Power high-voltage circuit breaker SOM feature quantification proxy modeling method

Abstract

According to the quantitative proxy modeling method for the SOM characteristics of the electric power high-voltage circuit breaker, the finite element simulation model of the operating mechanism is established in the ABAQUS environment, the simulation model is modified and verified through the displacement curve, and the simulation model with the good goodness of fit is obtained. The method comprises the following steps: firstly, carrying out part quantization on a part by using a feature modeling idea, dividing an infarction graph pixel into a structural pixel and a contour pixel, establishing and solving a constraint relationship of the structural pixel according to a geometrical relationship, calculating the contour pixel by using the obtained structural pixel, and completing part quantization by combining feature parameters. Creating an information model and performing secondary development to complete finite element part quantitative modeling of the SOM, and applying C # development part quantitative information management software to assist in managing modeling information. The motion deviation of the operating mechanism caused by the hinge gap and the influence on a hinge pair are researched and analyzed, assistance is provided for the design of the operating mechanism, and the design risk and hidden danger of the SOM are greatly reduced.

Description

Technical Field

[0001] The present application relates to a method for SOM proxy modeling of a power circuit breaker, and in particular to a method for SOM feature quantification proxy modeling of a power high-voltage circuit breaker, and belongs to the technical field of SOM modeling of circuit breakers. Background Art

[0002] High-voltage circuit breakers are one of the most important and complex equipment in the power grid system. The nature of the circuit breaker determines that it must withstand high-speed impact loads and ensure high reliability. The operating mechanism is an important component of the circuit breaker. The spring operating mechanism (SOM) uses the spring as the power element and drives the static and moving contacts through the transmission mechanism to complete the opening and closing. The good motion characteristics of the transmission mechanism are the basis for ensuring the performance and reliability of the mechanical system. Due to the limitations of mechanical design and processing and manufacturing, the kinematic pairs between the mechanisms cannot achieve ideal connection. The existence of the kinematic pair gap causes the mechanism movement to deviate from the ideal design value, and at the same time causes friction and wear of the kinematic pair itself, which in turn promotes the destruction of the joint surface and causes a larger gap. Traditional physical test methods are costly and easy to introduce random errors. With the development of computer technology, the computing power of simulation technology has been enhanced, and it has been used more and more widely in actual engineering. However, the pre- and post-processing of finite element simulation modeling takes up a lot of time in the entire process, which greatly increases the cost of variable parameters or optimization analysis and reduces the analysis efficiency.

[0003] SOM often uses connecting rod mechanism to transmit power and motion, and a large number of components are connected by hinges. Hinge will introduce hinge pins and sleeves, and inevitably introduce motion joint surfaces. The existence of joint surfaces makes the mechanical structure or system no longer continuous, resulting in the complexity of solving kinematic and dynamic problems. Due to the limitations of actual processing conditions, the requirements of engineering assembly, the inadequacy of design methods, and the existence of friction and wear problems during use, articulated connections will inevitably have gaps. The impact of the gap makes the impact problem in the connecting components more prominent, aggravates the friction and wear of the contact surface, and causes the morphology of the joint surface of the mechanical structure to change. The joint surface is closely related to the static characteristics, vibration and vibration control, and dynamic characteristics of the mechanical structure. At the same time, SOM is limited by design requirements and has serious high-speed impact. Therefore, the study of SOM articulation gap and wear is of great value.

[0004] The problems that need to be solved by the existing power circuit breaker SOM agent modeling method and the key technical difficulties of this application include:

[0005] (1) The high-voltage circuit breaker SOM uses a spring as a power element, and drives the static and moving contacts through a transmission mechanism to complete the opening and closing. The good motion characteristics of the transmission mechanism are the basis for ensuring the performance and reliability of the mechanical system. However, due to the limitations of mechanical design and processing and manufacturing, the kinematic pairs between the mechanisms cannot achieve ideal connection. The existence of the kinematic pair gap causes the mechanism movement to deviate from the ideal design value, and at the same time causes friction and wear of the kinematic pair itself, which in turn promotes the destruction of the joint surface and causes a larger gap. The existing physical test method is expensive and easy to introduce random errors. The existing technology uses computers for simulation technology. The calculation ability is short, and the pre- and post-processing of finite element simulation modeling takes up a lot of time in the whole process, which greatly increases the cost of variable parameters or optimization analysis, reduces the analysis efficiency, and makes it impossible to establish a finite element simulation model of the power high-voltage circuit breaker SOM, and it is impossible to use test data to modify and verify the model. It is impossible to achieve feature-based component quantitative modeling, and it is impossible to establish a finite element component quantitative model of the object. There is a lack of development parameter information management auxiliary software, and it is impossible to use the proxy model to establish its simulation approximate model, and it is impossible to analyze the influence of the articulation gap on the articulated pair and the mechanism, and it is impossible to provide feedback for the mechanism design.

[0006] (2) SOM uses a connecting rod mechanism to transmit power and motion. A large number of components are connected by hinges. The hinge will introduce hinge pins and sleeves, and inevitably introduce motion joint surfaces. The existence of the joint surface makes the mechanical structure or system no longer continuous, resulting in the complexity of solving kinematic and dynamic problems. Due to the limitations of actual processing conditions, the requirements of engineering assembly, the inadequacy of design methods, and the existence of friction and wear problems during use, the hinge connection will inevitably have gaps. The impact of the gap makes the impact problem in the connecting component more prominent, aggravates the friction and wear of the contact surface, and causes the morphology of the mechanical structure joint surface to change. The joint surface is closely related to the static characteristics, vibration and vibration control, and dynamic characteristics of the mechanical structure. At the same time, SOM is restricted by design requirements and has serious high-speed impact. However, the existing technology lacks research on the SOM articulation gap and its wear, lacks an efficient and reliable power circuit breaker SOM proxy modeling method, and cannot create a proxy modeling program based on the spatial optimal model to study and analyze the articulation wear of the operating mechanism, which is not conducive to the impact load and reliability research of the power high-voltage circuit breaker SOM, and brings hidden dangers to the safe operation of the power grid system.

[0007] (3) The prior art lacks the analysis of the structural composition and motion principle of the SOM of the power high-voltage circuit breaker, does not consider the high-speed impact problem in the working environment, does not reasonably select its material model and contact model, cannot establish the finite element simulation model of the operating mechanism in the ABAQUS environment, does not build a physical test bench to modify and verify the simulation model with the displacement curve, lacks the use of feature modeling ideas to quantify the parts, cannot establish and solve the constraint relationship of the structural elements according to the geometric relationship, cannot establish the vector ring equation and the z-direction specification chain of the mechanism, and complete the quantification of the parts of the assembly with the positioning specification, the length of the connecting rod, and the initial position as variables. There is a lack of C# to develop the parts quantification information management software to assist in the management of modeling information, and does not analyze the motion deviation of the operating mechanism caused by the articulation gap and its influence on the articulation pair. There is a lack of spatial optimal proxy model, and it is impossible to establish a proxy modeling model of the research object. The Archard model is used to analyze the degree of articulation wear as the target, which ultimately results in the failure to provide good help for the design of the operating mechanism, which is not conducive to the research and development, design and application of the circuit breaker SOM. Summary of the invention

[0008] This application analyzes the structural composition and motion analysis of the electric high-voltage circuit breaker SOM, simplifies it reasonably, considers the high-speed impact problem in the working environment, reasonably selects the material model and contact model, establishes a finite element simulation model of the operating mechanism in the ABAQUS environment, and uses the constructed physical test bench to modify and verify the simulation model with the displacement curve, and obtains a simulation model with good fit. Then, the feature modeling idea is used to quantify the parts of the parts, the outline pixels are divided into structural pixels and contour pixels, the constraint relationship of the structural pixels is established and solved according to the geometric relationship, the contour pixels are calculated using the obtained structural pixels, and the quantification of the parts is completed in combination with the feature parameters. An information model is created based on the relevant information of finite element modeling, and the Python language is used for secondary development to complete the finite element component quantitative modeling of SOM, and C# is used to develop component quantitative information management software to assist in the management of modeling information. The study analyzed the motion deviation of the operating mechanism caused by the articulated gap and its impact on the articulated pair. The Archard model was used to analyze the degree of articulated wear and the articulated fit tolerance (gap value and deviation value) as variables, which provided assistance for the design of the operating mechanism, greatly reduced the design risks and hidden dangers of SOM, and was conducive to the large-scale promotion and application of SOM of high-voltage circuit breakers.

[0009] In order to achieve the above technical effects, the technical solutions adopted in this application are as follows:

[0010] The SOM feature quantification agent modeling method for high-voltage power circuit breakers is used to establish a finite element simulation model of the operating mechanism in the ABAQUS environment, and the simulation model is modified and verified by the displacement curve. The parts are quantified based on the feature modeling. The outline pixels are divided into structural pixels and contour pixels. The constraint relationship of the structural pixels is established and solved according to the geometric relationship. The contour pixels are calculated using the obtained structural pixels. The parts are quantified in combination with the feature parameters. The vector ring equation and the z-direction specification chain of the mechanism are established. The parts quantification of the assembly is completed with the positioning specification, the connecting rod length, and the initial position as variables. The information model is created according to the finite element modeling related information, and the Python language is used for secondary development. The parts quantification information management software is developed using C# to assist in the management of modeling information. Based on the developed program, the motion deviation of the operating mechanism caused by the articulation gap and its influence on the articulated pair are studied and analyzed. The genetic space sampling strategy and the expected enhanced sequence addition strategy are used. Based on the spatial optimal agent model, the SOM agent modeling model is established in combination with the parts quantification modeling and the programmed post-processing technology. Based on the Archard model, the degree of articulation wear is taken as the target and the articulation fit tolerance is taken as the variable to analyze the SOM.

[0011] A-Circuit breaker SOM finite element parts quantitative modeling: 1) Establish a feature-based parts quantitative model, establish and solve the constraint relationship for the feature outline construction element, construct the mathematical representation of the contour element, and realize the quantitative construction of the parts geometric solid model; 2) Establish the plane position relationship of the parts assembly, solve the vector ring equation to obtain the position and initial posture of the parts on the two-dimensional projection surface, and combine the z-axis specification chain to complete the parts quantification of the spatial assembly; 3) Create an actuator finite element SOM information model, express the feature information in the form of abstract classes, establish a finite element parts quantitative model of the actuator, realize the reuse of the finite element SOM model, and develop a parts quantitative information management platform;

[0012] B-SOM proxy modeling and analysis of articulated characteristics: 1) Taking the clearance of the articulated pair as the variable, the deformation and relative displacement of the articulated sleeve and the pin are analyzed to obtain the variation law of the mechanism motion displacement deviation with the clearance specification; 2) Based on the spatial optimal proxy model, the genetic space sampling strategy and the expected enhanced point addition criterion are adopted to establish a finite element proxy modeling program in the ABAQUS environment; 3) Using the established proxy modeling program, with the articulated pair fit tolerance as the design variable and the wear condition during the articulated pair motion as the objective function, it is found that the articulated clearance has a greater impact on the wear, and a clearance that is too large or too small will lead to greater wear.

