drawn from the converged sampling distribution to form a population of candidate solutions. .

[0013] Furthermore, the specific method for multi-criteria point selection in step S5 is as follows: calculate each design variable in the candidate solution population. X Joint probability satisfying all inequality constraints Design variables with PoF(X) greater than a set threshold are retained; for the retained design variables, the hypervolume index increment caused by their inclusion in the current Pareto front solution set is calculated based on the predicted response of the co-kriging surrogate model. ,choose The k largest design variables are used as the addition point set.

[0014] The present invention also provides a performance optimization system for a hydrostatic guideway system of a heavy machine tool, and based on the performance optimization method for the hydrostatic guideway system of a heavy machine tool, includes: A multi-objective optimization model module is established to determine the design variables of the hydrostatic guide rail system. X Construct an optimization objective function that includes translational stiffness, overturning stiffness, and positioning posture deviation. F(X) And establish equation constraints based on the balance of force and torque. H (X ) and inequality constraints to prevent dry friction G(X) ; The multi-fidelity database module is used to obtain high-fidelity datasets through high-fidelity numerical model simulations. Low-fidelity datasets are obtained through simulations using low-fidelity numerical models with the same computational logic but lower grid density. ; The data fusion prediction model building module is used to utilize the high-fidelity dataset. With low-fidelity datasets A co-kriging surrogate model based on an autoregressive structure is constructed for handling arbitrary design variables. X The objective function and constraints are used to make predictions, and the predicted mean and predicted variance are output. Candidate solution population generation module, used for high-fidelity datasets The Pareto front solution set in the model is used to construct a probabilistic model. Combined with the sampling distribution constructed from the prediction variance, the cross-entropy method is used to fuse and generate a population of candidate solutions. ; The multi-criteria point selection module is used to select the candidate solution population. Perform constraint filtering to remove constraints that do not satisfy the inequality. G(X) Solutions with probabilities below a threshold; calculate the hypervolume index increment resulting from incorporating the remaining candidate solutions into the current Pareto front, and select several design variables with the largest increments to form an addition set. ; The simulation update module is used to update the point set using the high-fidelity numerical model. Perform simulation calculations to obtain accurate responses and update them to a high-fidelity database. At the same time, update the Pareto front solution set; The iterative optimization module is used to repeat steps S3 to S6 until the preset convergence condition or calculation budget is met, and output the final Pareto optimal solution set.

[0015] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the performance optimization method for a hydrostatic guideway system for heavy machine tools.

[0016] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the performance optimization method for a hydrostatic guideway system for heavy machine tools.

[0017] Compared with the prior art, the beneficial effects of the technical solution of the present invention are: 1. Significantly Improved Optimization Efficiency and Reduced Computational Costs: This invention employs multi-fidelity data fusion technology (based on a co-kriging surrogate model with an autoregressive structure). By fusing a large amount of low-cost, low-fidelity (sparse mesh) simulation data with a small amount of expensive, high-fidelity (dense mesh) simulation data, the number of calls to the time-consuming high-fidelity numerical model is significantly reduced while maintaining the model's prediction accuracy. This solves the problem that relying solely on high-fidelity models is extremely time-consuming, while relying solely on simplified models results in insufficient accuracy, thus significantly shortening the overall optimization time.

[0018] 2. Enhanced global optimization capability and avoidance of local optima: The hybrid optimization process of this invention innovatively uses the cross-entropy method to fuse two distributions: one is a Pareto frontier probability model learned based on historical high-performance solutions (responsible for development / utilization), and the other is an uncertainty sampling distribution constructed based on the variance of the surrogate model prediction (responsible for exploration). This mechanism ensures that the algorithm can both deeply explore known high-performance regions and actively explore those sparse "blind spots," effectively preventing it from getting trapped in local optima in a large design variable space (such as 126 dimensions).

[0019] 3. Improve the engineering feasibility and reliability of the optimized solution: This invention introduces a probability-based constraint filtering mechanism in the point selection stage. A surrogate model is used to predict the probability PoF(X) that the design variables satisfy the inequality constraints (preventing dry friction), and low-probability solutions are eliminated. This ensures that the solutions submitted to the high-fidelity model for verification are physically highly feasible, avoiding the waste of computational resources on infeasible design schemes and directly guaranteeing the safety of the hydrostatic guide rail under actual working conditions.

[0020] 4. Significantly improves machine tool machining performance and accuracy: This invention directly uses translational stiffness, overturning stiffness, and five-dimensional positioning posture deviation as joint optimization objectives. Through optimization, a Pareto optimal solution set can be obtained, enabling the guide rail system to have stronger vibration and deformation resistance when subjected to heavy cutting loads, while minimizing systematic positioning errors, thereby directly improving the machining surface quality and geometric accuracy of heavy-duty machine tools. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the method of the present invention.

[0022] Figure 2 This is a schematic diagram of the structure of the first layer of hydrostatic guide rails (7 pieces) of the heavy-duty floor-type milling and boring machine involved in the embodiment of the present invention. The second layer has the same distribution, and the fourth piece of each layer is an adjustable hydrostatic guide rail plate.

