A tolerance allocation multi-objective model parameter fusion optimization method and system

Abstract

The present invention provides a tolerance allocation multi-objective model parameter fusion optimization method and system, including: S1: constructing a geometric error spinor model of basic error source feature changes; S2: constructing a key feature posture change model considering matching errors; S3: constructing a single assembly error accumulation model and a multi-assembly coordination error model; S4: constructing multiple single-objective tolerance optimization models for improving performance according to product requirements; S5: establishing a multi-model weight parameter calculation method; S6: constructing a tolerance allocation multi-objective optimization model; S7: iteratively solving the tolerance allocation multi-objective optimization model. The present invention can greatly reduce the cost and time of repair and maintenance, and at the same time effectively avoids the problem of optimization deviating from the demand direction due to the inconsistent connotation of each target model data and the size of the numerical value, so that the optimized tolerance data meets the expected optimization effect.

Description

[Technical field]

[0001] The present invention relates to the field of digital assembly coordination technology of mechanical products in the field of mechanical manufacturing, and in particular to a tolerance allocation multi-objective model parameter fusion optimization method and system. [Background technology]

[0002] As an important link in the manufacturing process of complex mechanical products, tolerance allocation directly affects the service performance and economic benefits of products. Since the 1960s, how to use optimization technology to improve the tolerance allocation scheme of components has always been a research hotspot in the field of tolerance design. Nowadays, with the continuous improvement of the digitalization and intelligence of modern manufacturing technology, the introduction of technologies such as digital twins and intelligent algorithms has brought more efficient production efficiency and high-quality manufacturing performance, while also putting forward higher quality, lower cost, and more targeted performance requirements for product production. The traditional tolerance allocation scheme based only on engineering experience can no longer meet current production needs. The existing tolerance allocation methods are mostly centered on the optimization of manufacturing cost and quality loss, but the reduction of manufacturing cost in the optimization process is often accompanied by an increase in quality loss. At the same time, the various tolerance optimization objectives are difficult to coordinate with each other, which brings certain difficulties to the optimization of tolerance allocation schemes.

[0003] In view of the blindness of traditional tolerance allocation technology, in order to build a real and effective tolerance optimization model, domestic and foreign scholars have carried out a lot of research on the precise construction of models and the improvement of algorithms, mainly including the construction and fusion of product performance models related to tolerance and the use of intelligent algorithms to solve the tolerance multi-objective optimization model. While these works have achieved reliable tolerance allocation solutions, they also have some shortcomings: 1) The fusion of various optimization sub-goals in tolerance allocation lacks data support and demand guidance, and often uses a simple weighted summation method, which fails to fully consider the imbalance of data size and inconsistent connotation of each model, resulting in a large deviation between the fusion model and the actual model, causing the subsequent tolerance optimization direction to deviate from the actual production needs; 2) The key links of the assembly process can be identified to a certain extent through the tolerance optimization results, and a good assembly process can also effectively reduce the pressure of tolerance optimization, but the effective correlation between tolerance optimization design and the improvement of assembly process solutions needs to be strengthened.

[0004] Considering the high performance requirements of modern products for assembly quality, it is difficult to meet the production acceptance standards by using tolerance optimization methods, so the repair method is often used for assembly. Existing optimization allocation models rarely take the reduction of repair costs as an optimization goal. This will make it difficult to avoid the problems of increased assembly economic costs, extended production time, and decreased product service mechanical properties caused by the repair link. The research on repair solutions in the existing technology is mostly concentrated in the field of assembly process design, and there is a lack of a coordinated mechanism for reducing repair costs and improving assembly process solutions driven by tolerance optimization of assembly dimension chains.

[0005] In summary, certain progress has been made in the construction and solution of tolerance optimization models. However, there are still some deficiencies in the fusion of multi-objective models and the guarantee of optimization direction for product performance requirements.