[0013] Preferably, the assembly parts are quantified: when the parts are created, the initial points are set at the origin, and the relative position relationship between the parts is adjusted during assembly. In the assembly module, the relative position and initial motion posture of the parts are adjusted using the feature relationship of the parts;

[0014] The assembly parts are associated through coaxial lines, parallelism, relative distance, and relative rotation angle related constraint behaviors to ensure that under given initial conditions, each component is adjusted to the corresponding posture. The coordinate system in the assembly drawing is the global coordinate system. The coordinate system used by each component in modeling is called the local coordinate system. The position of the local coordinate system in the global coordinate system is calculated. The position of some local coordinate systems in the global coordinate system is relatively fixed and directly represented by the relative position vector. The actual linkage mechanism contains secondary rod groups and higher-level rod groups. The quantification of assembly parts needs to be solved. The operating mechanism contains two four-bar linkages, and their positions are solved;

[0015] The transmission part of the operating mechanism is composed of connecting rods l1, l2, l3, l4, l5, l6, l7 and frames l4, l8. Among them, l1, l2, l3, l4 constitute four-link I, which transmits motion and force from the upper connecting rod to the main shaft, and l5, l6, l7, l8 constitute four-link II, which transmits motion from the main shaft to the large crank arm. The four-link mechanism establishes a closed vector loop.

[0016]

[0017] In the formula, For vector The angle with the x-axis, l i For vector Model;

[0018] When solving, project them onto the coordinate axis respectively and convert them into scalar form equations. Equation 2 is its scalar form:

[0019]

[0020] A and D are external hinge points. After their coordinates are given as initial conditions, Equation 2 contains only the design variables. and The equation is solvable;

[0021] Solve the four-bar linkage I to obtain by The angle θ with the main axis is used as the initial condition to solve the four-bar linkage II, and the plane motion position of the transmission mechanism under the given initial condition is obtained. The specification driving relationship in the plane normal direction is established according to the assembly relationship;

[0022] The operating mechanism includes three phases, each phase includes a pair of main shaft crank arms, lower connecting rods, and large crank arms. The large crank arms and lower connecting rods are sub-assembled separately. The large crank arm sub-assembly parameter is Ddgb1 With D dgb2 , the lower connecting rod assembly parameters are D xlg With D dgb2 , obtained from the assembly structure:

[0023] D xlg =D dgbl -T dgb -T xlg Formula 3

[0024] The positioning specifications of the components are exported specifications, and the component positioning specifications are established by taking the connecting rod coordinate origin as the global coordinate origin.

[0025] Preferably, the finite element SOM model information structure is generally divided into CAD information and CAE information, wherein the CAD information includes component information and assembly information, and the CAE information includes material information, analysis step information, interaction information, boundary load information, mesh information, and task information;

[0026] The finite element SOM model information is taken as the set M, and its composition is expressed by formula 4:

[0027] M={I Part , I Assem , I Mat , I Bcload , I Mesh , I Int , I Step , I Job Formula 4

[0028] Where: I part Part information refers to the geometric topological information of the part entity and the auxiliary information used for finite element modeling. The geometric topological information includes the point, line, surface, and body information of the part entity. The auxiliary information includes reference points, lines, surfaces, and feature sets. The part information is a description of the part features and provides selection data for adding assembly contacts and meshing the model.

[0029] L assem Assembly information refers to the relevant geometric information and auxiliary information contained in the assembly. In addition to the instantiation of the part instance and the corresponding inherited part information, the assembly information also includes the reference points, lines, surfaces and feature sets defined under the assembly global coordinates; I mat Material information refers to the relevant material properties assigned to the part, including material density, elastic modulus, Poisson's ratio, plastic material model and related parameters, material direction information, and the material model also includes the cross-sectional properties of the part; I bcloadBoundary load information describes the load conditions of the model and the boundary conditions it depends on. Load conditions include force, moment loads, and the relationship between load and time. Boundary conditions refer to the constraints on the parts in the model. Boundary load information depends on the auxiliary information provided by assembly information. mesh Mesh information, including data related to meshing of part instances; I int In order to exchange information, the geometric entities of the hinge pin and the sleeve are established, and the whole is used as a hinge pair for mechanism analysis and solution; I step I is the analysis step information, which refers to the type of model analysis problem, the length of time used in the analysis step, the time increment, the mass magnification factor, and also includes the historical output variables reflecting the analysis process and result data, contact force and field output variables, stress strain and output frequency; I job It is the task information, which refers to the category of the established analysis task and the related solution control parameters. The analysis category includes conventional analysis and restart. The related solution control parameters refer to the number of solution cores for parallel operation, the number of calculation domains and the memory size for calculation and solution. SOM object finite element SOM information model: The information model is divided into a set of operations and parameters, and any subset in the information model is expressed as:

[0030]

[0031] The finite element SOM information model is described as:

[0032]

[0033] Subset I in the finite element SOM information model i Defined as class (Class) Cls, the operation in the information model is defined as function (Function) Fcn, and the parameter in the information model is defined as attribute (Attribute) Atr.

[0034] Preferably, the modeling process based on the finite element SOM information model: creating a design object based on the finite element SOM information model for analysis and solution, the specific steps are as follows:

[0035] Step 1: Call the model layer script, read the information in the information library, and initialize the model design related parameters, including component specifications, assembly parameters, material parameters, mesh parameters, boundary load parameters, and interaction parameters; Step 2: Instantiate the part class, call the geometric modeling function of each component, and establish the geometric model of the component; Step 3: Instantiate the material class, call the public function Material&assign() function, establish section properties for the model, add materials, and assign materials to the parts; Step 4: Instantiate the assembly class, call the Assem() function, calculate the assembly position of each component, establish auxiliary reference points, and assemble the components; Step 5: Call the Rp_Bulid() function to add auxiliary reference points to the model and modify the reference point properties; Step 6: Call the Set&Surface_Bulid() function to establish auxiliary surfaces and collections based on the association characteristics and geometric features of the model components; Step 7: Instantiate the analysis step class, call the Step() function, Establish an analysis step class, set the analysis type and analysis duration; Step 8: Instantiate the interaction class, call the Interaction_Bulid() function, establish contact properties, hinge properties, and establish interactive connections for components; Step 9: Call the Output() function to set output variables, output domains, and output frequencies for the model; Step 10: Instantiate the load boundary class, call the Load() function, add boundary conditions and input loads to the simulation model, including loading input load curves and constraining degrees of freedom; Step 11: Instantiate the task class, call the Job() function, and set the solution method, number of processors, and amount of memory occupied for the model; Step 12: Analyze the calculation results. If they are not ideal, use the Import() function to import the finite element SOM model, and then call the model layer script to set parameters; Step 13: Call the ConstrainDel() and PartDel() functions to delete the component and the interactive connections and features associated with the modified component, and call the component modeling function to model.

[0036] Preferably, the motion posture analysis of the articulated pair: during the movement of the pin and the sleeve in the articulated pair, there are several different relationships between their relative positions and the force conditions. According to the relationships they show, the motion postures of the articulated pair are classified, including 5 postures: Posture I: When the articulated pair is in posture I, the sleeve connected to the middle connecting rod, called the outer sleeve, moves downward under the force of the middle connecting rod. At this time, the pin and the sleeve connected to the main shaft become the inner sleeve, and there is no contact or collision due to the existence of the gap; Posture II: The outer sleeve contacts and collides with the pin, driving the pin to move, and the pin contacts and collides with the inner sleeve, driving the main shaft to rotate; Posture III: The pin is separated from the inner and outer sleeves and is in a free state, the transmission of the mechanism force is interrupted, and the downstream mechanism consumes its own energy to move; Posture IV: The characteristics of the mechanism's own structure and the driving input load, the inner sleeve drives the pin, the pin contacts and collides with the outer sleeve, and does work on the upstream mechanism; Posture V: During the movement of the mechanism, the sleeve is deformed by force, causing the pin to get stuck;

[0037] During the movement of the mechanism, the articulated pair switches between four postures: I, II, III, and IV. Posture V only appears when the gap is small and the articulated pair is stuck. The articulated flexible sleeve of the mechanism is deformed during the movement, and the circular inner diameter section becomes an ellipse. When the clearance value of the articulated pair is too small, the short diameter of the ellipse is smaller than the outer diameter of the pin shaft after the sleeve is deformed, and the articulated pair is stuck.

[0038] Preferably, the genetic space sampling strategy is: sample points are taken in the design interval to form a sample point set S, and the response data value Y is obtained through response solution. According to the sample data [S|Y], a response surface model is constructed based on the n of each sample point. v The minimum sphere radius without intersection in the design space is used as the measurement standard. The larger the minimum sphere radius, the better the spatial distribution characteristics of the sample points. p The smaller the value, the greater the distance between the design sample points. Its mathematical expression is:

[0039]

[0040] In the formula, p is a positive integer, n s is the number of sample points, d ij is the distance between two points composed of sample points, and the distance between any two points is expressed by formula 7:

[0041]

[0042] First, encode the data to describe the LHS sampling samples in the form of a matrix. In the matrix L of size m×n, each row represents a set of sample points, and each column contains a set of 1-m integer sequences:

[0043]

[0044] In the formula, xj is the jth group of sample points; the operation of matrix L initializes the population to generate N pop samples, N pop is an even number, the selection operation sorts the individuals in the population according to their fitness values ​​and selects the top N pop / 2 best samples; crossover operation uses the selected N pop / 2 The number of generated samples is N pop The new population is divided into S1 = [1, (N pop / 2+1)] and S2=[N pop / 2,N pop ] two intervals, S1 interval is based on the optimal individual, and [2,N pop / 2] randomly exchange a column to generate new individuals, and the S2 interval is [2, N pop / 2], randomly exchange a column with the best individual to generate a new individual, and the combination of S1 and S2 constitutes a new population; the mutation operation acts on all individuals in the new population except the best individual, and generates a random number in the interval [0, 1] for each column. If the random number is less than the given mutation rate value p mut , then randomly swap two elements in the column.