[0023] Figure 3a and Figure 3b These are schematic diagrams of the support force distribution of the basic structure of the hydrostatic guide plate and the FDM discretized model of the oil film.

[0024] Figure 4 This is a flowchart illustrating the process of generating a population of candidate solutions. Detailed Implementation

[0025] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention.

[0026] Example 1 This embodiment provides a performance optimization method for a hydrostatic guideway system of a heavy machine tool. (See...) Figure 1 Specifically, it includes the following steps: S1. Establish a multi-objective optimization model: In this embodiment, the hydrostatic guide rail performance optimization problem is formulated as the following multi-objective optimization problem: First, determine the design variables. X : X Among them, the design variables are determined based on the number of input variables. X It includes i Group n A matrix of design variables, which are... X In this implementation case n =9, i =14, representing that each hydrostatic guide plate has .

[0027] O i For the first i Set of outer contour dimensions of the block hydrostatic guide plate It includes length and width, a total of two design variables. S i For the first i The set of oil cavity size parameters for the block hydrostatic guide plate. There are four design variables, including the width of the four-sided sealing edge. For the first i The oil cavity depth of the block hydrostatic guide plate is a design variable. For the first i The diameter of the oil supply hole in the block hydrostatic guide plate is a design variable. For the first i The design value of the film thickness of the block hydrostatic guide plate is one design variable. There are a total of 9 design variables, and in this example, there are a total of 14 hydrostatic guide plates.

[0028] This design variable X The optimization metric is to minimize an objective function. : ; Determine the target variable based on the number of objectives to be optimized. F(X) It includes m The matrix of optimization objectives first maximizes translational stiffness. K static and overturning stiffness K moment The reciprocal is converted into flexibility, and then combined with minimizing the pose positioning deviation. Error pose .therefore, .in, This is the set of translational stiffness parameters for the hydrostatic guideway system in two directions other than the main feed direction within the coordinate system. There are two target variables in total. For the hydrostatic guide rail system around the coordinate system X,Y,Z The set of rotational stiffness parameters for the shaft. There are a total of three target variables. This is the set of five-dimensional attitude positioning deviation parameters for a hydrostatic guideway system in equilibrium, excluding the main feed motion direction (two translational degrees of freedom and three rotational degrees of freedom). There are a total of five target variables, therefore, , .

[0029] Translational stiffness and overturning stiffness are the load-bearing performance characteristics of the hydrostatic guideway system itself. These characteristics are closely related to the stability of the spindle box during machining. Greater translational stiffness and overturning stiffness can prevent vibration amplitude from being reduced when subjected to milling loads or other variable loads during machining, thereby improving the surface quality of the milled machined parts. Positioning posture deviation is the accuracy performance characteristic of the hydrostatic guideway system itself. Smaller positioning posture deviation allows for better initialization of machining position parameters, improving machining accuracy during milling.

[0030] Simultaneously satisfying a set of equality constraints and inequality constraints : ; ; H(X) For includes q The equality constraints of each function are responsible for controlling the spindle box to maintain the balance of forces and moments in all degrees of freedom under the action of the 14 hydrostatic guide plates. This is an important constraint condition for calculating the current pose, and the matrix form of this constraint is as follows: H(X) ; in, and They are the first i Block guide plates provide support for hydrostatic guide systems. X and Y Support force in the axial direction, ,、 and They are the first i Block guide plates provide winding for hydrostatic guide rail systems X,Y,Z The torque on the shaft, , For hydrostatic guide rail systems X and Y External load force in the axial direction, , , For hydrostatic guide rail systems in winding X, Y,Z The external load torque on the shaft.

[0031] For includes p Inequality constraints from the equations are used to prevent positioning and orientation deviations caused by dry friction. Based on the number of hydrostatic guide plates, the points on the outer contour of each hydrostatic guide plate furthest from the geometric center of the hydrostatic guide system—i.e., the points most prone to dry friction—are designated as constraint points. i The new coordinates of the constraint points on the block hydrostatic guide plate after the attitude positioning deviation of the entire system are: : ; in, p= 14, For the initial installation i The coordinates of the point on the outer contour of the block hydrostatic guide plate that is farthest from the geometric center of the hydrostatic guide system. In the i The scalar component in the normal direction of the block hydrostatic guide plate is .

[0032] To prevent the i Dry friction occurs at the first constraint point, requiring the definition of the matching first constraint point. i There are 1 reference point with coordinates of 1 ,coordinate Indicates the first i The coordinates of the point on the outer contour of the block hydrostatic guide plate that is farthest from the geometric center of the hydrostatic guide system, when it is about to experience dry friction with the hydrostatic guide section. In the i The scalar component in the normal direction of the block hydrostatic guide plate is .

[0033] Therefore, the first i Constraint conditions at each constraint point , i =1…14.

[0034] As a fluid system, hydrostatic guide rails are subject to mesh constraints in simulation calculations using commonly used numerical and finite element methods. Achieving high accuracy requires precise mesh generation, which significantly increases computation time. Therefore, the function... K static , Kmoment , Error pose , All rely on a high fidelity (HF) numerical model. Y HF In addition, this embodiment also constructs a computational logic AND... Y HF The same low-fidelity (LF) model with a lower mesh density Y LF .

