A Multidisciplinary Collaborative Design Optimization Method for Connectors Based on Adaptive Agent Model

Abstract

This invention discloses a multidisciplinary collaborative design optimization method for connectors based on an adaptive surrogate model. The method includes constructing an initial Kriging surrogate model, introducing an EI function to upgrade the Kriging surrogate model to an adaptive surrogate model, applying the adaptive surrogate model to multidisciplinary collaborative design optimization, constructing a multidisciplinary collaborative design optimization method based on the adaptive surrogate model, performing uncertainty analysis and design optimization modeling under the coupled state of multi-physics mechanical properties of connectors for special-purpose lifting slewing platforms, and finally performing design optimization. This invention, by combining uncertainty theory, surrogate model theory, and collaborative optimization theory, solves the problem that traditional design methods struggle to qualitatively and quantitatively construct optimization problem performance functions or constraints for complex engineering models, thus limiting the widespread application of collaborative design optimization methods in practical engineering fields. This allows the invention to meet the lightweight target requirements of connectors for special-purpose lifting slewing platforms.

Description

Technical Field

[0001] This invention belongs to the field of multidisciplinary design optimization technology for mechanical products that considers uncertainties, and particularly relates to a multidisciplinary collaborative design optimization method for connectors based on an adaptive surrogate model. Background Technology

[0002] With the rapid development of the manufacturing industry, mechanical equipment and system structures are becoming increasingly complex and diverse. Multidisciplinary design optimization methods, as a methodology for solving complex system optimization problems, can comprehensively evaluate different disciplines during disciplinary analysis. For disciplines with coupling effects, methods such as disciplinary decoupling or collaborative solving can be adopted to eliminate mutual influence to a certain extent. By introducing intelligent optimization algorithms and strategies, the entire system design optimization process can be rationally allocated and arranged, improving the efficiency of design optimization and system performance indicators, shortening development time, and saving costs. Therefore, research on multidisciplinary design optimization methods for mechanical systems is of great significance.

[0003] Collaborative optimization (CO) is a hierarchical optimization method that decomposes engineering tasks into two levels: system-level and discipline-level optimization. The basic idea is to construct a system layer to coordinate inconsistencies in the solutions from different sub-disciplines. Each sub-discipline, when optimizing, can temporarily disregard the influence of other disciplines and only need to satisfy its own constraints. The goal of discipline-level optimization is to minimize the difference between the optimization result of that discipline and the target value provided by the system-level optimization. Inconsistencies between the optimization results of different disciplines are coordinated by the system-level optimization. Through multiple iterations between the system-level and discipline-level optimizations, the results converge to a most accurate value.

[0004] Currently, some studies have combined multidisciplinary design optimization methods and collaborative optimization methods for optimization in many fields.

[0005] For example, Chinese invention patent document CN105205275A, published on December 30, 2015, discloses a multidisciplinary design optimization method for missile and engine integration based on variable correlation. This method includes constructing a quantitative hyperellipsoidal model of variable correlation based on variable intervals, establishing a multidisciplinary collaborative optimization model for missile and engine integration using a collaborative optimization method, and constructing and solving the multidisciplinary design optimization model for missile and engine integration under variable correlation conditions. This invention combines the hyperellipsoidal model with interval theory, solving the problem of inaccurate design results caused by artificially severing the correlation between variables, thus better meeting the actual design requirements of products.

[0006] For example, Chinese invention patent document CN110059415A, published on July 26, 2019, discloses a multidisciplinary design method for high-speed pantographs based on a collaborative optimization algorithm. Based on multidisciplinary design theory, it analyzes the design factors affecting the working performance of high-speed pantographs and classifies them into disciplinary categories. Then, it performs numerical analysis on the design parameters of each discipline, derives the mathematical expressions of the design objectives for each discipline, and establishes corresponding optimization design models. Next, based on the global sensitivity method, it analyzes the coupling relationship and coupling strength of the design variables in the optimization design models of each discipline. According to the coupling strength of the design variables, it determines the system-level and discipline-level design variables, and uses a multidisciplinary collaborative optimization algorithm to solve for the optimized values ​​of the design variables in the optimization models of each discipline. Based on the optimized values ​​of the design variables, it analyzes the optimization results of the pantograph and establishes a three-dimensional solid model of the pantograph. This invention improves the working performance of the pantograph, enhances the current collection quality of the pantograph-catenary system, and provides a new design research approach for high-speed pantographs.