[0006] Therefore, it is necessary to study a tolerance allocation multi-objective model parameter fusion optimization method and system to address the shortcomings of the existing technology in order to solve or alleviate one or more of the above problems. [Summary of the invention]

[0007] In view of this, the present invention provides a tolerance allocation multi-objective model parameter fusion optimization method and system, which mainly include three parts: the construction of an assembly error accumulation calculation model and a multi-assembly coordination error model, the fusion construction of a multi-objective tolerance optimization model and the solution of an intelligent algorithm, which can achieve: 1) the construction of an actual change relationship model between different features on the same assembly part and a matching error change relationship model between two assembly parts, thereby obtaining an error accumulation calculation model for the entire assembly and a coordination error model between multiple assemblies; 2) the precise construction of a product repair cost model based on the component ring tolerance, and the introduction of a tolerance optimization model to greatly reduce the repair cost and time; 3) while accurately and efficiently establishing a single-objective tolerance optimization model, a multi-objective fusion weight calculation method that integrates performance requirements and empirical data is proposed to achieve efficient and accurate construction of a multi-objective model, effectively avoiding the problem of optimization deviating from the demand direction due to the inconsistent data connotations of each target model and the size of the numerical value, so that the optimized tolerance data meets the expected optimization effect.

[0008] In one aspect, the present invention provides a tolerance allocation multi-objective model parameter fusion optimization method, which is used to improve the performance requirements and assembly coordination process of segmented wing structure parts during the production and manufacturing process. The fusion optimization method includes the following steps:

[0009] S1: According to the structural characteristics and assembly process of the parts to be produced, a geometric error screw model of the basic error source characteristic variation is constructed;

[0010] S2: Construct a key feature pose change model considering the matching error based on the geometric error screw model;

[0011] S3: Based on the geometric error screw model and the key feature pose change model, a single assembly error accumulation model and a multi-assembly coordination error model are constructed;

[0012] S4: Based on the product requirements of the parts to be produced, the single assembly error accumulation model and the multi-assembly coordination error model, multiple single-objective tolerance optimization models are constructed to improve performance according to product requirements;

[0013] S5: Establish a multi-model weight parameter calculation method that integrates performance requirements and empirical data;

[0014] S6: Through the multi-model weight parameter calculation method, multiple single-objective tolerance optimization models in S4 are integrated to construct a tolerance allocation multi-objective optimization model that characterizes the comprehensive performance of the product;

[0015] S7: Accelerated particle swarm optimization is used to iteratively solve the tolerance allocation multi-objective optimization model.

[0016] According to the above aspects and any possible implementation, an implementation is further provided, wherein S1 specifically includes:

[0017] S11: Analyze the assembly product structure and assembly performance requirements, obtain specific product assembly accuracy requirements, and convert the product assembly accuracy requirements into the form of feature points and surfaces according to the assembly accuracy characteristics, and obtain the assembly dimension chain accuracy index requirements expressed by the screw model;

[0018] S12: Taking the assembly dimension chain accuracy index requirement as the source and the assembly datum as the end point, find the error transmission path, identify each dimension link on the transmission path as the error source, and obtain the key measurement points and geometric feature information of each error source;

[0019] S13: The position change data of the geometric features of the assembly are obtained according to the key measurement points and geometric feature information of each error source, and the position change data are converted into an error spinor model in a matrix format using kinematic theory and small displacement spinor method, and the error spinor model of the position change is mapped to the tolerance domain.

[0020] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein S2 specifically includes:

[0021] S21: construct the global coordinate system of the assembly and the local coordinate systems of each component of the assembly, and set the homogeneous transformation matrix of each local coordinate system relative to the global coordinate system, and obtain the key features on each assembly part;

[0022] S22: The representation relationship of the key features of each part in its own local coordinate system is obtained through the error screw model in S1, and the homogeneous coordinate transformation matrix is ​​obtained according to the transformation relationship between each local coordinate system and the global coordinate system, and the fitting error between each part is calculated in the global coordinate system;

[0023] S23: According to the matching errors between the parts, the size and tolerance data of the matching area of ​​the parts are obtained, and based on the size and tolerance data, the part posture change data caused by the matching between the parts are calculated;

[0024] S24: Analyze and construct the transformation relationship matrix of two key features on the same part and the part posture change matrix caused by the matching errors of the two parts, and then multiply the two to obtain the assembly cumulative error matrix after the parts are matched;

[0025] S25: Convert the assembly cumulative error matrix into a vector form and ignore the influence of high-order micro-variations to obtain the component form of the cumulative error of part features in each direction.