[0035] This embodiment provides a multi-fidelity proxy-assisted hybrid optimization method that optimizes any design variable within the domain with the fewest high-fidelity calculations. X The response is predicted, and the optimal solution is sought.

[0036] S2. Construct a multi-fidelity database: Build a database to store evaluation data from multiple fidelity levels; the first is a high-fidelity database. Used to record the response of high-fidelity numerical models And the corresponding design variables, denoted as ,therefore The form is The second is a low-fidelity database. Used to record the response of low-fidelity numerical models and the corresponding independent variable ,therefore D LF The form is In subsequent optimization iterations, this module performs dynamic updates according to the instructions of the point addition decision submodule and is responsible for storing the currently known Pareto front solution set.

[0037] S3. Construct a data fusion prediction model: By constructing a co-kriging proxy model based on an autoregressive structure and using all the data from the aforementioned data management module, a corrected predictive response is generated. This prediction is more accurate than the Kriging interpolation model alone. The system is maintained in [number missing]. m+p These are several co-kriging proxy models based on autoregressive structures, respectively corresponding to m One optimization objective and p One inequality constraint.

[0038] The core mathematical formula of the co-kriging surrogate model based on autoregressive structures is a hierarchical autoregressive structure, which integrates... and All data in the dataset are used to express the high-fidelity numerical model response as the sum of the low-fidelity numerical model response multiplied by a scalar scaling factor and the deviation of an independent Gaussian process: ; in, It involves modeling a low-fidelity numerical model using a Kriging interpolation model. It is a scalar scaling factor used to characterize the correlation between the responses of high-fidelity and low-fidelity models, while It is the bias function, which is also modeled as an independent Gaussian process. It is used to compensate for the nonlinear deviation between the high-fidelity model response and the scaled low-fidelity model response, and is an essential part of achieving high-precision fusion. X These are the design variables that are input.

[0039] This embodiment employs an anisotropic Gaussian process covariance structure, where for any two design variables... and Its related function is defined as: ; in, and These represent design variables. and In the k Components in each design variable dimension. Among them, The anisotropic setting allows the model to automatically learn and distinguish the sensitivity differences of different hydrostatic guide plate design variable parameters, such as the oil sealing edge width and oil cavity depth, to the optimization objectives and constraints.

[0040] Referring to the structure of the covariance matrix in the Kriging interpolation model, a joint covariance matrix C is constructed that integrates the correlations between the high-fidelity model and the low-fidelity model: ; ; Where cov{} represents the covariance solution. Design variables for samples in a low-fidelity model. Design variables for samples in a high-fidelity model. Covariance structure of anisotropic Gaussian processes The variance of the low-fidelity process represents the magnitude of fluctuation in the low-fidelity model's response around its mean function. The variance of the bias process represents the high-fidelity model response after deducting the scaled low-fidelity response, i.e., the variance mentioned above. The fluctuation range of the residual deviation.

[0041] Five hyperparameters in the covariance matrix C The maximum likelihood estimation method is used to obtain the hyperparameters, which means finding a set of hyperparameters that... The probability density value of its occurrence reaches its maximum. Among them, This represents the total number of high-fidelity and low-fidelity models in the sample. For hyperparameters The covariance matrix under [the given conditions]. Joint response of high-fidelity and low-fidelity models. It follows a mean of 0 and a covariance matrix of . The probability density function of the Gaussian distribution is: ; Through derivation, the complete form of the approximate log-likelihood function can be obtained as follows: ; After iteration, seeking The required hyperparameters can then be determined.

[0042] Therefore, the predicted mean of the co-kriging proxy model based on the autoregressive structure and variance It can be represented as: ; .

[0043] S4. Generate a population of candidate solutions: S401. First, an evolutionary search framework based on a multi-objective probabilistic model is constructed to generate a population of candidate solutions. The framework first requires the construction of a Pareto front probabilistic model. This Pareto front probabilistic model is based on a current high-fidelity database. Pareto front solution set stored in A Gaussian mixture model (GMM) is used to fit the data, and its probability density function is expressed as: ; Among them, the clustering distribution of Pareto front solutions is determined by observing the Pareto front solution clustering distribution. K The number of Gaussian components, For the first i The weights of each Gaussian component, For the first i The mean vector of Gaussian components, For the first i The covariance matrix of Gaussian components, where Let be the probability density function of a multivariate Gaussian distribution, and its basic form is: ; in, X As a set of design variables, For the first i The mean vector of the design variables, Indicates the first i The covariance matrix of each cluster, n Design variables X Dimensions.

[0044] This probabilistic model clearly depicts the relationship between the Pareto front solution set distribution and the design variables.

[0045] S402. Subsequently, considering design variables... X The sheer size of the dimensionality and the complex coupling between various design variables and the objective function, especially the nonlinear correlation between stiffness and oil cavity size parameters, necessitate the introduction of an uncertainty-guided sampling distribution within the framework to explore unknown regions. This method utilizes the autoregressive structure-based co-kriging surrogate model discussed above for the first... i The prediction variance provided by the optimization objective Construct a sampling distribution that is proportional to the total prediction uncertainty: ; This distribution ensures that the algorithm actively samples regions in the parameter space where the surrogate model makes inaccurate predictions or lacks data, thereby effectively preventing the search process from getting trapped in local optima.