[0007] For example, Chinese patent document CN113239516A, published on August 10, 2021, discloses a multidisciplinary reliability design optimization method for turbine rotors considering uncertainties. This method includes a multi-level, multidisciplinary modeling and design optimization method for complex mechanical systems; a multi-source heterogeneous uncertainty analysis and reliability multidisciplinary design optimization method for complex mechanical systems based on approximate reliability analysis; a multi-source heterogeneous uncertainty analysis and reliability multidisciplinary design optimization method for complex mechanical systems based on numerical simulation reliability analysis; and the introduction of the UBMDO method into the field of turbine rotor design to perform multidisciplinary design optimization of turbine rotors considering uncertainties. This invention combines system hierarchical control strategies, collaborative optimization methods, saddle point approximation methods, and subset simulation uncertainty analysis methods to solve the problem of inaccurate design results caused by the artificial neglect of uncertainty factors in traditional design methods, thereby meeting the high reliability requirements of turbine rotor products.

[0008] The aforementioned published literature demonstrates from different fields that combining multidisciplinary design optimization methods and collaborative optimization methods can solve problems such as inaccuracy, performance, and reliability of the corresponding technologies.

[0009] Compared to connectors in general mechanical equipment, connectors used in specialized lifting platform slewing machines operate in environments characterized by high speed and high load. Their load conditions, operating conditions, environmental conditions, and structural layout are more complex. Damage to the mechanical structure during operation is caused by the continuous accumulation of factors such as constantly changing loads, changing stress distribution due to rotational speed, and continuous degradation of material properties. To achieve the optimization goal of structural lightweighting, it is necessary to analyze multiple disciplines affecting the performance of connectors in specialized lifting platform slewing machines, ensuring that the lightweighting goal is achieved without compromising performance. Furthermore, when the disciplines involved are difficult to quantitatively describe mathematically, the nonlinearity of the optimization problem increases, leading to biased results. Therefore, a multidisciplinary collaborative design optimization method using an adaptive surrogate model needs to be considered to solve the lightweighting problem of connectors in specialized lifting platform slewing machines. Summary of the Invention

[0010] To address the problems of low computational efficiency caused by the inability of traditional design optimization methods to perform parallel design across multiple disciplines, and excessive theoretical limitations due to the inability to provide quantitative mathematical descriptions for complex engineering problems, this invention proposes a multidisciplinary collaborative design optimization method for connectors based on an adaptive surrogate model. This method combines uncertainty theory, surrogate model theory, and collaborative optimization theory to overcome the limitations of the widespread application of collaborative design optimization methods in practical engineering fields. It is particularly suitable for connectors of special-purpose lifting slewing platforms, achieving the goal of lightweighting these products.

[0011] The technical solution of this invention is:

[0012] A multidisciplinary collaborative design optimization method for connectors based on an adaptive surrogate model includes the following steps:

[0013] A multidisciplinary collaborative design optimization model for the connector of a special-purpose lifting slewing platform is constructed based on an adaptive surrogate model. A design optimization model for the connector is then constructed based on the adaptive surrogate model. Reliability analysis of the coupled mechanical properties of the connector under multi-physics fields is performed, followed by design optimization. The multidisciplinary approach includes at least strength and vibration disciplines.

[0014] The reliability analysis includes:

[0015] (1) Conduct random uncertainty factor analysis on the connecting parts of the special lifting tool slewing platform;

[0016] (2) Make assumptions about the environmental factors of the special lifting tool slewing platform connector under complex working conditions;

[0017] (3) Consider the stress and failure modes of each design point of the special lifting tool slewing platform connector, and incorporate the strength performance and vibration into the design optimization and reliability analysis;

[0018] When constructing the design optimization model of the connecting parts of the special lifting tool slewing platform, it is required to achieve the best lightweight effect of the connecting parts. The influence of inertia on the connecting parts under normal working conditions needs to be considered, and the shear stress is applied vertically to the surface of the connecting parts to simulate the effect of force.

[0019] The specific steps in the above process are as follows:

[0020] S1. The construction of the adaptive surrogate model is as follows: First, an initial Kriging surrogate model related to the performance function and constraints of the optimization model is constructed. Then, the EI (exponential integral) function prediction point method and the self-loop optimization strategy are introduced into the initial Kriging surrogate model to obtain an adaptive surrogate model that can iteratively optimize the prediction point.