[0045] Preferably, the SOM proxy model: using known sample data [X m×n |Y m×p ] Establish an approximate mathematical expression relationship between the input variable x and the output response y, assuming that the design point set S = [s1, s2, ..., s m ] T ,s i ∈R n The system outputs a one-dimensional (q=1) response value Y=[y1, y2, ..., y m ] T ,y i ∈R q , expressed as a regression model and random function:

[0046]

[0047] The regression model uses a linear regression model consisting of k basis functions:

[0048]

[0049] Assuming the mean of the random function z is 0, the covariance between z(w) and z(x) is:

[0050] E[z(w)z(x)]=σ 2 R(θ,w,x) Formula 11

[0051] In the formula, σ 2 is the variance of the random function z, R(θ, w, x) is the association model, and θ is the association model parameter;

[0052] Therefore, for the design sample point set S, its design matrix F is expressed as:

[0053]

[0054] The correlation matrix R of the random function is:

[0055]

[0056] The correlation vector r between the unknown point x and the design sample point set S is:

[0057] r(x)=[R(θ,s1,x)…R(θ,s m , x)] T Formula 14

[0058] The predicted response value at x is expressed as:

[0059]

[0060] The prediction error is:

[0061]

[0062] Where Z = [z1,…,z m ] T The error of the design sample point set is kept unbiased by the prediction, F T c(x)-f(x)=0, the average variance of the prediction error for:

[0063]

[0064] beg The Lagrange equation for the minimum value of c while maintaining the unbiased property is:

[0065] L(c,λ)=σ 2 (1+c T Rc-2c T r)-λ T (F T cf)

[0066] L′(c,λ)=2σ 2 (Rc-r)-λF Equation 18

[0067] According to the necessary optimization conditions and the no-deviation constraint relationship, we can get:

[0068]

[0069] Solving equation 19 yields:

[0070]

[0071] Matrix R and R- 1 is a symmetric matrix, then:

[0072]

[0073] The least squares solution of the regression model with respect to R is:

[0074] β * =(F T R -1 F)F T R -1 Y formula 22

[0075] Combining equation 21 and equation 22, we get:

[0076]

[0077] In the formula, Rγ * =Y-Fβ * ;

[0078] The model prediction error is:

[0079]

[0080] In the formula, u and σ 2 are process quantities,

[0081] u=F T R -1 rf

[0082]

[0083] Construct a SOM agent model.

[0084] Preferably, the sequence optimization point adding criterion is: by analyzing the corresponding response value on the basis of the initial sample, new sample points are added to improve the accuracy of the model. In the optimization process, not only the minimum value in the prediction model is considered, but also the sparsity of the model sample points is considered. The sparser the sampling points, the lower the credibility of the prediction model.

[0085] For a design point x, its response value y(x) cannot be known before performing expensive calculations to predict its mean and mean square error σ 2 , the uncertainty of y(x) is taken as the uncertainty about the variable Y, and the mean and mean square error are and σ 2The normal distribution of , the point-adding criterion weighs all possible points that can obtain the optimal value, and its expression is:

[0086]

[0087] and Φ(·) are the density function and distribution function of the standard normal distribution, respectively. The first term is the difference between the current optimal value and the predicted value multiplied by the increased probability, and the second term is the standard deviation multiplied by the probability density. The larger the difference between the current optimal value and the predicted value and the smaller the prediction variance, the larger the first term is, that is, looking for a point where the predicted value is significantly smaller than the current optimal value. When the prediction variance is larger, the larger the second term is, but it does not deviate from the current optimal point due to the probability density constraint, that is, looking for a point with large uncertainty, weighing the optimal value and uncertainty of the prediction model, and selecting the next sample point.

[0088] Preferably, the optimization estimation based on the spatial optimal model includes single-step optimization and sequence optimization. The single-step optimization constructs a proxy model after sampling once, checks the accuracy of the proxy model, and optimizes based on the proxy model if the requirements are met. The sequence optimization increases samples by considering the predicted value and the predicted variance, continuously updates the model, improves the model accuracy, and finds the optimal value for the proxy model. When the convergence criterion is met, the iteration is stopped. The steps are as follows: Step 1: Adopting the genetic space sampling strategy, construct an initial design sample point S with a dimension size of m; Step 2: According to the design objective function value, solve the target response value Y corresponding to the design sample point S; Step 3: Construct the spatial optimal proxy model using the sample design point S and the response value Y; Step 4: Use the optimization algorithm to solve the proxy model and obtain the current maximum expected enhancement value; Step 5: Determine whether the iteration meets the convergence criterion. If it meets the convergence requirement, stop the iteration and use the optimization algorithm to solve the optimal value of the current proxy model as the optimization calculation result; otherwise, execute the sixth step; Step 6: Use the maximum expected enhancement value plus point criterion to obtain new sample points, add them to the sample S, and execute the second step.

[0089] Preferably, the wear analysis of the articulated pair of the operating mechanism: taking the wear of the articulated pair as the optimization target and the articulated fit tolerance (the articulated clearance and the tolerance band deviation value) as the design variable, the articulated pair is optimized;

[0090] 1) Design variables: In engineering, the clearance of the articulated joint is the difference between the inner diameter of the sleeve and the outer diameter of the pin, which is a relative change value, considering the reference deviation value ε s , with deviation value and gap value as design variables;

[0091] 2) Optimization goal: The wear model calculation formula is:

[0092]

[0093] Where V is the wear depth, s is the relative sliding distance, and Fn is the normal contact force, k is the dimensionless wear coefficient, H is the Brinell hardness of the softer material in the contact material, and the contact force is obtained by finite element simulation of the contact history variable normal contact force, and the relative relationship of the coordinate position of the relative sliding pin shaft sleeve mark point during the movement process is calculated;

[0094] 3) Constraints: Interval constraints are imposed on the design variable values. The interval constraint relationship is as follows:

[0095]

[0096] The unit of constraint condition is μm. The GALHS method is used for initial sampling, the size of initial sample point matrix is ​​10×2, and the number of iterations is 20. When the gap value is small or large, the wear degree is relatively high. When the gap value is around [60,110], the wear degree of the articulated joint is relatively low, and most of the optimal search points are concentrated in this interval.

[0097] Compared with the prior art, the innovations and advantages of this application are:

[0098] (1) This application analyzes the structural composition and motion analysis of the electric high-voltage circuit breaker SOM, simplifies it reasonably, considers the high-speed impact problem in the working environment, reasonably selects the material model and contact model, establishes a finite element simulation model of the operating mechanism in the ABAQUS environment, and uses the constructed physical test bench to modify and verify the simulation model with the displacement curve, and obtains a simulation model with good fit. Then, the idea of ​​feature modeling is used to quantify the parts of the parts, the outline pixels are divided into structural pixels and contour pixels, the constraint relationship of the structural pixels is established and solved according to the geometric relationship, the contour pixels are calculated using the obtained structural pixels, and the quantification of the parts is completed in combination with the feature parameters. An information model is created based on the relevant information of finite element modeling, and secondary development is carried out using the Python language to complete the finite element component quantitative modeling of the SOM, and C# is used to develop component quantitative information management software to assist in the management of modeling information. The study analyzed the motion deviation of the operating mechanism caused by the articulated gap and its impact on the articulated pair. The Archard model was used to analyze the degree of articulated wear and the articulated fit tolerance (gap value and deviation value) as variables, which provided assistance for the design of the operating mechanism, greatly reduced the design risks and hidden dangers of SOM, and was conducive to the large-scale promotion and application of SOM of high-voltage circuit breakers.

[0099] (2) This application establishes a finite element simulation model of the circuit breaker SOM and uses test data to correct and verify the model to ensure the reliability of the model; uses a feature-based component quantitative modeling method to establish a finite element component quantitative model of the object and develops parameter information management auxiliary software; uses proxy model technology to establish its simulation approximate model, studies and analyzes the impact of the articulated clearance on the articulated pair and mechanism, provides feedback for mechanism design, and is specifically used to solve the most important and complex design and test problems of SOM. After continuous experimentation and improvement of SOM feature quantitative proxy modeling, it can withstand high-speed impact loads and ensure high reliability. SOM drives the static and moving contacts to complete the opening and closing of the switch through a transmission mechanism, has good motion characteristics of the transmission mechanism, can ensure the performance and reliability of the mechanical system, and solves the problem that the kinematic pairs between the mechanisms cannot achieve ideal connection due to limitations in mechanical design and processing and manufacturing. The existence of the kinematic pair clearance causes the mechanism movement to deviate from the ideal design value, and at the same time causes friction and wear of the kinematic pair itself, promotes the damage of the joint surface, and causes a larger gap. The cost of the test method of this application is greatly reduced, no random errors are introduced, the computing power of the simulation technology is enhanced, and the time occupied by the pre- and post-processing of finite element simulation modeling is greatly shortened, which greatly reduces the cost of variable parameter or optimization analysis and improves analysis efficiency.

[0100] (3) The present application constructs a method for quantitative modeling of components of a high-voltage circuit breaker SOM. Based on the quantification of components based on feature modeling, the outline image pixels are divided into structural pixels and contour pixels, and the model is calculated and solved by constraint solving to enhance applicability; and the mechanism is solved to complete the quantitative modeling of the assembly components of the target object, and a finite element component quantitative model of the object is established. At the same time, information software for managing relevant parameter data is developed, and the influence of the articulation gap on the movement of the mechanism is analyzed. The genetic space sampling strategy is adopted, and the "expected enhancement" is used as the sequence adding criterion to establish a spatial optimal agent model. In combination with ABAQUS simulation software, an agent modeling program based on the spatial optimal model is developed, and the program is applied to the research and analysis of the articulation wear of the operating mechanism, which plays a very important role in eliminating or reducing the design risks and hidden dangers of the use of the high-voltage circuit breaker SOM. BRIEF DESCRIPTION OF THE DRAWINGS

[0101] Figure 1 It is a schematic diagram of the connecting rod structure of the operating mechanism assembly.

[0102] Figure 2 It is a schematic diagram of fire hazard estimation and analysis of solar photovoltaic cell modules.

[0103] Figure 3 It is a flow chart for creating design object analysis and solution based on the finite element SOM information model.

[0104] Figure 4It is a schematic diagram of the dispersion degree of the maximum deformation area of ​​the sleeve.

[0105] Figure 5 It is a schematic diagram of the parameterized system architecture of the actuator.

[0106] Figure 6 It is a flow chart of optimizing estimation using the spatial optimal model.

[0107] Figure 7 It is a diagram of the deformation statistics of the dumpling sleeve at the end face and the center trajectory of the pin shaft.

[0108] Figure 8 It is a schematic diagram of the calculation and expression of the pin shaft flexural deformation deflection.

[0109] Fig. 9 It is a flow chart of optimizing estimation using the spatial optimal model. DETAILED DESCRIPTION

[0110] The following, in conjunction with the accompanying drawings, further describes the technical solution of the SOM feature quantification agent modeling method for a high-voltage power circuit breaker provided in the present application, so that those skilled in the art can better understand the present application and implement it.

[0111] (A) Taking a certain type of circuit breaker operating mechanism as the object, the composition of the mechanism is analyzed and reasonably simplified. The material model provided in the simulation software and the differences in the related settings of the contact model are analyzed. The relevant parameters are reasonably selected and set, and a finite element simulation model is established. The model is modified and verified using the test data from the built test bench, and a finite element SOM model with good accuracy is obtained, which provides a basic guarantee for the subsequent quantitative modeling and computational analysis of components.

[0112] (B) In order to solve the time-consuming and laborious problem of finite element modeling, the finite element component quantification technology is used to establish the finite element component quantification model of the object. Based on the feature modeling concept, the component entity is modeled quantitatively. The pixels of the feature outline are divided into structural pixels and contour pixels. The structural pixel information is obtained by analyzing, establishing and solving the constraint relationship of the contour pixels. The contour pixels are solved based on this, and the component quantification is realized by combining the feature parameters. According to the characteristics of the assembly connecting rod, the vector ring equation of the connecting rod is established to solve the connecting rod, and the specification chain in the Z axis is established according to the assembly relationship. The component quantification of the assembly is completed through the solution results of the two. Following the object-oriented and information modeling concepts, the finite element modeling information is abstractly organized, and the finite element SOM information model is established. The Python language is used to carry out secondary development of ABAQUS to realize the finite element component quantification modeling. In order to facilitate the management of modeling information, the simulation information management software of the object is developed using the C# language.