[0046] S403. Finally, construct a cross-entropy fusion processor to fuse data using the cross-entropy method. and And generate a candidate solution population. The specific fusion and generation methods are as follows: First, initialize the sampling distribution. In subsequent iterations, the first t Wheel distribution is A co-kriging surrogate model based on autoregressive structure is used for the first... i The prediction variance provided by the optimization objective Generate a sampling distribution that is proportional to the prediction uncertainty of the current sampling distribution. As the current distribution From the current distribution Multiple design variables in X A co-kriging surrogate model based on an autoregressive structure was used to predict the response and assess the hypervolume index increment it brings to the Pareto front. The model was then selected from among those with the highest hypervolume index increments. as a high-quality candidate solution population .

[0047] The physical meaning of the hypervolume index is the volume of the closed region formed by the non-dominated front and the reference point, and the Pareto front solution set. For reference point The hypervolume index can be expressed as: ; in, This is the value of the excess volume index. The Lebesgue measure is used to measure volume. Represents the Pareto front solution set The number of solutions For the first i The volume of the region dominated by the solutions and the reference point.

[0048] Therefore, the increment of the hypervolume index can be expressed as: ; in, This indicates that the predicted response is incorporated into the hypervolume index after the current Pareto front. .

[0049] Then, update the next distribution. Make it comparable to the population of high-quality candidate solutions of KL Minimize divergence, that is, minimize the difference between two probability distributions. The calculation method is as follows: ; in The probability density function is denoted as continuous distribution p The probability density function is continuous distribution q Between KL Divergence is defined as: ; go through t Round iteration, from the final distribution The sample drawn from the middle constitutes the candidate solution population. .

[0050] S5. Multiple Criterion Point Selection: It includes two stages: constraint filtering and multi-objective selection.

[0051] First, receive the candidate solution population. For candidate solution population Each set of candidate design variables X The satisfaction of the autoregressive co-kriging surrogate model is calculated. p Inequality constraints mean and variance Therefore, candidate solution population A candidate design variable XSatisfying the first i Inequality constraints probability for: ; Therefore, for design variables X ,when p Inequality constraints The joint probability when both conditions are met for: ; Set an acceptable probability value All Design variables X Save as a candidate point set.

[0052] Then, for all design variables in the candidate addition set X The incremental hypervolume index resulting from incorporating the predicted response calculated using a co-kriging surrogate model based on an autoregressive structure into the current Pareto front is calculated. .

[0053] ； Choosing from the candidate addition point set yields the greatest increase in supervolume index. k Design variables X As a set of additions , .

[0054] After completing the above set of added points, use a high-fidelity numerical model to calculate the design variables for each group in the set of added points. response , ,Will Add to high-fidelity database Therefore, the updated .

[0055] S6. Iterative Optimization: Repeat steps S3 to S6 until the improvement amount of the optimization target is lower than the preset accuracy requirement or the number of iterations is reached, and finally from The optimal solution is extracted from the final Pareto front stored in the database.

[0056] Example 2 This embodiment provides a supplementary explanation of the performance optimization method for the hydrostatic guideway system of heavy machine tools, based on specific applications and data, as follows: This embodiment aims to optimize the performance of the Y-axis hydrostatic guideway system of a heavy-duty floor-type milling and boring machine. Figure 2 For this heavy-duty floor-type milling and boring machine YThe basic structure of the shaft consists of a column and a spindle box. Fourteen hydrostatic guide plates are installed between the column and the spindle box for load-bearing and constraint in two horizontal directions. These 14 guide plates are arranged in two layers, with seven plates in each layer. One of the guide plates is positioned... It can be changed, meaning the morphology of the hydrostatic oil film can be actively controlled. The optimization objective of this system is to minimize the static stiffness in both translational directions. Minimize the overturning stiffness when rotating about three axes. Minimize the five-dimensional attitude positioning deviation, excluding the main feed motion direction (i.e., two translational degrees of freedom and three rotational degrees of freedom). ) Optimization Objective .

[0057] The equilibrium equations, which maintain the balance of forces and moments across all degrees of freedom using the spindle box under the force of 14 hydrostatic guide plates, serve as equality constraints. .

[0058] H(X) ; The point on the outer contour of the 14 hydrostatic guide plates that is furthest from the geometric center of the hydrostatic guide system, i.e. the point most prone to dry friction, is deemed not to experience dry friction, which is used as an inequality constraint. .

[0059] ; S100, Construct a high-fidelity numerical model: Responsible for calculating the performance of the hydrostatic guide rail, namely stiffness and position error.

[0060] In this example, for Figure 2 Any hydrostatic guide plate in the system, all guide plates adopt constant pressure oil supply, and their oil film pressure distribution is described by the Reynolds equation for incompressible fluids: ; in, p Indicates pressure , h Indicates oil film thickness , Indicates the dynamic viscosity of oil , U Indicates fluid along x Flow velocity in direction , V Indicates fluid along y Flow velocity in direction .