[0021] The specific process of constructing the initial Kriging proxy model is as follows: Based on the data near the design variable sample points, the information of other unknown variables is fitted by weighted combination of the data near the design variable sample points. A parametric polynomial responsible for the linear regression analysis and a nonparametric polynomial responsible for the random distribution of the model are established, expressed as:

[0022] (1)

[0023] In formula (1): It is a response function; It is a zero-order, first-order, or second-order polynomial regression model; polynomial coefficients The regression parameters can be solved using the least squares method; It is a polynomial basis functions, polynomials The basis functions can be combined to form a basis function vector; This represents a stochastic process, provides an approximation of local biases in the simulation, and further... It should possess the following statistical characteristics, corresponding in order to mean, variance, and oblique variance:

[0024] (2)

[0025] In formula (2): and Represents random sample points; Represents a parameter-based The correlation function can be used to describe the specific relationship between different sample points, and this specific relationship is also related to the distance between two sample points; furthermore, the... It can be represented as:

[0026] (3)

[0027] In formula (3): ; m Indicates the number of design variables; and Represents sample points and The k Each component.

[0028] Using the Gaussian function As a related function of the model, equation (3) can be transformed into the following form:

[0029] (4)

[0030] In equation (4), It is a random parameter greater than zero. Different values ​​of will cause changes in the correlation of the related functions. It can be solved by the maximum likelihood estimation method.

[0031] Furthermore, step S1 includes the following sub-steps:

[0032] S11. Using the Genetic Algorithm (GA) as the solver for the system optimization process, the initial Kriging surrogate model is analyzed and solved to obtain the predicted point N1, and the EI function expression is defined as follows:

[0033] (5)

[0034] In equation (5): Indicates input design variables Simulation results; The CDF (cumulative distribution function) represents the standard normal distribution. PDF (probability density function) representing the standard normal distribution; and This indicates the initial Kriging model at the sample points. The corresponding predicted values ​​and predicted standard deviations are given, and their specific values ​​can be calculated using formulas (6) and (7), respectively:

[0035] (6)

[0036] (7)

[0037] In equations (6) and (7): The dimension is n , unit column vector; Indicates based on parameters Correlation parameters; The predicted value representing the standard deviation. Represents a polynomial parameter vector; This represents the correlation vector between design variables and sample points. Indicates transpose; where, It can be expressed by equation (8):

[0038] (8)

[0039] S12. Substitute the predicted point N1 into the EI function expression, find its maximum value through the monotonicity of the EI function, and use it as the new predicted point N2 to replace the original predicted point. Repeat the iterative steps of solving the replacement through the self-loop optimization strategy until no new predicted point is generated or the value of the new predicted point changes within a small interval.

[0040] S13. Then, we obtain the adaptive proxy model based on the EI function and the self-loop optimization strategy.

[0041] S2. Apply the adaptive surrogate model obtained in step S1 to the collaborative design optimization method to construct a multidisciplinary collaborative design optimization model with approximate performance functions at the system level and approximate constraints at the discipline level. The specific steps are as follows:

[0042] S21. Considering various uncertainties and treating them as design variables, construct optimization mathematical models for the system layer and discipline layer of the collaborative optimization method, expressed as follows:

[0043] (9)

[0044] (10)

[0045] In equations (9) and (10): Indicates the first i The first in each sub-discipline j One target design variable; Indicates the first j One target design variable; This represents the system-level optimization objective performance function; This indicates that after subject-level and sub-subject analysis The optimization result * represents subject-level data information; Indicates the first i Constraints of each sub-discipline; The system-level consistency constraints and the subject-level objective function are represented as shown in equation (11):

[0046] (11)

[0047] S22. Introducing the adaptive surrogate model into the system layer of the collaborative optimization method yields the mathematical optimization model of the system layer of the collaborative optimization method, expressed as:

[0048] (12)

[0049] In equation (12), The approximate fitting objective function is expressed as:

[0050] (13)

[0051] In equation (13): Represents the set of polynomial vectors at the system layer; Represents the polynomial parameter matrix of the system layer; A vector representing the correlation between the sample set and the unknown design variables; Represents the response matrix; Represents the regression multinomial matrix;

[0052] Furthermore, the aforementioned This can be interpreted as equation (14):

[0053] (14)