[0113] (C) The high cost of simulation calculation makes it difficult to carry out optimization analysis of actual engineering systems using traditional optimization methods. The proxy model technology solves this problem better. This application focuses on the influence of articulated clearance on the mechanism and articulated pair, analyzes the mechanism motion deviation characteristics caused by different clearance values, and finds that the clearance value increases so that the motion deviation changes from small multiple oscillations to large few oscillations. Small clearance often leads to contact area concentration. When the clearance is large, the regional concentration phenomenon is weakened. However, due to the inherent characteristics of the mechanism and the limitations of input conditions, regional concentration is inevitable. Based on the spatial optimal proxy model, considering the importance of the initial sample points, the genetic algorithm is used to optimize the sampling to obtain the initial sample, and the expected enhanced point addition criterion is used to perform sequential point addition, establish an approximate model, and verify its approximate ability with a known function. The approximate model is applied to the simulation optimization analysis of the research object, with the contact wear degree calculated by the Archard model as the target, and the articulated pair matching tolerance (tolerance band and deviation) as the variable. Analysis and calculation are carried out, and it is found that the tolerance specification is a factor affecting wear. Too large or too small clearance will cause greater wear. The research object has a smaller degree of wear when it is about 60 to 110um.

[0114] 1. Quantitative modeling of circuit breaker SOM finite element components

[0115] Based on the quantitative modeling of finite element parts, the existing finite element modeling pre-processing mostly adopts the GUI method, which has a large modeling workload and low efficiency, especially when performing variable parameter analysis. This application adopts the quantitative modeling method of parts of the information model, encapsulates the operations required for modeling in the program code, focuses on the design parameters, and avoids the modeling failure or error caused by human factors in the GUI modeling method.

[0116] 1) Adopting the framework of feature modeling, establish a feature-based parts quantification model, establish and solve the constraint relationship for the feature outline construction elements, construct the mathematical representation of the contour elements, and realize the parts quantification construction of the parts geometric solid model; 2) Establish the planar position relationship of the parts assembly, solve the vector ring equation to obtain the position and initial posture of the parts on the two-dimensional projection surface, and combine the Z-axis specification chain to complete the parts quantification of the spatial assembly; 3) Create a finite element SOM information model of the actuator mechanism, express the feature information in the form of abstract classes, establish a finite element parts quantification model of the actuator mechanism, realize the reuse of the finite element SOM model, and develop a parts quantification information management platform. Compared with the traditional GUI modeling method, this method has significantly improved the creation efficiency and provided support for simulation optimization.

[0117] (I) Quantitative construction of component features

[0118] 1. Component quantification based on feature modeling

[0119] The construction of a part is a stacking of basic features, which converts the quantification of the components of the part into the quantification of the components of the features that constitute the part. Most features are based on 2D sketches and are swept along the third direction of the space to form entities. Sweeping includes translation along a specified vector, rotation along an axis, and movement along a specific curve. The component design parameters are divided into feature parameters and sketch parameters.

[0120] 2. Quantification of feature outline components

[0121] The quantitative construction process of parts is divided into three steps: drawing a skeleton diagram, adding constraints, and solving constraints. In the skeleton diagram drawing stage, the designer defines the topological characteristics of the skeleton diagram without paying attention to the precise specifications of the pixels. In the constraint adding stage, corresponding constraints are added to the pixels according to the design requirements. In the constraint solving stage, the corresponding algorithm is used to solve the established constraint relationship to obtain a skeleton diagram instance.

[0122] (1) Analysis of constraints in the outline diagram

[0123] Component quantification is performed based on the morphological features of geometric entities, and constraint types are divided into topological constraints, geometric constraints, and specification constraints. Topological constraints refer to the relationship between the point, line, and ring topological elements that constitute the graphics. They are formed after the component design is completed and are an implicit constraint relationship. Topological constraint information is used to solve ambiguity problems. Geometric constraints refer to the relative position relationship between the pixels that constitute the graphics. The final form of the constraint is mainly to constrain the position of the graphics. Constraint solving is the process of solving new feature points based on known specification design parameters and known feature points. Specification constraints constrain the shape and size of the pixels, and the specifications are divided into variable specifications and derived specifications. Variable specifications are specifications adjusted during the design process and are independent of other specifications. Derived specifications refer to specification values ​​that are restricted by other specifications and are calculated based on the dependent specification parameters and specification constraint relationships.

[0124] (2) Pixel expression method

[0125] It only includes the topological elements that constitute the skeleton diagram instance. The pixels are divided into construction pixels and contour pixels. The construction pixels analyze and solve the quantitative model of the parts. The points, lines, and circles of the characteristic points are calculated according to the constraints. The contour pixels use the pixels obtained by calculation to draw the contour loop, which is composed of line segments and arcs. The composition parameters of all construction pixels are independent of each other, and the complete degree of freedom is retained. The contour pixels are established based on the construction pixels, and the composition parameters are restricted by the constraints and construction pixels.

[0126] Different expressions of pixels are regarded as unresolved constraints, and only one expression is used for each pixel. A two-dimensional point has two degrees of freedom, which are represented by coordinates. A two-dimensional line contains two degrees of freedom, which are represented by a point on the line and its vector angle. A two-dimensional circle contains three degrees of freedom, which are represented by the center and radius. Line segments and arcs are represented by two points and three points respectively.

[0127] (3) Constraint solving

[0128] The pixels of a single loop are divided into structural pixels and contour pixels, and the solution of constraints is divided into structural pixel solution and contour pixel solution. The skeleton diagram structural pixel solution is the process of solving feature points based on the graphic constraint relationship and the input design parameters using the structural pixel solution method. The contour pixel solution is to calculate and solve the contour pixel using the structural pixels obtained by the solution, and the constraint on the structural pixel is positioning;

[0129] The constraints between pixels are unidirectional, meaning that one pixel O depends on other pixels O. i , when pixel O i When it changes, it will cause the pixel О to change, but the change of O will not affect O. i Cause impact, Pixel O i is the main variable, O is the secondary variable;

[0130] The constraint solving of pixels is divided into two stages: analysis stage and construction stage. In the analysis stage, the outline drawing process is analyzed according to the outline structure and the construction pixels are established. In the construction stage, constraint relationships are established for pixels according to the drawing process.

[0131] (II) Quantification of assembly parts

[0132] When creating parts, the initial point is set at the origin. When assembling, the relative position relationship between parts is adjusted. In the assembly module, the relative position and initial motion posture of parts are adjusted by using the feature relationship of parts.

[0133] The assembly parts are associated through coaxial lines, parallelism, relative distance, and relative rotation angle related constraint behaviors to ensure that under given initial conditions, each component is adjusted to the corresponding posture. The coordinate system in the assembly drawing is the global coordinate system. The coordinate system used by each component in modeling is called the local coordinate system. The position of the local coordinate system in the global coordinate system is calculated. The position of some local coordinate systems in the global coordinate system is relatively fixed and directly represented by the relative position vector. The actual linkage mechanism contains secondary rod groups and higher-level rod groups. The quantification of assembly parts needs to be solved. The operating mechanism contains two four-bar linkages, and their positions are solved;

[0134] like Figure 1 As shown, the transmission part of the operating mechanism is composed of connecting rods l1, l2, l3, l4, l5, l6, l7 and frames l4, l8, among which l1, l2, l3, l4 constitute four-link I, which transmits motion and force from the upper connecting rod to the main shaft, and l5, l6, l7, l8 constitute four-link II, which transmits motion from the main shaft to the large crank arm. The four-link mechanism establishes a closed vector loop.

[0135]

[0136] In the formula, For vector The angle with the x-axis, l i For vector Model;

[0137] When solving, project them onto the coordinate axis respectively and convert them into scalar form equations. Equation 2 is its scalar form:

[0138]

[0139] A and D are external hinge points. After their coordinates are given as initial conditions, Equation 2 contains only the design variables. and The equation is solvable;

[0140] Solve the four-bar linkage I to obtain by The angle θ with the main axis is used as the initial condition to solve the four-bar linkage II, and the plane motion position of the transmission mechanism under the given initial condition is obtained. The specification driving relationship in the plane normal direction is established according to the assembly relationship;

[0141] The operating mechanism includes three phases, each phase includes a pair of main shaft crank arms, lower connecting rods, and large crank arms. The large crank arms and lower connecting rods are sub-assembled separately. The large crank arm sub-assembly parameter is D dgb1 With D dgb2 , the lower connecting rod assembly parameters are D xlg With D dgb2 , obtained from the assembly structure:

[0142] D xlg =D dgbl -T dgb -T xlg Formula 3

[0143] The positioning specifications of the parts are exported. The connecting rod coordinate origin is used as the global coordinate origin. The positioning specifications of the parts are established as follows: Figure 2 shown.

[0144] 3. Feature-based finite element SOM information model

[0145] 1. Finite element SOM model information structure

[0146] In terms of overall structure, it is divided into CAD information and CAE information. CAD information includes component information and assembly information, and CAE information includes material information, analysis step information, interaction information, boundary load information, mesh information, and task information.

[0147] The finite element SOM model information is taken as the set M, and its composition is expressed by formula 4:

[0148] M={I Part , I Assem , I Mat , I Bcload , I Mesh , I Int , I Step , I Job Formula 4

[0149] Where: I part Part information refers to the geometric topological information of the part entity and the auxiliary information used for finite element modeling. The geometric topological information includes the point, line, surface, and body information of the part entity. The auxiliary information includes reference points, lines, surfaces, and feature sets. The part information is a description of the part features and provides selection data for adding assembly contacts and meshing the model.

[0150] L assem Assembly information refers to the relevant geometric information and auxiliary information contained in the assembly. In addition to the instantiation of the part instance and the corresponding inherited part information, the assembly information also includes the reference points, lines, surfaces and feature sets defined under the assembly global coordinates;

[0151] I mat Material information refers to the relevant material properties assigned to the part, including material density, elastic modulus, Poisson's ratio, plastic material model and related parameters, material direction information. The material model also includes the cross-sectional properties of the part.

[0152] I bcload Boundary load information describes the load conditions of the model and the boundary conditions it depends on. Load conditions include force, moment loads, and the relationship between load and time. Boundary conditions refer to the constraints on the parts in the model. Boundary load information depends on the auxiliary information provided by assembly information.