[0061] Among them, film thickness h The distribution satisfies ,in Indicates the thickness of the center oil film , Indicates the depth of the oil cavity , Indicates the hydrostatic guide plate winding x Inclination angle of the axis , Indicates the hydrostatic guide plate winding y Inclination angle of the axis .

[0062] The Reynolds equations are discretized using the finite difference method (FDM), and the discretization points along the path are discretized on the hydrostatic guide plate surface. X direction m points, along Y direction n There are a total of [number] points. These points determine the point where the oil pressure is released (i.e., the pressure is 0 MPa) and the boundary condition where the pressure at the oil supply port equals the supply pressure. Therefore, the oil film pressure... p exist The partial derivative at point can be expressed in difference form as , where i From 1 to m integers, j From 1 to n Integers: ; ; Therefore, this The first and second partial derivatives of the Reynolds equation at each point can be expressed in terms of the surrounding points, forming a total of A system of equations. Solving the system of equations using boundary conditions yields the calculated pressure distribution for each equation. Then, the pressure is integrated over the area of ​​each plate to obtain a vector form of the load-bearing capacity of that plate. The remaining 13 plates are calculated using the same method. See [link to calculation]. Figure 3a and Figure 3b . Through the force and moment balance equations, i.e., equality constraints The equilibrium position of the system is determined iteratively, thus forming the current design variables. X In the response .

[0063] A small variable is applied in turn to each of the five responses in the above positioning pose deviation response. That is, the following five operating conditions of hydrostatic guide rail systems. : ; ; ; ; ; These 5 hydrostatic guide rail systems are operating under the following conditions. All Extract the vector. The first element , The second element , The third element , The fourth element , The fifth element .

[0064] Therefore, design variables X The remaining responses in the response are: ; Therefore, the optimization objective It can be represented as: ; The core of this example lies in using the hybrid optimization engine described in the invention description section to efficiently process optimizations. Functions, whose procedures strictly follow Figure 4 The flowchart shown is executed.

[0065] S200, Initialization Phase: Define the optimization problem to be performed on this set of hydrostatic guide rail systems. In this example, design the variable vector. X The total dimensions are 126. The calculation method and optimization objective function are shown in S100.

[0066] In this implementation case, grid density is used to differentiate the models. The high-fidelity model uses 100,000+ grids, while the low-fidelity model uses around 5,000 grids.

[0067] To construct the initial sample set, the Latin hypercube sampling (LHS) method was used to generate sample points in a 126-dimensional parameter space. To ensure the initial quality of the surrogate model, [the following steps were taken]: A low-fidelity sample of design variables and The high-fidelity design variable samples are mostly distributed under common design parameters or initial design parameters in implementation case equipment. The low-fidelity sample points are widely distributed over a wider range that can meet the conditions, including some design variables under boundary conditions.

[0068] Using a high-fidelity numerical model simulation module Simulate and calculate the response of 100 samples, and then... Stored in a high-fidelity database ; Use high-fidelity numerical model simulation module Simulate the response of 100 samples and compare the data. Stored in a high-fidelity database .in These 100 high-performance sample points will be used to initialize the first-generation Pareto front solution set. This serves as the initial solution set for subsequent iterative optimization.

[0069] S300. Construct a co-kriging agent model based on an autoregressive structure, utilizing... and Based on all the data, construct 10 optimization objective functions. and 14 inequality constraints A co-Kriging surrogate model based on an autoregressive structure is proposed. This surrogate model represents the high-fidelity numerical model response as the sum of the low-fidelity numerical model response multiplied by a scalar scaling factor and the bias of an independent Gaussian process: ; in, It involves modeling a low-fidelity numerical model using a Kriging interpolation model. It is a scalar scaling factor used to characterize the correlation between the responses of high-fidelity and low-fidelity models, while It is a deviation function.

[0070] The covariance matrix C is constructed based on the structure of the covariance matrix in the Kriging interpolation model, and the five hyperparameters in the covariance matrix C are obtained using the maximum likelihood estimation method. And the predicted mean of the co-kriging surrogate model based on the autoregressive structure was calculated. and variance .

[0071] After training the co-kriging proxy model based on an autoregressive structure, for new design variables not included in the sample set... X Each proxy model can provide not only the predicted mean of the target. It is used in the S330 as a fast performance estimate and can also provide prediction variance. This is used as a quantitative indicator of predictive uncertainty. These 24 models... and It forms the basis for all subsequent decisions.

[0072] S310. A population of potential candidate guide rail designs is generated through the constructed multi-objective probabilistic model. First, using... These 100 high-performance sample points are used to initialize the first-generation Pareto front solution set. A Pareto front probability model was constructed. These solutions represent a series of guide rail designs that achieve optimal trade-offs among 10 objective functions: two maximizing translational stiffness, three maximizing overturning stiffness, and five minimizing positioning and pose deviations. The distribution of these high-performance designs was fitted using the Gaussian mixture model (GMM) mentioned in the invention, where the number of mixture components is... K Set to 5. This model can learn the combination rules and correlations of key design variables such as sealing edge width and oil cavity depth in high-performance design.