[0054] In equation (14): Represents several polynomial vectors; Represents a vector of parameters for a polynomial; This represents a correlation vector between several sample sets and unknown design variables. A method for calculating the correlation vector between several sample sets and unknown design variables. Represents a polynomial vector of several system layers. Represents several fitted target response functions;

[0055] S23. Introducing the adaptive surrogate model into the subject layer of the collaborative optimization method yields the subject-layer mathematical optimization model of the collaborative optimization method, expressed as:

[0056] (14)

[0057] In equation (14), Indicates the first i The fitting constraints for each sub-discipline are expressed as follows:

[0058] (15)

[0059] In equation (15): A set of polynomial vectors representing sub-disciplines; Represents the polynomial parameters of a sub-discipline; A vector representing the correlation between unknown design variables and the sample set; The matrix representing the responses of the sample points; Representing vectors The matrix.

[0060] The beneficial effects of this invention are as follows:

[0061] This invention constructs a multidisciplinary collaborative design optimization model; introduces the EI function and self-looping optimization strategy to build an adaptive surrogate model based on the EI function and self-looping optimization strategy; applies the adaptive surrogate model to the collaborative design optimization method, constructing collaborative optimization system-level and discipline-level design optimization models based on the adaptive surrogate model; applies the collaborative design optimization method based on the adaptive surrogate model to the connector of a special lifting slewing platform, constructs a multidisciplinary collaborative design optimization model for the connector based on the adaptive surrogate model, conducts reliability analysis of the mechanical performance coupling of the connector under multi-physics fields, and finally performs design optimization; by combining uncertainty theory, surrogate model theory, and collaborative optimization theory, this invention solves the problem that traditional design methods are difficult to qualitatively and quantitatively construct optimization problem performance functions or constraints for complex engineering models, which restricts the widespread application of collaborative design optimization methods in practical engineering fields, thereby meeting the lightweight target requirements of special lifting slewing platform connector products. Attached Figure Description

[0062] Figure 1 This is a flowchart illustrating the present invention.

[0063] Figure 2 This is a roadmap for uncertainty analysis and design optimization modeling of the multi-physics field mechanical performance coupling state of the connecting parts of the special lifting device rotary platform of this invention. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0065] like Figure 1 As shown, the basic process steps of this invention are as follows:

[0066] S1. Construct an initial Kriging proxy model related to the performance function and constraints of the optimization model;

[0067] S2. The method of solving the prediction point using the EI function and the self-looping optimization strategy are introduced into the initial Kriging surrogate model to obtain an adaptive surrogate model that can iteratively optimize the prediction point.

[0068] S3. Apply the adaptive surrogate model based on EI function and self-loop optimization strategy constructed in step S2 to the collaborative design optimization method to construct a multi-disciplinary collaborative design optimization model with approximate performance function of collaborative optimization system layer and approximate constraint conditions of collaborative optimization discipline layer.

[0069] S4. Apply the multidisciplinary collaborative design optimization model based on the adaptive surrogate model constructed in step S3 to the connector of the special lifting tool slewing platform, construct the design optimization model of the connector based on the adaptive surrogate model, conduct reliability analysis of the mechanical performance coupling of the connector under multiple physics fields, and finally optimize the design.

[0070] In this embodiment, the multidisciplinary field includes at least strength discipline and vibration discipline.

[0071] In step S1, an initial Kriging surrogate model related to the performance function and constraints of the optimization model is constructed. Specifically, based on data near the design variable sample points, information about other unknown variables is fitted by a weighted combination of the data near the design variable sample points. A parametric polynomial responsible for the linear regression analysis and a nonparametric polynomial responsible for the random distribution of the model are established, expressed as follows:

[0072] (1)

[0073] In formula (1): It is a response function; It is a zero-order, first-order, or second-order polynomial regression model; polynomial coefficients The regression parameters can be solved using the least squares method; It is a polynomial The basis functions, and polynomial basis functions can be combined to form a basis function vector; It represents a stochastic process and provides an approximation of local deviations in the simulation.

[0074] Furthermore, the aforementioned It should possess the following statistical characteristics, corresponding in order to mean, variance, and oblique variance:

[0075] (2)

[0076] In formula (2): and Represents random sample points; Represents a parameter-based The correlation function can be used to describe the unique relationship between different sample points, and this relationship is also related to the distance between two sample points.