[0153] I mesh Mesh information, including data related to meshing of part instances;

[0154] I int In order to exchange information, the geometric entities of the hinge pin and the sleeve are established, and the whole is used as a hinge pair for mechanism analysis and solution;

[0155] I step Analysis step information, which refers to the type of model analysis problem, the length of time used in the analysis step, the time increment, the mass magnification factor, and also includes historical output variables reflecting the analysis process and result data, contact force and field output variables, stress strain and output frequency;

[0156] I jobTask information refers to the type of analysis task established and related solution control parameters. Analysis types include regular analysis and restart. Related solution control parameters refer to the number of solution cores for parallel computing, the number of computational domains, and the memory size for computing solutions.

[0157] 2. SOM object finite element SOM information model

[0158] The information model is divided into a set of operations and parameters, and any subset of the information model is represented as:

[0159]

[0160] The finite element SOM information model is described as:

[0161]

[0162] Subset I in the finite element SOM information model i Defined as class (Class) Cls, the operation in the information model is defined as function (Fcn), and the parameter in the information model is defined as attribute (Attribute) Atr;

[0163] In the organizational structure of the finite element SOM information model, A1 is the component quantification module, which solves the target object constraints and assembly positions according to the design parameters, the design object components and their assembly relationships, and obtains the component and assembly geometric entity templates; A2 is the template class, which stores the subset I in the form of a class. i The part class in A2 is the manifestation of the specific part structure; the public class includes classes with strong versatility such as assembly and material, and the public class is applied to different finite element SOM models; A3 is the finite element SOM model of the design object, which is the final expression of the required model.

[0164] The finite element SOM information model consists of four layers: model layer, component layer, function layer, and code layer;

[0165] (1) The code layer is located at the bottom layer and provides the necessary finite element software interface for the information model. It receives parameters transmitted by the upper layer and transmits commands to the finite element software to enable it to perform corresponding operations.

[0166] (2) The function layer is located above the code layer. The function layer contains functions and organizes unrelated codes.

[0167] (3) The construction layer is the superior of the function layer. The construction layer exists in the form of classes, reflecting information subsets. The attributes in the class represent the parameter set required to build the model. The functions in the class are the operation sets contained in the finite element creation of the object. The creation of the design object is the instantiation of the construction class.

[0168] (4) The model layer is the top layer, which is the description of the finite element SOM model of the design object. It instantiates the construction class according to the design requirements, establishes the finite element SOM model, controls and modifies the design parameters, creates a design object that meets the requirements, and outputs the log at the same time.

[0169] 3. Modeling process based on finite element SOM information model

[0170] Create design objects based on the finite element SOM information model for analysis and solution. The flow chart is as follows Figure 3 , where B1 is the finite element modeling process, the specific steps are as follows:

[0171] Step 1: Call the model layer script, read the information in the information library, and initialize the model design related parameters, including component specifications, assembly parameters, material parameters, mesh parameters, boundary load parameters, and interaction parameters;

[0172] Step 2: Substantiate the part class, call the geometric modeling function of each component, and establish the geometric model of the component;

[0173] Step 3: Instantiate the material class, call the public function Material&assign() to create cross-section properties for the model, add materials, and assign materials to the parts;

[0174] Step 4: Instantiate the assembly class, call the Assem() function, calculate the assembly position of each component, establish auxiliary reference points, and assemble the components;

[0175] Step 5: Call Rp_Bulid() function to add auxiliary reference points to the model and modify the reference point properties;

[0176] Step 6: Call the Set&Surface_Bulid() function to create auxiliary surfaces and collections based on the associated characteristics and geometric features of the model components;

[0177] Step 7: Instantiate the analysis step class, call the Step() function, create an analysis step class for the model, and set the analysis type and analysis duration;

[0178] Step 8: Instantiate the interaction class, call the Interaction_Bulid() function, establish contact properties and hinge properties, and establish interactive connections for components;

[0179] Step 9: Call the Output() function to set the output variable, output domain, and output frequency for the model;

[0180] Step 10: Instantiate the load boundary class and call the Load() function to add boundary conditions and input loads to the simulation model, including loading the input load curve and constraining the degrees of freedom;

[0181] Step 11: Instantiate the task class, call the Job() function, and set the solution method, number of processors, and memory usage for the model.

[0182] Step 12: Analyze the calculation results. If they are not ideal, use the Import() function to import the finite element SOM model, and then call the model layer script to set the parameters.

[0183] Step 13: Call ConstrainDel() and PartDel() functions to delete the component and the interactive connections and features associated with the modified component, and call the component modeling function to model;

[0184] Figure 3 B2 represents the model reuse process, which uses CreateOdb() and OdbFill() functions to read the nodes, units and analysis result information contained in the specified set in the model, and uses the read information to create a new result model database file for analysis and calculation.

[0185] (IV) Quantitative information management platform for circuit breaker operating mechanism components

[0186] The components, assembly and finite element SOM information of SOM are quantified, and feature-based finite element component quantification modeling is performed. The ABAQUS / PYTHON secondary development interface is used to establish the circuit breaker SOM finite element SOM information model. The core calculation is implemented in Python language, and the component quantification information management platform is developed using C# software.

[0187] Quantitative system architecture of actuator components Figure 5 As shown, it includes a core calculation module and an information management module. The core calculation model receives design parameters, calculates the structural shape of parts and components, the position and posture of the assembly according to the parameters, interprets the finite element modeling information, and outputs commands to the ABAQUS software to realize the finite element modeling of the actuator. The information management module manages the design parameters required for modeling, including historical query, parameter report output, design parameter output, and interactive input module, providing convenient and fast parameter interaction and information management, and improving the efficiency of quantitative modeling of parts.

[0188] When modeling, follow the modeling process in the navigation bar, fill in the relevant parameters in the workspace, and complete the input and setting of CAD / CAE related design data such as the geometric structure specifications of parts, assembly positioning, material properties, mesh parameters and division algorithms, interactive information such as contact, hinge property settings, and boundary load conditions.

[0189] 2. SOM agent modeling and articulation characteristics analysis

[0190] Due to design requirements, processing accuracy, and long-term operation, there are often gaps in the mechanism, which causes the mechanism to deviate from the theoretical design point, affecting the motion accuracy and performance. The circuit breaker SOM has the characteristics of large impact and rapid response, and the articulated pin is subjected to severe loads, which can easily lead to pin deformation and wear.

[0191] This application utilizes the established quantitative model of the operating mechanism components and takes the clearance as a variable to analyze the impact of the clearance value change on the articulated pin and the mechanism. The proxy model is used to build a simulation optimization program to optimize the SOM. The pin and the sleeve are collectively referred to as an articulated pair, and the Hinge connection method is called an ideal articulation.

[0192] Based on the quantitative model of parts and components, the articulated clearance is used as a variable for analysis. The motion and dynamic data of the articulated pair and the mechanism under various working conditions are extracted and calculated using the post-processing program. The articulated pair motion characteristics caused by the articulated clearance and its influence on the mechanism motion are analyzed, and an agent modeling program is established based on the agent model; 1) Taking the articulated clearance as a variable, the deformation and relative displacement of the articulated sleeve and the pin are analyzed, and the variation law of the mechanism motion displacement deviation with the clearance specification is obtained. It is found that the clearance specification affects the degree of deformation concentration of the articulated pair. The smaller the clearance, the higher the concentration and the smaller the relative slip rotation; the increase in clearance causes the mechanism displacement deviation to change from small multiple oscillations to large few oscillations; 2) Based on the spatial optimal agent model, the genetic space sampling strategy and the expected enhanced point addition criterion are adopted to establish a finite element agent modeling program in the ABAQUS environment; 3) Using the established agent modeling program, the articulated pair fit tolerance is taken as the design variable, and the wear condition of the articulated pair during motion is taken as the objective function, and the influence of tolerance on the wear of the articulated pair is obtained: the articulated clearance has a greater influence on wear, and excessive or too small clearance will lead to greater wear.

[0193] (I) Analysis of the articulated characteristics of circuit breaker SOM

[0194] The circuit breaker SOM contains three-phase output. There are 9 articulated pairs in the transmission mechanism that play a connecting transmission role. Only one articulated gap is introduced in the finite element SOM model, and the rest are connected by ideal articulated pairs. An articulated gap is introduced at the connection between the middle connecting rod and the main shaft. The focus is on the closing and opening movement process, rather than the vibration after the closing and opening movement is in place, which reduces the simulation analysis time and improves the efficiency of motion analysis calculation.

[0195] 1. Articulated joint motion posture analysis

[0196] There are several different relationships between the relative position and force of the pin and sleeve in the articulated joint during movement. According to the relationships shown, the articulated joint motion postures are classified into five types:

[0197] Posture I: When the articulated joint is in posture I, the sleeve connected to the middle link, called the outer sleeve, moves downward under the force of the middle link. At this time, the pin and the sleeve connected to the main shaft become the inner sleeve, and there is no contact or collision due to the existence of the gap; Posture II: The outer sleeve contacts and collides with the pin, driving the pin to move, and the pin contacts and collides with the inner sleeve, driving the main shaft to rotate; Posture III: The pin is separated from the inner and outer sleeves and is in a free state. The transmission of the mechanism force is interrupted, and the downstream mechanism consumes its own energy to move; Posture IV: The characteristics of the mechanism's own structure and the driving input load, the inner sleeve drives the pin, the pin contacts and collides with the outer sleeve, and does work on the upstream mechanism; Posture V: During the movement of the mechanism, the sleeve is deformed by force, causing the pin to get stuck;

[0198] During the movement of the mechanism, the articulated pair switches between four postures: I, II, III, and IV. Posture V only appears when the gap is small and the articulated pair is stuck.

[0199] The flexible sleeve of the mechanism articulated joint deforms during movement, and the circular inner diameter section becomes an ellipse. When the clearance value of the articulated joint is too small, the short diameter of the ellipse is smaller than the outer diameter of the pin shaft after the sleeve is deformed, and the articulated joint becomes stuck.

[0200] 2. The influence of articulation clearance on articulated joints

[0201] The change of the gap of the circuit breaker SOM will cause discontinuity in the transmission of the mechanism's motion, causing a delay in the mechanism's motion transmission. At the same time, the gap will cause impact problems in the process of force transmission of the mechanism. As for the articulated pair, the gap affects the relative motion and dynamic characteristics of the articulated pair's pin shaft and sleeve. The existence of the gap increases the discontinuity time between the articulated pin shaft and the sleeve, and changes the stiffness of the pin shaft and the sleeve, thus affecting the deformation of the articulated pair. The deformation of the sleeve is observed by the maximum and minimum distances of the inner diameter on the section parallel to the end face, and the deformation of the pin shaft is observed by the bending deflection on the axis line.