[0026] According to the above aspects and any possible implementation, an implementation is further provided, wherein S3 specifically includes:

[0027] S31: Analyze the range and tolerance domain of geometric deviation in S1, determine the error geometric model of the assembled parts equivalent to the coordinate system, define the form of matching error transmission between the parts on the assembly, and then determine the posture change of the subsequent parts in the assembly transmission;

[0028] S32: In combination with the assembly process of each component part of the assembly, the error transfer model is performed for each component part, and according to the calculation method in S23, the error of the previous part is introduced into the next part to determine the final assembly error of the entire assembly and the single assembly error accumulation model;

[0029] S33: Obtain the error transmission direction and transmission path of the assembly, and use the final assembly error of the entire assembly calculated in S32 as data support, and construct an assembly coordination error transmission accumulation model between assemblies through the absolute value of the difference between the final cumulative error values ​​of the two assemblies.

[0030] According to the above aspects and any possible implementation, an implementation is further provided, wherein S3 further includes:

[0031] S34: The Monte Carlo method is used to analyze the assembly coordination error dimension chain, obtain the accumulation law of coordination error and calculate the error mean;

[0032] S35: Analyze the assembly errors of each component and the coordination error data of the assembly obtained in S33, identify the key assembly elements and key assembly processes in the assembly process of the mechanical product, and optimize the assembly process links that have a greater impact on the coordination error of the assembly based on the accumulation law of the coordination error and the statistical results of the error mean.

[0033] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein S4 specifically includes:

[0034] S41: Constructing a manufacturing cost-tolerance model: Analyze the tolerance-cost data generated in the manufacturing process and select a suitable tolerance cost model for fitting to obtain an accurately established tolerance-cost function; secondly, sort out the tolerance-cost functions of each component link in the assembly dimension chain, calculate the processing cost of each tolerance and sum them up to obtain the total processing cost of the product assembly tolerance, and complete the construction of the manufacturing cost-tolerance model;

[0035] S42: Constructing a quality loss-tolerance model: The deviation between the actual performance and the ideal performance of the product is identified as quality loss; the tolerance of each component ring is used to represent the product assembly accuracy index, and the ideal performance value of the assembly and the actual performance calculated by the component ring tolerance are calculated respectively. The Taguchi quality loss cost function is introduced to complete the construction of the quality loss-tolerance model;

[0036] S43: Construct a repair cost-tolerance model: verify whether the current assembly accuracy can meet the performance requirements and determine whether repair work is needed; then calculate the expected value of the repair quantity based on the tolerance data of each component ring, and combine the repair economic cost, repair time and repair area size data generated in the mechanical product manufacturing process to obtain the assembly repair cost function represented by the tolerance, and complete the construction of the repair cost-tolerance model.

[0037] According to the above aspects and any possible implementation, an implementation is further provided, wherein the multi-model weight parameter calculation method in S5 is specifically as follows:

[0038] S51: According to the factory MBD digital model or the tolerance data set by the engineering staff’s production experience, the dimension values, tolerance values ​​and processing feature information of each component ring of the assembly are obtained, and the specific values ​​of the manufacturing cost function model, quality loss function model and repair cost model under the current tolerance data are preliminarily calculated, which are expressed as C M (T m ), C Q (T m ) and C R (T m ), and the weight parameters of each model are set to be α, β and γ respectively;

[0039] S52: Combined with factory production and manufacturing analysis, determine the key links of the assembly process and the performance requirements of product production, obtain the key target model that has a significant impact on the required performance improvement in the tolerance optimization process, determine the importance of each target model to the final target performance improvement, and obtain the initial ratio of the weight parameter of each target model to the product of the model value, that is, α·C M (T m ), β·C Q (T m ) and γ·C R (Tm )

[0040] S53: Decomposing the above initial proportional equation into multiple inequivalent equations, each weight parameter in each equation is an unknown quantity, and the remaining values ​​are all known, and at the same time, setting α+β+γ=1, combining multiple inequivalent equations, a set of linear equations in which all unknown quantities are weight parameters is obtained;

[0041] S54: Solve the linear equations to obtain specific values ​​of each weight parameter, and complete the construction of a multi-model weight parameter calculation method that integrates performance requirements and empirical data, where the linear equations are:

[0042]

[0043] In the formula, C M (T m ) is the manufacturing cost value under empirical data, C Q (T m ) is the mass loss value under empirical data, C R (T) is the value of the repair cost loss under empirical data, c1 and c2 are constants that can be obtained from the proportional relationship, and α, β, and γ are the weight parameters of each model to be determined.