[0073] S320. Subsequently, in order to explore the region of design variable combinations that were not fully validated beyond the initial 100 high-fidelity design variable samples, the system invoked a co-kriging surrogate model based on an autoregressive structure to calculate the prediction variance of all points in the entire 126-dimensional design space. By predicting the parameter combinations corresponding to regions with high variance, especially the positions of the two adjustable plates and the design parameters of the corresponding hydrostatic guide plates providing the opposite direction, an exploratory distribution is constructed. This distribution is proportional to the product of the variances of all target predictions. It has high values ​​in the design variable regions where high-fidelity design variable samples are sparse, providing a wider set of design variables for the whole algorithm and reducing the limitations of subsequent candidate sets.

[0074] S330. After initializing the sampling distribution, the goal is to merge the two distribution types mentioned above. Through cross-entropy adaptive iterative sampling, the goal is to generate 100 candidate guide rail design schemes from the current distribution.

[0075] First of all as the initial sampling distribution The predicted mean was obtained using a co-kriging surrogate model based on an autoregressive structure. We quickly predict the 10 optimization objective functions for each of the 100 schemes and calculate the hypervolume increment for each scheme. We then select the top 10% (10 schemes) with the highest hypervolume increments and label them as "elite samples." These samples maximize overall performance in the prediction. Based on these 10 elite samples, we update the sampling distribution. This minimizes the KL divergence between the distribution of this sample and the distribution of elite samples, thus making it more likely to produce high-performance designs similar to elite samples.

[0076] S340. After 5 iterations of cross-entropy, 100 candidate design variable schemes are extracted from the final distribution to form the candidate population for this round. .

[0077] S400. Next, we will make a decision on adding points. First, we will perform constraint filtering on the candidate population generated in S340. The 100 candidate design variable schemes were sent to the constraint filter.

[0078] Using 14 inequality constraints The co-kriging surrogate model based on autoregressive structure calculates the effects of 100 candidate design variable schemes on 14 inequality constraints. The probability of simultaneously satisfying : ; After that, any Below the preset threshold, in this case Guide rail designs with a friction factor below the design threshold of 0.95 were discarded because they were predicted to be highly likely to experience dry friction. After eliminating these designs, the remaining designs were compiled into a feasible candidate set. .

[0079] Then, for the feasible candidate set Each design variable in X Each was incorporated into a high-fidelity database. The study uses a co-kriging surrogate model based on an autoregressive structure to calculate the resulting hypervolume index increment, and evaluates its potential for use in high-fidelity databases. Current Pareto front solution set The impact.

[0080] Finally, determine the skill points to allocate and select... The 5 guide rail design schemes with the largest increase in volume index in the Chinese Super League As a set of dots.

[0081] S410. The following is a high-fidelity evaluation of the design scheme to ensure its effectiveness. The resulting performance parameters are accurate, and this design scheme... The data is sent to a high-fidelity numerical model simulation module for a complete hydrostatic guide rail performance simulation to calculate its accurate response. This process ensures that each accurate model call and calculation yields the greatest information benefit or optimizes the performance of the target hydrostatic guide rail.

[0082] S420. Update the current database and extract the 10 responses calculated by the high-fidelity numerical model simulation module in S410. ,get Will Add to high-fidelity database Therefore, the updated And update the Pareto frontier solution set. .

[0083] S430, The termination definition of the iterative steps is required, defining whether the system has converged. First, check... Does the total number of updates meet the budget setting? After the update, if the increase in the hypervolume index of the Pareto front brought about by the newly added response is greater than 1%, then... Return to S300 and use the updated version. Retrain all autoregressive co-kriging agent models and begin a new cycle of "candidate generation - point addition decision - validation update". The continuous updating of the algorithm is key to its continuous performance improvement and accurate optimization results. Iteration stops when the budget setting is met. Furthermore, to minimize the computation cycle and prevent overcomputation, auxiliary judgment conditions are added; iteration also stops after five consecutive iterations.

[0084] S500 outputs results after system optimization, but the output is not a single design parameter solution, but rather... The final Pareto front is stored in the database. This can provide designers with a set of optimal trade-offs to achieve the highest load-bearing stiffness under the machining conditions that the machine tool often experiences, or to provide a higher positioning pose for decision-making.