[0077] Furthermore, the aforementioned It can also be expressed in the following form:

[0078] (3)

[0079] In formula (3): ; m Indicates the number of design variables; and Represents sample points and The k Each component.

[0080] Using the Gaussian function As a related function of the model, equation (3) can be transformed into the following form:

[0081] (4)

[0082] In equation (4), It is a random parameter greater than zero. Different values ​​of will cause changes in the correlation of the related functions. It can be solved by the maximum likelihood estimation method.

[0083] The specific steps of step S2 are as follows:

[0084] S21. Using the genetic algorithm as the solver for the system optimization process, the initial Kriging surrogate model is analyzed and solved to obtain the predicted point N1, and the EI function expression is defined as follows:

[0085] (5)

[0086] In equation (5): Indicates input design variables Simulation results; The CDF represents the standard normal distribution; PDF representing the standard normal distribution; and This indicates the initial Kriging model at the sample points. The corresponding predicted value and prediction standard deviation are given.

[0087] The and The specific values ​​can be calculated using formulas (6) and (7) respectively:

[0088] (6)

[0089] (7)

[0090] In equations (6) and (7): Indicates dimension as n , unit column vector; Indicates based on parameters The correlation parameters; The predicted value representing the standard deviation. Represents a polynomial parameter vector; This represents the correlation vector between design variables and sample points. Indicates transpose; where, It can be expressed by equation (8):

[0091] (8)

[0092] S22. Substitute the predicted point N1 into the expression of the EI function, find its maximum value through the monotonicity of the EI function, and replace the original predicted point with it as the new predicted point N2. Repeat the iterative steps of the replacement through the self-loop optimization strategy until no new predicted point is generated or the value of the new predicted point changes within a small interval.

[0093] S23. Combining step S1, we obtain an adaptive proxy model based on the EI function and a self-loop optimization strategy.

[0094] The specific steps of step S3 are as follows:

[0095] S31. Considering various uncertainties and treating them as design variables, construct optimization mathematical models for the system layer and discipline layer of the collaborative optimization method, expressed as follows:

[0096] (9)

[0097] (10)

[0098] In equations (9) and (10): Indicates the first i The first in each sub-discipline j One target design variable; Indicates the first j One target design variable; This represents the system-level optimization objective performance function; This indicates that after subject-level sub-subject analysis The optimization result * represents subject-level data information; Indicates the first i Constraints of each sub-discipline; The system-level consistency constraints and the subject-level objective function are represented as shown in equation (11):

[0099] (11)

[0100] S32. Introducing the adaptive surrogate model into the system layer of the collaborative optimization method yields the mathematical optimization model of the system layer of the collaborative optimization method, expressed as:

[0101] (12)

[0102] In equation (12), The approximate fitting objective function is expressed as:

[0103] (13)

[0104] In equation (13): Represents the set of polynomial vectors at the system layer; Represents the polynomial parameter matrix of the system layer; A vector representing the correlation between the sample set and the unknown design variables; Represents the response matrix; Represents the regression multinomial matrix;

[0105] Furthermore, the aforementioned This can be interpreted as equation (14):

[0106] (14)

[0107] In equation (14): Represents several polynomial vectors; Represents a vector of parameters for a polynomial; This represents a correlation vector between several sample sets and unknown design variables. A method for calculating the correlation vector between several sample sets and unknown design variables. Represents a polynomial vector of several system layers. Represents several fitted target response functions;

[0108] S33. Introducing the adaptive surrogate model into the subject layer of the collaborative optimization method yields the subject-layer mathematical optimization model of the collaborative optimization method, expressed as:

[0109] (14)

[0110] In equation (14), Indicates the first iThe fitting constraints for each sub-discipline are expressed as follows:

[0111] (15)

[0112] In equation (15): A set of polynomial vectors representing sub-disciplines; Represents the polynomial parameters of a sub-discipline; A vector representing the correlation between unknown design variables and the sample set; The matrix representing the responses of the sample points; Representing vectors The matrix.