[0044] According to the above aspects and any possible implementation, an implementation is further provided, wherein the multi-objective optimization model for tolerance allocation that characterizes the comprehensive performance of the product is integrated and constructed in S6 and specifically includes:

[0045] S61: Obtain the manufacturing cost C established in S5 M , quality loss C Q and Repair Cost Model C R The function model is uniformly adjusted to a tolerance T i is the format of the independent variable.

[0046] S62: Obtain the numerical data of the weight parameters of each model obtained in the multi-model weight parameter calculation method, and use this to weight the sum of each target model to obtain the comprehensive performance index of the product Transform tolerance optimization from a multi-objective optimization problem into a single-objective function optimization process;

[0047] S63: According to production constraints such as tolerance processing capability, corresponding constraints are imposed on the optimization model, and the models in S61 are integrated to obtain a mathematical model including the product comprehensive performance index function, each target optimization model function, tolerance range constraints and weight parameters, thus completing the fusion construction of the tolerance allocation multi-objective optimization model.

[0048] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein S7 specifically includes:

[0049] S71: Define the objective function file and obtain the product comprehensive performance index model R (T i ), Manufacturing Cost Model C M (T i ), quality loss model C Q (T i ) and repair cost model C R (T i ) and then convert the above model into a format that can be recognized by the algorithm program, import it into the objective function file, and complete the construction of the objective model module;

[0050] S72: define a constraint function file, obtain the constraint space data set in S5, and convert the constraint space data into the form of inequalities and further write them into a format recognizable by the algorithm program and import them into the constraint function file one by one to complete the construction of the constraint application module;

[0051] S73: Define and write the accelerated example swarm algorithm program, and at the same time build multiple sub-function files that need to be called for algorithm operation, and then call the target model module, constraint application module and sub-function module in the algorithm implementation program. At the same time, define the population size, number of iterations, and particle random attenuation factor parameters of the accelerated example swarm, and run the accelerated particle swarm algorithm program to obtain the optimized tolerance allocation plan.

[0052] According to the aspects described above and any possible implementation method, a tolerance allocation multi-objective model parameter fusion optimization system is further provided, wherein the optimization system comprises a memory and a processor, wherein the memory is connected to the processor, and the fusion optimization method is stored in the memory.

[0053] Compared with the prior art, the present invention can achieve the following technical effects:

[0054] 1) By analyzing the error coupling relationship between the matching features of complex assemblies and the matching relationship between the key features of each component, the error accumulation calculation model of a single assembly and the coordination error model between multiple assemblies can be obtained;

[0055] 2) By constructing a repair cost-tolerance optimization model, the repair cost generated by the repair process can be effectively controlled according to engineering requirements, minimizing the negative impact of the repair process on product assembly production performance, and effectively avoiding problems such as increased economic costs, reduced production efficiency, and labor safety hazards caused by the repair process;

[0056] 3) By adopting a multi-model weight parameter calculation method that integrates performance requirements and empirical data, the problem of difficulty in determining weight parameters due to inconsistent data content and unbalanced values ​​of each model can be effectively solved. At the same time, relying on the weighted summation of each model based on the obtained weight, the optimization space allocation can be more reasonable, and the model optimization direction is consistent with the expected optimization target and performance requirements, effectively ensuring the assembly coordination quality of mechanical products.

[0057] Of course, any product implementing the present invention does not necessarily need to achieve all of the above-mentioned technical effects at the same time.

Brief Description of the Drawings

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0059] Figure 1 It is a parameter fusion optimization flow chart of tolerance allocation multi-objective model for performance requirements and assembly process improvement provided by an embodiment of the present invention;

[0060] Figure 2 It is a schematic diagram of the process of constructing the assembly coordination error accumulation transfer model proposed by the present invention;

[0061] Figure 3 is a schematic diagram of a spacecraft wing assembly structure provided by an embodiment of the present invention;

[0062] Figure 4 It is a schematic diagram of a spacecraft wing docking structure provided by an embodiment of the present invention;

[0063] Figure 5 It is the assembly coordination error calculation method proposed by the present invention;

[0064] Figure 6 is a schematic diagram of the error variation of a wing surface in the normal direction provided by an embodiment of the present invention;