[0202] The articulated sleeve is deformed by extrusion and collision impact during movement. The distribution of deformation area is related to the movement characteristics of the mechanism. Due to the characteristics of uneven distribution and regional concentration, when the gap value is small, the regional concentration phenomenon is more obvious. As the gap value increases, the maximum deformation area becomes more dispersed. The articulated pin is mostly in posture II and posture IV, and less in posture III. While postures II and IV are contact movements, their relative positions remain unchanged. Posture III is a free state, and the pin can move in multiple directions during movement. In the case of a small gap, the pin and the sleeve have a relatively fixed contact area during movement. The impact and collision of the pin causes the sleeve to deform, making the deformation area of ​​the sleeve more concentrated. On the contrary, as the gap increases, the proportion of posture III in the articulated pair during movement increases, and the probability of the pin moving in multiple directions increases, which increases the change in the contact area between the pin and the sleeve during movement, and the deformation area of ​​the sleeve becomes dispersed. Figure 7As shown in Figures (b) and (d), the outer circle is the gap value, indicating the range of motion of the center trajectory. Figure 7 As shown in (a), the interval is divided into 36 parts. When the hinge gap is 8um, the deformation area is concentrated between 60 and 65 degrees, with 67 times, accounting for 33.67% of all times, and the statistical variance is 130.47; in the case of large gap, such as Figure 7 As shown in (c), when the hinge gap is 208um, the deformation area is relatively scattered, which contains two maximum deformation areas, -84 to 79 degrees and 27 to 32 degrees, with a frequency of 26 times, accounting for 13.07% of all times, and a statistical variance of 38.25. Figure 4 As shown in the figure, the larger the variance of the dispersion, the more concentrated the maximum deformation area is. Figure 4 It can be seen that the variance of the degree of dispersion decreases as the gap value increases, and the variance of the degree of dispersion maintains a relatively stable value after reaching a certain level.

[0203] The pin is subjected to impact and bends during movement. The line connecting the centers of the two ends of the pin (O1, O2) is used as the standard line. Figure 8 As shown in the figure, the maximum vertical distance from the center line to the standard line is taken as the current deflection. The deflection deformation is related to the gap value and the motion characteristics of the mechanism. The magnitude of the impact load is related to the relative movement speed and acceleration of the two before the collision. When the gap value is small, the pre-movement distance of the articulated pin and the sleeve before the collision impact is short, and the collision impact load is small, resulting in a smaller pin deflection deformation; on the contrary, when the gap increases, the pre-movement distance before the collision increases, resulting in a larger relative speed difference before the collision, a more violent collision, a greater impact, and a greater deflection deformation.

[0204] The maximum deflection of the pin changes with the gap value during movement: the deflection value increases with the increase of the gap value. When the gap value increases to a certain extent, the maximum deflection value tends to a relatively stable state, and the maximum relative speed between the two rods is a constant value. The existence of the gap causes it to fluctuate within a certain range. When the gap is small, the interval between two adjacent contacts is short, the speed direction difference is small, and the relative speed difference maintains a small value. When the gap is large, the time interval between two adjacent large contacts is long, the speed direction changes greatly, and the speed difference between the two connecting rods is large. The change in speed direction is constrained by the motion characteristics of the mechanism and maintains a relatively stable value within a similar time interval, so that when the gap is large, the maximum deflection of the pin fluctuates within a relatively stable value.

[0205] (II) Response surface SOM proxy model

[0206] 1. Genetic space sampling strategy

[0207] Sample points are taken in the design interval to form a sample point set S. The response data value Y is obtained through response solution. According to the sample data [S|Y], a response surface model is constructed. Based on the n of each sample point v The minimum sphere radius without intersection in the design space is used as the measurement standard. The larger the minimum sphere radius, the better the spatial distribution characteristics of the sample points. p The smaller the value, the greater the distance between the design sample points. Its mathematical expression is:

[0208]

[0209] In the formula, p is a positive integer, n s is the number of sample points, d ij is the distance between two points composed of sample points, and the distance between any two points is expressed by formula 7:

[0210]

[0211] The LHS is improved and the algorithm flow is as follows:

[0212] CreateLHS(); Create an initial population

[0213] FitnessLHS(); Population individual evaluation

[0214] _stop = FALSE;

[0215] while(not_stop)

[0216] SelectLHS0; Select operation

[0217] CrossLHS); Cross operation

[0218] MutateLHS); mutation operation

[0219] If convergence condition is met then_stop=TURE;

[0220] end

[0221] First, encode the data to describe the LHS sampling samples in the form of a matrix. In the matrix L of size m×n, each row represents a set of sample points, and each column contains a set of 1-m integer sequences:

[0222]

[0223] In the formula, x j is the jth group of sample points; the operation of matrix L initializes the population to generate N pop samples, N popis an even number, the selection operation sorts the individuals in the population according to their fitness values ​​and selects the top N pop / 2 best samples; crossover operation uses the selected N pop / 2 The number of generated samples is N pop The new population is divided into S1 = [1, (N pop / 2+1)] and S2=[N pop / 2,N pop ] two intervals, S1 interval is based on the optimal individual, and [2,N pop / 2] randomly exchange a column to generate new individuals, and the S2 interval is [2, N pop / 2], randomly exchange a column with the best individual to generate a new individual, and the combination of S1 and S2 constitutes a new population; the mutation operation acts on all individuals in the new population except the best individual, and generates a random number in the interval [0, 1] for each column. If the random number is less than the given mutation rate value p mut , then randomly swap two elements in the column.

[0224] 2. SOM Agent Model

[0225] Using known sample data [X m×n |Y m×p ] Establish an approximate mathematical expression relationship between the input variable x and the output response y, assuming that the design point set S = [s1, s2, ..., s m ] T ,s i ∈R n The system outputs a one-dimensional (q=1) response value Y=[y1, y2, ..., y m ] T ,y i ∈R q , expressed as a regression model and random function:

[0226]

[0227] The regression model uses a linear regression model consisting of k basis functions:

[0228]

[0229] Assuming the mean of the random function z is 0, the covariance between z(w) and z(x) is:

[0230] E[z(w)z(x)]=σ 2 R(θ,w,x) Formula 11

[0231] In the formula, σ 2is the variance of the random function z, R(θ, w, x) is the association model, and θ is the association model parameter;

[0232] Therefore, for the design sample point set S, its design matrix F is expressed as:

[0233]

[0234] The correlation matrix R of the random function is:

[0235]

[0236] The correlation vector r between the unknown point x and the design sample point set S is:

[0237] r(x)=[R(θ,s1,x)…R(θ,s m , x)] T Formula 14

[0238] The predicted response value at x is expressed as:

[0239]

[0240] The prediction error is:

[0241]

[0242] Where Z = [z1,…,z m ] T The error of the design sample point set is kept unbiased by the prediction, F T c(x)-f(x)=0, the average variance of the prediction error for:

[0243]

[0244] beg The Lagrange equation for the minimum value of c while maintaining the unbiased property is:

[0245] L(c,λ)=σ 2 (1+c T Rc-2c T r)-λ T (F T cf)

[0246] L′(c,λ)=2σ 2 (Rc-r)-λF Equation 18

[0247] According to the necessary optimization conditions and the no-deviation constraint relationship, we can get:

[0248]

[0249] Solving equation 19 yields:

[0250]

[0251] Matrix R and R- 1 is a symmetric matrix, then:

[0252]

[0253] The least squares solution of the regression model with respect to R is:

[0254] β * =(F T R -1 F) F TR -1 Y formula 22

[0255] Combining equation 21 and equation 22, we get:

[0256]

[0257] In the formula, Rγ * =Y-Fβ * ; The model prediction error is:

[0258]

[0259] In the formula, u and σ 2 are process quantities,

[0260] u=F T R -1 rf

[0261]

[0262] Construct a SOM agent model.

[0263] 3. Sequence optimization and point addition criteria

[0264] By analyzing the corresponding response values ​​on the basis of the initial samples, new sample points are added to improve the accuracy of the model. In the optimization process, not only the minimum value in the prediction model is considered, but also the sparsity of the model sample points. The sparser the sampling points, the lower the credibility of the prediction model, and there may be a minimum value.

[0265] For a design point x, its response value y(x) cannot be known before performing expensive calculations to predict its mean and mean square error σ 2 , the uncertainty of y(x) is taken as the uncertainty about the variable Y, and the mean and mean square error are and σ 2The normal distribution of , the point-adding criterion weighs all possible points that can obtain the optimal value, and its expression is:

[0266]

[0267] and Φ(·) are the density function and distribution function of the standard normal distribution, respectively. The first term is the difference between the current optimal value and the predicted value multiplied by the increased probability, and the second term is the standard deviation multiplied by the probability density. The larger the difference between the current optimal value and the predicted value and the smaller the prediction variance, the larger the first term is, that is, looking for a point where the predicted value is significantly smaller than the current optimal value. When the prediction variance is larger, the larger the second term is, but it does not deviate from the current optimal point due to the probability density constraint, that is, looking for a point with large uncertainty, weighing the optimal value and uncertainty of the prediction model, and selecting the next sample point.

[0268] 3. Proxy estimation method based on spatial optimal model

[0269] 1. Optimization estimation based on spatial optimal model

[0270] It includes single-step optimization and sequence optimization. Single-step optimization constructs a proxy model after sampling once, checks the accuracy of the proxy model, and optimizes based on the proxy model when it meets the requirements. Sequence optimization increases samples by considering the predicted value and predicted variance, continuously updates the model, improves the model accuracy, and finds the optimal value for the proxy model. It stops iterating when the convergence criterion is met. The steps are as follows:

[0271] The first step is to use the genetic space sampling strategy to construct an initial design sample point S with a dimension of m; the second step is to solve the target response value Y corresponding to the design sample point S according to the design objective function value; the third step is to use the sample design point S and the response value Y to construct the spatial optimal proxy model; the fourth step is to use the optimization algorithm to solve the proxy model and obtain the current maximum expected enhancement value; the fifth step is to determine whether the iteration meets the convergence criterion. If the convergence requirement is met, the iteration is stopped and the optimization algorithm is used to solve the optimal value of the current proxy model as the optimization calculation result; otherwise, the sixth step is executed; the sixth step is to use the maximum expected enhancement value plus point criterion to obtain new sample points, add them to the sample S, and execute the second step.

[0272] The flowchart of using the spatial optimal model for optimization estimation is as follows Fig. 9 shown.

[0273] 2. Space optimization based on proxy estimation

[0274] Based on the parts quantitative model, the Python language is used to establish the parts quantitative modeling program, and the programmed model post-processing is realized; MATLAB is used to establish the spatial optimal proxy model, and the spatial optimization program of the proxy model is established.