[0113] The specific steps of step S4 are as follows:

[0114] S41. Conduct random uncertainty factor analysis on the connecting parts of the special lifting tool rotary platform;

[0115] S42. Make assumptions about the environmental factors of the special lifting tool rotary platform connector under complex working conditions;

[0116] S43. Considering the stress and failure modes at each design point of the special lifting tool slewing platform connector, strength performance and vibration conditions are incorporated into the design optimization and reliability analysis. The influence of inertia on the special lifting tool slewing platform connector under normal operating conditions is considered, and shear stress is applied vertically to the connector surface to simulate the effect of force. Combining steps S1, S2, and S3, a design optimization model for the special lifting tool slewing platform connector is constructed to achieve the best lightweight effect. A reliability analysis of the mechanical performance coupling under multi-physics fields is performed, and finally, design optimization is carried out.

[0117] To enable those skilled in the art to more clearly understand the above-described method for constructing a multidisciplinary collaborative design optimization of the connector of a special lifting slewing platform based on an adaptive surrogate model, further details are provided below.

[0118] like Figure 2As shown, considering the data characteristics of the design variables of the connecting parts for the special-purpose lifting platform, this invention establishes a parallel multidisciplinary system design optimization framework based on the adaptive surrogate model by employing a collaborative design optimization method—a parallel multidisciplinary optimization strategy—and leveraging the advantages of the adaptive surrogate model in handling complex engineering problems. Simultaneously, it considers the coupling relationship between the lightweight design goal and performance of the connecting parts for the special-purpose lifting platform. During the product design and production stages, the connecting parts undergo a repetitive process based on "design-simulation-improvement-redesign-resimulation" and "production-testing-improvement-reproduction-retesting," which maximizes the achievement of the lightweight design goal while ensuring the performance of the connecting parts. Furthermore, due to the long-term existence of uncertainties in actual operation, this invention uses the saddle point approximation method to uniformly quantify and analyze various uncertainties such as random uncertainty and fuzzy uncertainty that may occur in the application. A reliability analysis model based on the mean second-order moment saddle point approximation method is established, thereby continuously performing reliability analysis during the design optimization process of the connecting parts for the special-purpose lifting platform, effectively ensuring its stability and safety. Ultimately, a system is formed as follows: Figure 2 The diagram illustrates a multidisciplinary collaborative design optimization technique for the connector of a dedicated lifting slewing platform based on an adaptive surrogate model, considering uncertainties.

[0119] Considering the inertia experienced by the connecting parts of the special lifting slewing platform during operation, shear stress is applied to the surface of the connecting parts to simulate the effect of force. Combined with steps S2 and S3, a design optimization model for the special lifting slewing platform connecting parts is constructed to achieve the best lightweight effect. In the embodiment, for each uncertainty constraint, its reliability index... β It should be greater than 0.9.

[0120] This invention combines uncertainty theory, surrogate model theory, and collaborative optimization theory to solve the problem that traditional design methods are unable to qualitatively and quantitatively construct optimization problems for performance functions or constraints of complex engineering models, which restricts the widespread application of collaborative design optimization methods in practical engineering fields. This allows the invention to meet the lightweight target requirements of special lifting tool slewing platform connector products.