[0065] Figure 7 It is a wing step difference assembly dimension chain diagram provided by one embodiment of the present invention;

[0066] Figure 8 It is a flow chart of a multi-model fusion weight calculation method for fusion performance requirements and empirical data proposed in the present invention;

[0067] Fig. 9 It is a schematic diagram of wing shape assembly using shape cardboard provided by one embodiment of the present invention;

[0068] Fig.10 It is a comparison diagram of various model indicators before and after optimization under the initial assembly process scheme provided by an embodiment of the present invention;

[0069] Fig.11 It is a comparison chart of various model indicators before and after optimization under an improved assembly process scheme provided by an embodiment of the present invention. [Specific implementation method]

[0070] In order to better understand the technical solution of the present invention, the embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0071] It should be clear that the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0072] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "said" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.

[0073] The present invention provides a tolerance allocation multi-objective model parameter fusion optimization method and system, and further provides a tolerance allocation multi-objective model parameter fusion optimization method oriented to performance requirements and assembly process improvement, such as Figure 1 As shown, the fusion optimization method includes the following steps:

[0074] S1: Construct a geometric error spinor model with characteristic changes of basic error sources;

[0075] S2: Construct a key feature pose change model considering matching errors;

[0076] S3: Construct single assembly error accumulation and multi-assembly coordination error model;

[0077] S4: Build multiple single-objective tolerance optimization models for product performance improvement needs;

[0078] S5: Establish a multi-model weight parameter calculation method that integrates performance requirements and empirical data;

[0079] S6: Fusion constructs a tolerance allocation multi-objective optimization model that characterizes the comprehensive performance of the product;

[0080] S7: Accelerated particle swarm optimization is used to iteratively solve the tolerance allocation multi-objective optimization model.

[0081] The S1 specifically includes:

[0082] S11: Analyze the assembly product structure and assembly performance requirements, obtain specific product assembly accuracy requirements, and convert the accuracy requirements into feature points and surfaces according to the assembly accuracy characteristics. Obtain the assembly dimension chain accuracy index requirements that can be represented by the screw model.

[0083] S12: Find the error transmission path with the assembly accuracy requirement as the source and the assembly reference as the end point. Identify each dimension link on the transmission path as the error source, and obtain the key measurement points and geometric feature information of each error source.

[0084] S13: Obtain the position change data of the geometric features of the assembly, and use kinematic theory and small displacement screw method to convert the position change data into a matrix format error screw model. And map the position change error screw model to the tolerance domain. The relationship between the error screw model of typical feature points and surfaces and the constraint inequality is described as follows:

[0085]

[0086]

[0087] In the formula, α, β, γ, u, v, and w represent the changes of the feature in the moving direction and rotation direction of each coordinate axis respectively; t is the tolerance range of the feature size; and L is the size of the surface feature in each direction.

[0088] The S2 specifically includes:

[0089] S21: Construct the global coordinate system of the assembly O A The local coordinate systems O1, O2, ..., O n , and set For each local coordinate system O i Homogeneous transformation matrix relative to the global coordinate system. At the same time, the key features on each assembly part are obtained, and the error spinor model in S1 is used to obtain the representation relationship of the key features of each part in its respective local coordinate system, and the homogeneous coordinate transformation matrix is ​​obtained according to the transformation relationship between each local coordinate system and the global coordinate system, and the matching error between each part is calculated in the global coordinate system. The calculation formula is:

[0090]

[0091] In the formula, F 1i 、F 2i They are the key features of the matching on parts Ⅰ and Ⅱ respectively; Key feature F on part I 1i Error relative to the global coordinate system; is the key feature F on part II 2iThe error is relative to the global coordinate system.

[0092] S22: Obtain the size and tolerance data of the area where the parts cooperate with each other, and based on the size and tolerance data, calculate the part posture change data caused by the cooperation between the parts. The calculation formula is:

[0093]

[0094] Where, (0,0,0,Δh x1 +Δh x2 ,Δh y1 +Δh y2 ,Δh z1 +Δh z2 ) T The size variation range of the mutual matching area between parts I and II; (α1', β1', γ1', u1', v1', w1') T The key feature F 1i Errors relative to the global coordinate system; (α2', β2', γ2', u2', v2', w2') T The key feature F 2i The error is relative to the global coordinate system.