[0085] Example 3 Based on the same inventive concept, this application also provides a performance optimization system for a hydrostatic guideway system of a heavy machine tool, which can be used to implement the method described in the above embodiments, and specifically includes the following: A multi-objective optimization model module is established to determine the design variables of the hydrostatic guide rail system. X Construct an optimization objective function that includes translational stiffness, overturning stiffness, and positioning posture deviation. F(X) And establish equation constraints based on the balance of force and torque. H (X ) and inequality constraints to prevent dry friction G(X) ; The multi-fidelity database module is used to obtain high-fidelity datasets through high-fidelity numerical model simulations. Low-fidelity datasets are obtained through simulations using low-fidelity numerical models with the same computational logic but lower grid density. ; The data fusion prediction model building module is used to utilize the high-fidelity dataset. With low-fidelity datasets A co-kriging surrogate model based on an autoregressive structure is constructed for handling arbitrary design variables. X The objective function and constraints are used to make predictions, and the predicted mean and predicted variance are output. Candidate solution population generation module, used for high-fidelity datasets The Pareto front solution set in the model is used to construct a probabilistic model. Combined with the sampling distribution constructed from the prediction variance, the cross-entropy method is used to fuse and generate a population of candidate solutions. ; The multi-criteria point selection module is used to select the candidate solution population. Perform constraint filtering to remove constraints that do not satisfy the inequality. G(X) Solutions with probabilities below a threshold; calculate the hypervolume index increment resulting from incorporating the remaining candidate solutions into the current Pareto front, and select several design variables with the largest increments to form an addition set. ; The simulation update module is used to update the point set using the high-fidelity numerical model. Perform simulation calculations to obtain accurate responses and update them to a high-fidelity database. At the same time, update the Pareto front solution set; The iterative optimization module repeats steps S3 to S6 until a preset convergence condition is met. After the update, the growth rate of the hypervolume index at the Pareto front resulting from the newly added response is less than 1%, or After five consecutive iterations, the iteration stops, and the final Pareto optimal solution set is output.

[0086] Preferably, embodiments of this application also provide a specific implementation of an electronic device capable of implementing all steps in the performance optimization method for a heavy-duty machine tool hydrostatic guideway system described in the above embodiments. The electronic device specifically includes the following components: Processor, memory, communications interface, and bus; The processor, memory, and communication interface communicate with each other via a bus; the communication interface is used to realize information transmission between server-side devices, metering devices, and user-side devices.

[0087] The processor is used to call a computer program in memory, and when the processor executes the computer program, it implements all the steps in the performance optimization method for the hydrostatic guideway system of heavy machine tools in the above embodiments.

[0088] Embodiments of this application also provide a computer-readable storage medium capable of implementing all steps of the performance optimization method for a heavy-duty machine tool hydrostatic guideway system described in the above embodiments. The computer-readable storage medium stores a computer program that, when executed by a processor, implements all steps of the performance optimization method for a heavy-duty machine tool hydrostatic guideway system described in the above embodiments.

[0089] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on its differences from other embodiments. In particular, hardware + program embodiments are relatively simple in description because they are fundamentally similar to method embodiments; relevant parts can be referred to the descriptions in the method embodiments.

[0090] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0091] While this application provides method operation steps as shown in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-inventive labor. The order of steps listed in the embodiments is merely one possible execution order among many and does not represent the only execution order. In actual device or client product execution, the method can be executed in the order shown in the embodiments or drawings or in parallel (e.g., in a parallel processor or multi-threaded processing environment).

[0092] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0093] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0094] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0095] This invention is not limited to the embodiments described above. The above description of specific embodiments is intended to illustrate and explain the technical solutions of this invention. The specific embodiments described above are merely illustrative and not restrictive. Without departing from the spirit and scope of the claims, those skilled in the art can make many specific modifications based on the teachings of this invention, and these modifications all fall within the scope of protection of this invention.

Claims

1. A performance optimization method for a hydrostatic guideway system in heavy-duty machine tools, characterized in that, include: S1. Establish a multi-objective optimization model: Determine the design variables of the hydrostatic guide rail system. X Construct an optimization objective function that includes translational stiffness, overturning stiffness, and positioning posture deviation. F(X) And establish equation constraints based on the balance of force and torque. H(X ) and inequality constraints to prevent dry friction G(X) ; S2. Constructing a multi-fidelity database: Obtaining high-fidelity datasets through high-fidelity numerical model simulations. Low-fidelity datasets are obtained through simulations using low-fidelity numerical models with the same computational logic but lower grid density. ; S3. Construct a data fusion prediction model: using the aforementioned high-fidelity dataset With low-fidelity datasets A co-kriging surrogate model based on an autoregressive structure is constructed for handling arbitrary design variables. X The objective function and constraints are used to make predictions, and the predicted mean and predicted variance are output. S4. Generating a candidate solution population: Based on a high-fidelity dataset The Pareto front solution set in the model is used to construct a probabilistic model. Combined with the sampling distribution constructed from the prediction variance, the cross-entropy method is used to fuse and generate a population of candidate solutions. ; S5. Multi-criteria selection: For the candidate solution population... Perform constraint filtering to remove constraints that do not satisfy the inequality. G (X) Solutions with a probability below a threshold; Calculate the hypervolume index increment resulting from incorporating the remaining candidate solutions into the current Pareto front, and select the design variables with the largest increments to form the addition set. ; S6. Simulation Update: Use the high-fidelity numerical model to update the point set. Perform simulation calculations to obtain accurate responses and update them to a high-fidelity database. At the same time, update the Pareto front solution set; S7. Iterative optimization: Repeat steps S3 to S6 until the preset convergence condition or computation budget is met, and output the final Pareto optimal solution set.

2. The performance optimization method for a hydrostatic guideway system of a heavy machine tool according to claim 1, characterized in that, The design variables X It contains i sets of parameters, corresponding to i hydrostatic guide rails; each set of parameters includes: a set of outer contour dimensions of the hydrostatic guide rail. Oil cavity size parameter set Oil cavity depth Oil supply hole diameter and film thickness design value .