Claims

1. A multidisciplinary collaborative design optimization method for connectors based on an adaptive surrogate model, characterized in that, The process includes the following steps: constructing a multidisciplinary collaborative design optimization model for the connector of a special-purpose lifting slewing platform based on an adaptive surrogate model; constructing a design optimization model for the connector of the special-purpose lifting slewing platform based on an adaptive surrogate model; then conducting a reliability analysis of the mechanical performance coupling of the connector under multiple physics fields; and finally, performing design optimization. The multidisciplinary approach includes at least strength science and vibration science. The reliability analysis includes: (1) Conduct random uncertainty factor analysis on the connecting parts of the special lifting tool slewing platform; (2) Make assumptions about the environmental factors of the special lifting tool slewing platform connector under complex working conditions; (3) Consider the stress and failure modes of each design point of the special lifting tool slewing platform connector, and incorporate the strength performance and vibration into the design optimization and reliability analysis; When constructing the design optimization model of the special lifting tool slewing platform connector, it is required to achieve the best lightweight effect of the special lifting tool slewing platform connector. The influence of inertia on the special lifting tool slewing platform connector under normal working conditions needs to be considered, and the shear stress is applied vertically to the surface of the connector to simulate the effect of force. The specific steps in the above process are as follows: S1. The construction of the adaptive surrogate model is as follows: First, an initial Kriging surrogate model is constructed to optimize the model performance function and constraints. Then, the EI function prediction point method and the self-loop optimization strategy are introduced into the initial Kriging surrogate model to obtain an adaptive surrogate model that can iteratively optimize the prediction points. The specific steps are as follows: S11. Using the genetic algorithm as the solver for the system optimization process, the initial Kriging surrogate model is analyzed and solved to obtain the predicted point N1, and the EI function expression is defined as follows: （5） In formula (5): Indicates input design variables Simulation results; The CDF represents the standard normal distribution; PDF representing the standard normal distribution; and This indicates the initial Kriging model at the sample points. The corresponding predicted value and predicted standard deviation are given, and their specific values ​​are calculated using formulas (6) and (7): （6） （7） In equations (6) and (7): The dimension is n , unit column vector; Indicates based on parameters The correlation parameters; The predicted value representing the standard deviation. Represents a polynomial parameter vector; This represents the correlation vector between design variables and sample points. Indicates transpose; wherein, the Equation (8) is expressed as follows: ；（8） S12. Substitute the predicted point N1 into the expression of the EI function, find its maximum value through the monotonicity of the EI function, and use it as the new predicted point N2 to replace the original predicted point. Repeat the iterative steps of solving the replacement through the self-loop optimization strategy until no new predicted point is generated or the value of the new predicted point changes within a small interval. S13. Obtain the adaptive proxy model based on the EI function and the self-loop optimization strategy; S2. Apply the adaptive surrogate model obtained in step S1 to the collaborative design optimization method to construct a multidisciplinary collaborative design optimization model with approximate performance functions at the system level and approximate constraints at the discipline level. The specific steps are as follows: S21. Considering various uncertainties and treating them as design variables, construct optimization mathematical models for the system layer and discipline layer of the collaborative optimization method, expressed as follows: （9） （10） In equations (9) and (10): Indicates the first i The first in each sub-discipline j One target design variable; Indicates the first j One target design variable; This represents the system-level optimization objective performance function; This indicates that after subject-level sub-subject analysis The optimization results are shown in *, which indicates subject-level data information. Indicates the first i Constraints of each sub-discipline; The system-level consistency constraints and the subject-level objective function are represented as shown in equation (11): ；（11） S22. Introducing the adaptive surrogate model into the system layer of the collaborative optimization method yields the mathematical optimization model of the system layer of the collaborative optimization method, expressed as: （12） In equation (12), The approximate fitting objective function is expressed as: （13） In equation (13): Represents the set of polynomial vectors at the system layer; Represents the polynomial parameter matrix of the system layer; This represents the set of correlation vectors between the sample set and the unknown design variables. Represents the response matrix; Represents the regression multinomial matrix; S23. Introducing the adaptive surrogate model into the subject layer of the collaborative optimization method yields the subject-layer mathematical optimization model of the collaborative optimization method, expressed as: （14） In equation (14), Indicates the first i The fitting constraints for each sub-discipline are expressed as follows: （15） In equation (15): A set of polynomial vectors representing sub-disciplines; Represents the polynomial parameters of a sub-discipline; A vector representing the correlation between unknown design variables and the sample set; The matrix representing the responses of the sample points; Representing vectors The matrix.

2. The multidisciplinary collaborative design optimization method for connectors based on an adaptive surrogate model as described in claim 1, characterized in that, The specific process of constructing the initial Kriging surrogate model for optimizing the model performance function and constraints is as follows: Based on the data near the design variable sample points, information of other unknown variables is fitted by weighted combination of the data near the design variable sample points, establishing the parametric polynomial responsible for the linear regression analysis of the model and the nonparametric polynomial responsible for the random distribution of the model, expressed as: （1） In formula (1): It is a response function; It is a zero-order, first-order, or second-order polynomial regression model; polynomial coefficients The regression parameters are obtained by solving for their values ​​using the least squares method. It is a polynomial basis functions, polynomials The basis functions are combined to form a basis function vector; This represents a stochastic process, providing an approximation of local deviations in the simulation; and It should possess the following statistical characteristics: （2） In equation (2), and Represents random sample points; Represents a parameter-based The correlation function is used to describe the specific relationship between different sample points; Represented as: （3） In formula (3): ; m Indicates the number of design variables; and Represents sample points and The k Components; utilizing the Gaussian function. As a related function of the model, equation (3) is transformed into the following form: （4） In equation (4), It is a random parameter greater than zero. Different values ​​of will cause changes in the correlation of the related functions. It can be solved by the maximum likelihood estimation method.

Images