[0095] S23: Analyze and construct the transformation relationship matrix of two key features on the same part and the part posture change matrix caused by the matching errors of the two parts, and then multiply the two to obtain the assembly cumulative error matrix after the parts are matched. Finally, the assembly cumulative error matrix is ​​converted into a vector form and the influence of high-order micro changes is ignored to obtain the component form of the cumulative error of the part feature changes in each direction. The calculation formula is:

[0096]

[0097] In the formula, It is the matrix form of the cumulative error of part assembly; is the key feature F on part II 2i and F 2j The transformation matrix between It is the matrix form of the changes in part posture caused by the cooperation between parts.

[0098] like Figure 2 As shown, as mentioned above, S3 specifically includes

[0099] S31: Analyze the range and tolerance domain of geometric deviation in S1, determine the error geometric model of the assembled parts equivalent to the coordinate system, define the form of matching error transmission between the parts on the assembly, and then determine the posture changes of subsequent parts in assembly transfer.

[0100] S32: Combined with the assembly process of each component part of the assembly, the error transfer model is built for each component part, and according to the calculation method in S23, the error of the previous part is introduced into the next part, and then the final assembly error of the entire assembly is determined. The calculation formula is:

[0101]

[0102] In the formula, is the key feature F on part k ki and F kj The transformation matrix between (k-1)k The position change value of part k caused by the cooperation between part k-1 and part k.

[0103] S33: Obtain the error transmission direction and transmission path of the assembly, and use the assembly error of the single component calculated in S32 as data support to build an assembly coordination error transmission accumulation model between assemblies. The final cumulative errors of the two assemblies A and B calculated by S32 are δ An , δ Bn , subtract the two and take the absolute value to obtain the assembly coordination error of the two, and complete the construction of the multi-assembly coordination error model. The formula is:

[0104] Σ A,B =|δ An -δ Bn |

[0105] Furthermore, the Monte Carlo method is used to analyze the assembly coordination error dimension chain, obtain the accumulation law of coordination error and calculate the error mean. The generation formula is as follows:

[0106]

[0107] Where ξ is a random number that follows a normal distribution within (0, 1); U1 and U2 are independent random numbers that follow a uniform distribution within the range (0, 1); μ and σ are the mean and standard deviation of the assembly error; T U , T L are the upper and lower limits of the assembly error variation range; Z is the set standardized normal number, usually Z=3; X is the standard deviation of N(μ,σ 2 )Normal distributed random numbers.

[0108] S34: Analyze the assembly errors of each component and the coordination error data of the assembly obtained in S33, identify the key assembly elements and key assembly processes in the assembly process of the mechanical product, and determine and optimize the assembly process links that have a greater impact on the coordination error of the assembly.

[0109] In the step of constructing multiple single-objective tolerance optimization models for product performance improvement requirements in step 4, the actual production requirements of mechanical products are considered, and a manufacturing cost-tolerance model, a quality loss-tolerance model and a repair cost-tolerance model are established respectively to complete a preliminary quantitative analysis of assembly performance requirements. Step 4 specifically includes:

[0110] S41: Constructing a manufacturing cost-tolerance model: Analyze the tolerance-cost data generated in the manufacturing process and select a suitable tolerance cost model for fitting to obtain an accurately established tolerance-cost function; secondly, sort out the tolerance-cost functions of each component link in the assembly dimension chain, calculate the processing cost of each tolerance and sum them up to obtain the total processing cost of the product assembly tolerance, and complete the construction of the manufacturing cost-tolerance model. The overall processing cost of the assembly tolerance is:

[0111]

[0112] In the formula, C M (T i ) is the total processing cost of assembly tolerance; n is the number of rings in the assembly dimension chain; T i is the tolerance of the ith component ring; C Mi (T i ) is the manufacturing cost caused by the tolerance of the i-th component ring.

[0113] S42: Constructing the quality loss-tolerance model: First, the deviation between the actual performance and the ideal performance of the product is identified as the quality loss; the tolerance of each component ring is used to represent the product assembly accuracy index, and the ideal performance value of the assembly and the actual performance calculated by the component ring tolerance are calculated respectively. The Taguchi quality loss cost function is introduced to complete the construction of the quality loss-tolerance model. The expression of the assembly quality loss is:

[0114] C Q = k(ym) 2

[0115]

[0116] In the formula, C Q is the quality loss, k is the quality loss coefficient, y is the quality characteristic value of the product, m is the target value of the product quality, A is the loss coefficient of part function failure, and x0 is the maximum deviation of the parameter from the target value.