[0275] The spatial optimal model includes a spatial modeling program and a file system. The spatial modeling program includes an agent model building program, a finite element simulation modeling program, a data processing program, and a system scheduling program. The agent model program is built in the MATLAB environment using the m language. It builds an agent model by sampling design points S and response values, solves the agent model, calculates the current maximum expected value, determines the next sample point, and solves the model optimal value when the convergence criterion is met; the finite element simulation modeling program is built in the Python language, and the component quantitative modeling is established according to the corresponding parameters of the design sample points; the data processing program calculates the corresponding objective function value using the finite element simulation model according to the design objective function, including extracting data from the simulation results, data preprocessing, and objective function value calculation. At the same time, the calculation intermediate result file is output, which is written in the m language and Pyth on language, the data processing program is related to the objective function; the scheduling program coordinates each program, controls each software to execute the corresponding command, calls the simulation software for modeling and simulation solution after the sample point is updated, and calls the data processing program to calculate the objective function value after the simulation solution is completed. The agent construction model is called according to the objective function response and sample points, the agent model is updated and the maximum expected enhancement is solved, and the calculation log and report file are output at the same time; the file system refers to the relevant information about agent modeling created during the calculation process, including parameter tables and result files for recording simulation modeling, agent model modeling, objective function parameter files, calculation report files, agent modeling architecture such as Fig. 9 As shown,

[0276] (IV) Wear analysis of the hinged pair of the operating mechanism

[0277] Due to processing conditions or design requirements, the existence of clearance in the articulated pair of the operating mechanism is inevitable. However, the size of the clearance value has a nonlinear effect on the mechanism. If the clearance is too large, the mechanism will have a high delay and a large impact; if the clearance is too small, the mechanism will have a significant friction contact effect and will be easy to get stuck. The friction and wear of the articulated pair of the operating mechanism affects the performance of the mechanism and may cause mechanism failure. Therefore, the wear of the articulated pair is taken as the optimization target, and the articulated fit tolerance (the deviation value of the articulated clearance and the tolerance band) is taken as the design variable to optimize the articulated pair.

[0278] 1) Design variables: In engineering, the clearance of the articulated joint is the difference between the inner diameter of the sleeve and the outer diameter of the pin, which is a relative change value, considering the reference deviation value ε s , with deviation value and gap value as design variables;

[0279] 2) Optimization goal: The wear model calculation formula is:

[0280]

[0281] Where V is the wear depth, s is the relative sliding distance, and F nis the normal contact force, k is the dimensionless wear coefficient, H is the Brinell hardness of the softer material in the contact material, and the contact force is obtained by finite element simulation of the contact history variable normal contact force, and the relative relationship of the coordinate position of the relative sliding pin shaft sleeve mark point during the movement process is calculated;

[0282] 3) Constraints: Interval constraints are imposed on the design variable values. The interval constraint relationship is as follows:

[0283]

[0284] Among them, the unit of constraint condition is μm;

[0285] The GALHS method is used for initial sampling, the initial sample point matrix size is 10×2, and the number of iterations is 20 times; when the gap value is small and the gap value is large, the wear degree is relatively high. When the gap value is around [60,110], the wear degree of the articulated joint is relatively low, and the optimal search points are mostly concentrated in this interval; when the contact material is certain, the factors affecting the wear degree of the articulated pair include the contact force of the articulated pair, the relative slip between the articulated pairs, and the friction contact action time of the articulated pair; when the gap value is small, the contact time of the articulated pair is long and the frequency is high. Long-term contact friction is the factor causing the wear of the articulated pair, and the space between the articulated pin and the sleeve is narrow; during the movement of the pin, After the rebound after the last collision, after a short distance (less than 28) of movement displacement, it contacts and collides with the sleeve again. In addition, the operating mechanism moves at high speed and the impact effect is obvious. The pin and the sleeve are deformed at the same time. When the gap value is extremely small, jamming occurs. The full-time contact and slippage of the jam will greatly promote wear; when the gap value is too large, the impact load of the articulated pair increases during the movement, and the relative slip between the articulated components is large. The larger contact force and the larger relative slip will also lead to a larger amount of wear, and the contact time decreases in a nonlinear relationship with the increase of the gap value. From the wear perspective of the articulated pair, the articulated gap should not be too large or too small when designing the circuit breaker operating mechanism.

Claims

1. A SOM feature quantification agent modeling method for high-voltage power circuit breakers, characterized in that: In the ABAQUS environment, a finite element simulation model of the actuator is established, and the simulation model is modified and verified by the displacement curve. The parts are quantified based on feature modeling. The outline pixels are divided into structural pixels and contour pixels. The constraint relationship of the structural pixels is established and solved according to the geometric relationship. The contour pixels are calculated using the obtained structural pixels. The parts are quantified in combination with the feature parameters. The vector ring equation and the z-direction specification chain of the mechanism are established. The parts quantification of the assembly is completed with the positioning specification, connecting rod length, and initial position as variables. The information model is created according to the relevant information of finite element modeling. The Python language is used for secondary development, and C# is used to develop parts quantification information management software to assist in the management of modeling information. Based on the developed program, the motion deviation of the actuator caused by the articulation clearance and its influence on the articulated pair are studied and analyzed. The genetic space sampling strategy and the expected enhanced sequence addition strategy are adopted. Based on the spatial optimal agent model, the SOM agent modeling model is established in combination with the parts quantification modeling and the programmed post-processing technology. Based on the Archard model, the degree of articulation wear is taken as the target and the articulation fit tolerance is taken as the variable to analyze the SOM. A-Circuit breaker SOM finite element parts quantitative modeling: 1) Establish a feature-based parts quantitative model, establish and solve the constraint relationship for the feature outline construction element, construct the mathematical representation of the contour element, and realize the quantitative construction of the parts geometric solid model; 2) Establish the plane position relationship of the parts assembly, solve the vector ring equation to obtain the position and initial posture of the parts on the two-dimensional projection surface, and combine the z-axis specification chain to complete the parts quantification of the spatial assembly; 3) Create an actuator finite element SOM information model, express the feature information in the form of abstract classes, establish a finite element parts quantitative model of the actuator, realize the reuse of the finite element SOM model, and develop a parts quantitative information management platform; B-SOM proxy modeling and analysis of articulated characteristics: 1) Taking the clearance of the articulated pair as the variable, the deformation and relative displacement of the articulated sleeve and the pin are analyzed to obtain the variation law of the mechanism motion displacement deviation with the clearance specification; 2) Based on the spatial optimal proxy model, the genetic space sampling strategy and the expected enhanced point addition criterion are adopted to establish a finite element proxy modeling program in the ABAQUS environment; 3) Using the established proxy modeling program, with the articulated pair fit tolerance as the design variable and the wear condition during the articulated pair motion as the objective function, it is found that the articulated clearance has a greater impact on the wear, and a clearance that is too large or too small will lead to greater wear.

2. According to claim 1, the SOM feature quantification agent modeling method for high-voltage circuit breaker is characterized in that: Quantification of assembly parts: When creating parts, the initial point is set at the origin, and the relative position relationship between parts is adjusted during assembly. In the assembly module, the relative position and initial motion posture of parts are adjusted using the feature relationship of parts; The assembly parts are associated through coaxial lines, parallelism, relative distance, and relative rotation angle related constraint behaviors to ensure that under given initial conditions, each component is adjusted to the corresponding posture. The coordinate system in the assembly drawing is the global coordinate system. The coordinate system used by each component in modeling is called the local coordinate system. The position of the local coordinate system in the global coordinate system is calculated. The position of some local coordinate systems in the global coordinate system is relatively fixed and directly represented by the relative position vector. The actual linkage mechanism contains secondary rod groups and higher-level rod groups. The quantification of assembly parts needs to be solved. The operating mechanism contains two four-bar linkages, and their positions are solved; The transmission part of the operating mechanism is composed of connecting rods l1, l2, l3, l4, l5, l6, l7 and frames l4, l8. Among them, l1, l2, l3, l4 constitute four-link I, which transmits motion and force from the upper connecting rod to the main shaft, and l5, l6, l7, l8 constitute four-link II, which transmits motion from the main shaft to the large crank arm. The four-link mechanism establishes a closed vector loop. In the formula, For vector The angle with the x-axis, l i For vector Model; When solving, project them onto the coordinate axis respectively and convert them into scalar form equations. Equation 2 is its scalar form: A and D are external hinge points. After their coordinates are given as initial conditions, Equation 2 contains only the design variables. and The equation is solvable; Solve the four-bar linkage I to obtain by Solve the four-bar linkage II with the angle θ with the main axis as the initial condition, obtain the plane motion position of the transmission mechanism under the given initial condition, and establish the specification driving relationship in the plane normal direction according to the assembly relationship; The operating mechanism includes three phases, each phase includes a pair of main shaft crank arms, lower connecting rods, and large crank arms. The large crank arms and lower connecting rods are sub-assembled separately. The large crank arm sub-assembly parameter is D dgb1 With D dgb2 , the lower connecting rod assembly parameter is D xlg With D dgb2 , obtained from the assembly structure: D xlg =D dgbl -T dgb -T xlg Formula 3 The positioning specifications of the components are exported specifications, and the component positioning specifications are established by taking the connecting rod coordinate origin as the global coordinate origin.

3. According to claim 1, the SOM feature quantification agent modeling method for high-voltage power circuit breaker is characterized in that: Finite element SOM model information structure: In terms of overall structure, it is divided into CAD information and CAE information. CAD information includes component information and assembly information. CAE information includes material information, analysis step information, interaction information, boundary load information, mesh information, and task information. The finite element SOM model information is taken as the set M, and its composition is expressed by formula 4: M = {I Part , I Assem , I Mat , I Bcload , I Mesh , I Int , I Step , I Job} Formula 4 Where: I part Part information refers to the geometric topological information of the part entity and the auxiliary information used for finite element modeling. The geometric topological information includes the point, line, surface, and body information of the part entity. The auxiliary information includes reference points, lines, surfaces, and feature sets. The part information is a description of the part features and provides selection data for adding assembly contacts and meshing the model. L assem Assembly information refers to the relevant geometric information and auxiliary information contained in the assembly. In addition to the instantiation of the part instance and the corresponding inherited part information, the assembly information also includes the reference points, lines, surfaces and feature sets defined under the assembly global coordinates; I mat Material information refers to the relevant material properties assigned to the part, including material density, elastic modulus, Poisson's ratio, plastic material model and related parameters, material direction information. The material model also includes the cross-sectional properties of the part. I bcload Boundary load information describes the load conditions of the model and the boundary conditions it depends on. Load conditions include force, moment loads, and the relationship between load and time. Boundary conditions refer to the constraints on the parts in the model. Boundary load information depends on the auxiliary information provided by assembly information. I mesh Mesh information, including data related to meshing of part instances; I int In order to exchange information, the geometric entities of the hinge pin and the sleeve are established, and the whole is used as a hinge pair for mechanism analysis and solution; I step Analysis step information, which refers to the type of model analysis problem, the length of time used in the analysis step, the time increment, the mass magnification factor, and also includes historical output variables reflecting the analysis process and result data, contact force and field output variables, stress strain and output frequency; I job Task information refers to the type of analysis task established and related solution control parameters. Analysis types include regular analysis and restart. Related solution control parameters refer to the number of solution cores for parallel computing, the number of computational domains, and the memory size for computing solutions. SOM object finite element SOM information model: The information model is divided into a set of operations and parameters, and any subset in the information model is represented as: The finite element SOM information model is described as: Subset I in the finite element SOM information model i Defined as class (Class) Cls, the operation in the information model is defined as function (Function) Fcn, and the parameter in the information model is defined as attribute (Attribute) Atr.