3. The performance optimization method for a hydrostatic guideway system of a heavy machine tool according to claim 1, characterized in that, The optimization target vector F(X) Specifically, this includes: maximizing translational stiffness. The reciprocal of the maximization of overturning stiffness The reciprocal of the position and minimizing the positioning pose deviation Wherein, the positioning pose deviation The bearing capacity of each guide rail plate is obtained by solving the Reynolds equation, and then the constraint is based on the aforementioned equation. H(X) =0 is the five-dimensional attitude positioning deviation obtained by iterative solution.

4. The performance optimization method for a hydrostatic guideway system for heavy-duty machine tools according to claim 1, characterized in that, The inequality constraint G(X) The construction method is as follows: Select the points on the outer contour of each hydrostatic guide plate that are farthest from the geometric center of the machine tool hydrostatic guide system as constraint points, and calculate the actual position normal components of each constraint point after the attitude positioning deviation occurs. Normal component of the critical position where dry friction is about to occur The difference must be greater than or equal to 0.

5. The performance optimization method for a hydrostatic guideway system of a heavy machine tool according to claim 1, characterized in that, The formula for the co-kriging surrogate model based on the autoregressive structure in step S3 is as follows: ,in, For high-fidelity numerical model response, Kriging interpolation model for low-fidelity numerical model response. Scalar scaling factor The deviation function is modeled as an independent Gaussian process; the hyperparameters of the covariance matrix C in the co-Kriging proxy model are obtained by the maximum likelihood estimation method.

6. The performance optimization method for a hydrostatic guideway system of a heavy machine tool according to claim 1, characterized in that, The specific steps in step S4 for generating the candidate solution population include: fitting the current Pareto front solution set with a Gaussian mixture model to obtain the Pareto front probability model. Based on the prediction variance provided by the co-kriging agent model, a sampling distribution proportional to the total prediction uncertainty is constructed. Initialize the sampling distribution and iteratively update it using the cross-entropy fusion method; the cross-entropy fusion method specifically involves: combining multiple design variables from the sampling distribution... X The response is predicted using a co-Kriging agent model, and multiple design variables X and their corresponding predicted responses are included in the Pareto front solution set to evaluate these design variables. X And the predicted response as the hypervolume index increment brought by the Pareto front, selecting the hypervolume index increment before Design variables X The corresponding predicted responses are used as the population of high-quality candidate solutions; the sampling distribution is updated to minimize the KL divergence between it and the population of high-quality candidate solutions, and then a new round of iteration is started using the updated sampling distribution; finally, samples are drawn from the converged sampling distribution to form the population of candidate solutions. .

7. The performance optimization method for a hydrostatic guideway system of a heavy machine tool according to claim 1, characterized in that, The specific method for multi-criteria point selection in step S5 is as follows: calculate each design variable in the candidate solution population. X Joint probability satisfying all inequality constraints Design variables with PoF(X) greater than a set threshold are retained; for the retained design variables, the hypervolume index increment caused by their inclusion in the current Pareto front solution set is calculated based on the predicted response of the co-kriging surrogate model. ,choose The k largest design variables are used as the addition point set.

8. A performance optimization system for a hydrostatic guideway system of a heavy-duty machine tool, based on the performance optimization method for a hydrostatic guideway system of a heavy-duty machine tool according to any one of claims 1-7, characterized in that, include: A multi-objective optimization model module is established to determine the design variables of the hydrostatic guide rail system. X Construct an optimization objective function that includes translational stiffness, overturning stiffness, and positioning posture deviation. F(X) And establish equation constraints based on the balance of force and torque. H(X ) and inequality constraints to prevent dry friction G(X) ; The multi-fidelity database module is used to obtain high-fidelity datasets through high-fidelity numerical model simulations. Low-fidelity datasets are obtained through simulations using low-fidelity numerical models with the same computational logic but lower grid density. ; The data fusion prediction model building module is used to utilize the high-fidelity dataset. With low-fidelity datasets A co-kriging surrogate model based on an autoregressive structure is constructed for handling arbitrary design variables. X The objective function and constraints are used to make predictions, and the predicted mean and predicted variance are output. Candidate solution population generation module, used for high-fidelity datasets The Pareto front solution set in the model is used to construct a probabilistic model. Combined with the sampling distribution constructed from the prediction variance, the cross-entropy fusion method is used to generate a population of candidate solutions. ; The multi-criteria point selection module is used to select the candidate solution population. Perform constraint filtering to remove constraints that do not satisfy the inequality. G(X) Solutions with a probability below a threshold; Calculate the hypervolume index increment resulting from incorporating the remaining candidate solutions into the current Pareto front, and select the design variables with the largest increments to form the addition set. ; The simulation update module is used to update the point set using the high-fidelity numerical model. Perform simulation calculations to obtain accurate responses and update them to a high-fidelity database. At the same time, update the Pareto front solution set; The iterative optimization module is used to repeat steps S3 to S6 until the preset convergence condition or calculation budget is met, and output the final Pareto optimal solution set.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the performance optimization method for a hydrostatic guideway system for heavy machine tools as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the performance optimization method for a hydrostatic guideway system for heavy-duty machine tools as described in any one of claims 1 to 7.

Images