[0117] S43: Constructing the repair cost-tolerance model: First, verify whether the current assembly accuracy can meet the performance requirements and determine whether repair work is needed; then, calculate the expected value of the repair quantity based on the tolerance data of each component ring, and integrate the repair economic cost, repair time, repair area size and other data generated in the mechanical product manufacturing process to obtain the assembly repair cost function represented by the tolerance, and complete the construction of the repair cost-tolerance model. The assembly repair cost expression formula is:

[0118]

[0119] Among them, S(T i ) is the expected step difference calculated from the tolerance data; R(T) is the maximum step difference that can be accepted to achieve assembly performance guarantee; C R (T i ) is the repair cost; C Area is the area of ​​the repair surface that can be calculated; C Cost is the economic cost coefficient of repair related to the repair surface material and shape characteristics; C Time It is the time cost coefficient required to complete the repair work under the factory processing efficiency.

[0120] In the multi-model weight parameter calculation method of establishing the integration of performance requirements and experience data in step 5, the tolerance data of each component ring obtained by the factory MBD model data or the experience of engineering personnel is used as the main support, and the assembly process and performance requirement analysis are used as the auxiliary to calculate the specific value of each model weight parameter. Step 5 specifically includes:

[0121] S51: Tolerance data T established based on the factory MBD digital model or the production experience of engineering personnel m , and obtain the size values, tolerance values, and processing features of each component ring of the assembly. Based on this, the specific values ​​of the manufacturing cost function model, quality loss function model, and repair cost model under the current tolerance data are preliminarily calculated, which are expressed as C M (T m ), C Q (T m ) and C R (T m ), and the weight parameters of each model are set to be α, β and γ respectively.

[0122] S52: Combined with factory manufacturing analysis, determine the key links of the assembly process and the performance requirements of product production, so as to identify the key target models that have a significant impact on the required performance improvement in the tolerance optimization process, roughly determine the importance of each target model to the final target performance improvement, and obtain the initial ratio of the weight parameter of each target model to the product of the model value, that is, α·C M (T m), β·C Q (T m ) and γ·C R (T m ) ratio.

[0123] S53: Decompose the above initial proportional formula into multiple unequal equations, each weight parameter in each equation is an unknown quantity, and the remaining values ​​are known. At the same time, let α+β+γ=1, and combine the above multiple equations to obtain a set of linear equations in which all unknown quantities are weight parameters. Finally, solve the linear equations to obtain the specific values ​​of each weight parameter, and complete the construction of the multi-model weight parameter calculation method integrating performance requirements and empirical data, where the linear equations are:

[0124]

[0125] In the formula, C M (T m ) is the manufacturing cost value under empirical data, C Q (T m ) is the mass loss value under empirical data, C R (T m ) is the value of the repair cost loss under empirical data, c1 and c2 are constants that can be obtained from the proportional relationship, and α, β, γ are the weight parameters of each model to be determined.

[0126] In the fusion construction of the tolerance allocation multi-objective optimization model that characterizes the comprehensive performance of the product in step six, each optimization objective model is selected and uniformly adjusted, and then the weight parameters of each model are obtained by the above-mentioned multi-model weight value calculation method. The multi-objective optimization problem is converted into a single-objective function optimization problem by weighted summation. Finally, a certain constraint space is imposed on the optimization process to complete the fusion construction of the tolerance allocation multi-objective optimization model.

[0127] S61: Obtain the manufacturing cost C established in step 4 M , quality loss C Q and Repair Cost Model C R The function model is uniformly adjusted to a tolerance T i is the format of the independent variable.

[0128] S62: Obtain the numerical data of the weight parameters of each model obtained in step 5, and use them to weight the sum of each target model to obtain the comprehensive performance index of the product Thereby, tolerance optimization is transformed from a multi-objective optimization problem to a single-objective function optimization problem.