4. According to claim 1, the SOM feature quantification agent modeling method for high-voltage power circuit breaker is characterized in that: Modeling process based on finite element SOM information model: Create a design object based on the finite element SOM information model for analysis and solution. The specific steps are as follows: Step 1: Call the model layer script, read the information in the information library, and initialize the model design related parameters, including component specifications, assembly parameters, material parameters, mesh parameters, boundary load parameters, and interaction parameters; Step 2: Substantiate the part class, call the geometric modeling function of each component, and establish the geometric model of the component; Step 3: Instantiate the material class, call the public function Material&assign() to create cross-section properties for the model, add materials, and assign materials to the parts; Step 4: Instantiate the assembly class, call the Assem() function, calculate the assembly position of each component, establish auxiliary reference points, and assemble the components; Step 5: Call Rp_Bulid() function to add auxiliary reference points to the model and modify the reference point properties; Step 6: Call the Set&Surface_Bulid() function to create auxiliary surfaces and collections based on the associated characteristics and geometric features of the model components; Step 7: Instantiate the analysis step class, call the Step() function, create an analysis step class for the model, and set the analysis type and analysis duration; Step 8: Instantiate the interaction class, call the Interaction_Bulid() function, establish contact properties and hinge properties, and establish interactive connections for components; Step 9: Call the Output() function to set the output variable, output domain, and output frequency for the model; Step 10: Instantiate the load boundary class and call the Load() function to add boundary conditions and input loads to the simulation model, including loading the input load curve and constraining the degrees of freedom; Step 11: Instantiate the task class, call the Job() function, and set the solution method, number of processors, and memory usage for the model; Step 12: Analyze the calculation results. If they are not ideal, use the Import() function to import the finite element SOM model, and then call the model layer script to set the parameters. Step 13: Call ConstrainDel() and PartDel() functions to delete the component and the interactive connections and features associated with the modified component, and call the component modeling function to model.

5. According to claim 1, the SOM feature quantification agent modeling method for high-voltage power circuit breaker is characterized in that: Analysis of the kinematic posture of the articulated joint: During the motion of the pin and the sleeve in the articulated joint, there are several different relationships between their relative positions and the force conditions. According to the relationships they show, the kinematic postures of the articulated joint are classified into five types: Posture I: When the articulated joint is in posture I, the sleeve connected to the middle link, called the outer sleeve, moves downward under the force of the middle link. At this time, the pin and the sleeve connected to the main shaft become the inner sleeve. Due to the existence of the gap, there is no contact or collision. Posture II: The outer sleeve contacts and collides with the pin, driving the pin to move, and the pin contacts and collides with the inner sleeve, driving the spindle to rotate; Posture III: The pin is separated from the inner and outer sleeves and is in a free state. The transmission of the mechanism force is interrupted, and the downstream mechanism consumes its own energy to move. Posture IV: The characteristics of the mechanism's own structure and the drive input load, the inner sleeve drives the pin, the pin contacts and collides with the outer sleeve, and works on the upstream mechanism; Posture V: During the movement of the mechanism, the sleeve is deformed due to force, causing the pin to get stuck; During the movement of the mechanism, the articulated pair switches between four postures: I, II, III, and IV. Posture V only appears when the gap is small and the articulated pair is stuck. The flexible sleeve of the mechanism articulated joint deforms during movement, and the circular inner diameter section becomes an ellipse. When the clearance value of the articulated joint is too small, the short diameter of the ellipse is smaller than the outer diameter of the pin shaft after the sleeve is deformed, and the articulated joint becomes stuck.

6. The SOM feature quantification agent modeling method for high-voltage circuit breakers according to claim 1 is characterized in that: Genetic space sampling strategy: sample points are taken in the design interval to form a sample point set S. The response data value Y is obtained through response solution. According to the sample data [S|Y], a response surface model is constructed based on the n of each sample point. v The minimum sphere radius without intersection in the design space is used as the measurement standard. The larger the minimum sphere radius, the better the spatial distribution characteristics of the sample points. p The smaller the value, the greater the distance between the design sample points. Its mathematical expression is: In the formula, p is a positive integer, n s is the number of sample points, d ij is the distance between two points composed of sample points, and the distance between any two points is expressed by formula 7: The LHS is improved and the algorithm flow is as follows: CreateLHS(); Create an initial population FitnessLHS(); Population individual evaluation _stop = FALSE; while(not_stop) SelectLHS0; Select operation CrossLHS); Cross operation MutateLHS); mutation operation If convergence condition is met then_stop=TURE; end First, encode the data to describe the LHS sampling samples in the form of a matrix. In the matrix L of size m×n, each row represents a set of sample points, and each column contains a set of 1-m integer sequences: In the formula, x j is the jth group of sample points; the operation of matrix L initializes the population to generate N pop samples, N pop is an even number, the selection operation sorts the individuals in the population according to their fitness values ​​and selects the top N pop / 2 best samples; The crossover operation uses the selected N pop / 2 The number of generated samples is N pop The new population is divided into S1 = [1, (N pop / 2+1)] and S2=[N pop / 2,N pop ] two intervals, S1 interval is based on the optimal individual, and [2, N pop / 2] randomly exchange a column to generate new individuals, and the S2 interval is [2, N pop / 2], randomly exchange a column with the best individual to generate new individuals, and the combination of S1 and S2 constitutes a new population; The mutation operation acts on all individuals in the new population except the best individual, and generates a random number in the interval [0, 1] for each column. If the random number is less than the given mutation rate value p mut , then randomly swap two elements in the column.

7. The SOM feature quantification agent modeling method for high-voltage circuit breakers according to claim 1 is characterized in that: SOM surrogate model: using known sample data [X m×n |Y m×p ] Establish an approximate mathematical expression relationship between the input variable x and the output response y, assuming that the design point set S = [s1, s2, ..., s m ] T ,s i ∈R n The system outputs a one-dimensional (q=1) response value Y=[y1, y2, ..., y m ] T ,y i ∈R q , expressed as a regression model and random function: The regression model uses a linear regression model consisting of k basis functions: Assuming the mean of the random function z is 0, the covariance between z(w) and z(x) is: E[z(w)z(x)] = σ 2 R(θ, w, x) Equation 11 In the formula, σ 2 is the variance of the random function z, R(θ, w, x) is the association model, and θ is the association model parameter; Therefore, for the design sample point set S, its design matrix F is expressed as: The correlation matrix R of the random function is: The correlation vector r between the unknown point x and the design sample point set S is: r(x) = [R(θ, s1, x) … R(θ, s m , x)] T Equation 14 The predicted response value at x is expressed as: The prediction error is: Where Z = [z1,…,z m ] T The error of the design sample point set is kept unbiased by the prediction, F T c(x)-f(x)=0, the average variance of the prediction error for: beg The Lagrange equation for the minimum value of c while maintaining the unbiased property is: L(c,λ)=σ 2 (1+c T Rc-2c T r)-l T (F T (cf) L′(c, λ) = 2σ 2 (Rc - r) - λF Equation 18 According to the necessary optimization conditions and the no-deviation constraint relationship, we can get: Solving equation 19 yields: Matrix R and R -1 is a symmetric matrix, then: The least squares solution of the regression model with respect to R is: β * = (F T R -1 F)F T R -1 Y formula 22 Combining equation 21 and equation 22, we get: Where, Rγ*=Y-Fβ*; The model prediction error is: In the formula, u and σ 2 are process quantities, u=F T R -1 r-f Construct a SOM agent model.

8. The SOM feature quantification agent modeling method for high-voltage circuit breakers according to claim 1 is characterized in that: Sequence optimization and point addition criteria: by analyzing the corresponding response values ​​on the basis of the initial samples, new sample points are added to improve the accuracy of the model. In the optimization process, not only the minimum value in the prediction model is considered, but also the sparsity of the model sample points. The sparser the sampling points, the lower the credibility of the prediction model. For a design point x, its response value y(x) cannot be known before performing expensive calculations to predict its mean and mean square error σ 2 , the uncertainty of y(x) is taken as the uncertainty about the variable Y, and the mean and mean square error are and σ 2 The normal distribution of , the point-adding criterion weighs all possible points that can obtain the optimal value, and its expression is: and Φ(·) are the density function and distribution function of the standard normal distribution, respectively. The first term is the difference between the current optimal value and the predicted value multiplied by the increased probability, and the second term is the standard deviation multiplied by the probability density. The larger the difference between the current optimal value and the predicted value and the smaller the prediction variance, the larger the first term is, that is, looking for a point where the predicted value is significantly smaller than the current optimal value. When the prediction variance is larger, the larger the second term is, but it does not deviate from the current optimal point due to the probability density constraint, that is, looking for a point with large uncertainty, weighing the optimal value and uncertainty of the prediction model, and selecting the next sample point.

9. The SOM feature quantification agent modeling method for high-voltage circuit breakers according to claim 1 is characterized in that: Optimization estimation based on spatial optimal model: including single-step optimization and sequence optimization. Single-step optimization constructs a proxy model after sampling once, checks the accuracy of the proxy model, and optimizes based on the proxy model when it meets the requirements. Sequence optimization increases samples by considering the predicted value and predicted variance, continuously updates the model, improves the model accuracy, and finds the optimal value for the proxy model. When the convergence criterion is met, the iteration stops. The steps are as follows: Step 1: Adopt the genetic space sampling strategy to construct the initial design sample point S with a dimension of m; Step 2: According to the design objective function value, solve the target response value Y corresponding to the design sample point S; Step 3: Use sample design points S and response values ​​Y to construct a spatial optimal proxy model; Step 4: Use the optimization algorithm to solve the proxy model and obtain the current maximum expected enhancement value; Step 5: Determine whether the iteration meets the convergence criteria. If so, stop the iteration and use the optimization algorithm to solve the optimal value of the current proxy model as the optimization calculation result. Otherwise, execute step 6; Step 6: Use the maximum expected enhancement value addition criterion to obtain new sample points, add them to sample S, and execute the second step.

10. The SOM feature quantification agent modeling method for high-voltage circuit breakers according to claim 1 is characterized in that: Wear analysis of the articulated pair of the operating mechanism: Taking the wear of the articulated pair as the optimization target and the articulated fit tolerance (the deviation value of the articulated clearance and tolerance band) as the design variable, the articulated pair is optimized; 1) Design variables: In engineering, the clearance of the articulated joint is the difference between the inner diameter of the sleeve and the outer diameter of the pin, which is a relative change value, considering the reference deviation value ε s , with deviation value and gap value as design variables; 2) Optimization goal: The wear model calculation formula is: Where V is the wear depth, s is the relative sliding distance, and F n is the normal contact force, k is the dimensionless wear coefficient, H is the Brinell hardness of the softer material in the contact material, and the contact force is obtained by finite element simulation of the contact history variable normal contact force, and the relative relationship of the coordinate position of the relative sliding pin shaft sleeve mark point during the movement process is calculated; 3) Constraints: Interval constraints are imposed on the design variable values. The interval constraint relationship is as follows: Among them, the unit of constraint condition is μm; The GALHS method was used for initial sampling, the initial sample point matrix size was 10 × 2, and the number of iterations was 20; When the gap value is small or large, the wear degree is relatively high. When the gap value is around [60,110], the wear degree of the articulation is relatively low, and most of the optimal search points are concentrated in this interval.

Images