[0129] S63: Considering production constraints such as tolerance processing capability, corresponding constraints are imposed on the optimization model, and the above models are integrated to obtain a mathematical model that includes the product comprehensive performance index function, each target optimization model function, tolerance range constraints and weight parameters, and complete the fusion construction of the tolerance allocation multi-objective optimization model.

[0130] In the step 7, in the process of iteratively solving the tolerance allocation multi-objective optimization model based on the accelerated particle swarm algorithm, the optimization algorithm program is divided into four parts: target model module, constraint application module, algorithm implementation module and sub-function module. The above mathematical model is input into each module and the optimization program is run to obtain the tolerance values ​​of each component ring after optimization. The step S7 specifically includes:

[0131] S71: Define the objective function file and obtain the product comprehensive performance index model R (T i ), Manufacturing Cost Model C M (T i ), quality loss model C Q (T i ) and repair cost model C R (T i ), then convert the above model into a format recognizable by the algorithm program, import it into the objective function file, and complete the construction of the target model module.

[0132] S72: Define the constraint function file, obtain the constraint space data set in S6, and convert the constraint space data into the form of inequality, and then further write it into a format recognizable by the algorithm program and import it into the constraint function file one by one to complete the construction of the constraint application module.

[0133] S73: Define and write the accelerated example swarm algorithm program, and build multiple sub-function files that need to be called for algorithm operation, such as the accelerated particle swarm algorithm implementation code file, the initialization particle function file, the particle update file, etc. Then, call the target model module, the constraint application module and the sub-function module in the algorithm implementation program, and define various parameters such as the population size, the number of iterations, and the particle random attenuation factor of the accelerated example swarm. Run the accelerated particle swarm algorithm program to obtain the optimized tolerance allocation plan.

[0134] The present invention also provides a tolerance allocation multi-objective model parameter fusion optimization system, the optimization system comprises a memory and a processor, the memory is connected to the processor, and the fusion optimization method is stored in the memory.

[0135] Embodiment 1:

[0136] In this embodiment, the step difference optimization of a certain type of spacecraft wing docking assembly is taken as an example. Figure 3As shown in the figure, the assembly consists of two wing sections, and the two wings are bolted together at the connecting strip plate using a multi-hole fit method. While meeting the assembly accuracy of the wing component sections, it is also necessary to ensure the high-demand assembly coordination accuracy index expressed in terms of step difference. Taking the step difference coordination accuracy optimization as an example, the implementation steps of the tolerance allocation multi-objective model parameter fusion optimization method for performance requirements and assembly process improvement are explained.

[0137] The overall idea adopted by the present invention to solve its technical problems is: by analyzing the error coupling relationship between the matching features of complex assemblies, constructing an assembly feature geometric error spinor model based on the error source tolerance domain, and according to the matching relationship between the key features of each component, clarifying the transmission path and network of the assembly coordination error, establishing the actual change relationship model between different features on the same assembly part, and obtaining the error accumulation calculation model of a single assembly and the coordination error model between multiple assemblies. Considering the actual production needs of mechanical products, multiple single-objective tolerance optimization models are constructed. Then, the proposed multi-model weight parameter calculation method that integrates performance requirements and empirical data is used to obtain the values ​​of each weight, and the fusion construction of the multi-objective tolerance optimization model is completed in combination with relevant constraints. Finally, the target model is optimized and solved based on the accelerated particle swarm algorithm, and a tolerance allocation scheme that can effectively guarantee assembly performance is obtained, which provides an effective solution to the problems of model fusion difficulties and optimization direction deviation from actual needs in the tolerance optimization process.

[0138] The present invention provides a tolerance allocation multi-objective model parameter fusion optimization method for performance requirements and assembly process improvement, specifically a tolerance allocation optimization method combining mechanical product assembly process improvement, multi-model weight parameter calculation, multi-objective optimization model fusion construction and intelligent algorithm solution, which is used for mechanical product tolerance optimization and process improvement. The tolerance allocation multi-objective model parameter fusion optimization method for performance requirements and assembly process improvement includes the following steps:

[0139] S1: Construct a geometric error spinor model with characteristic changes of basic error sources;

[0140] Specifically, the assembly structure and assembly process of the two wing sections are first analyzed: there are two interfaces at the bottom of the two wing sections, which are in contact with the fixture. The center of the oblong hole at the intersection of the two farthest interfaces is used as the reference for positioning and clamping, and then bolts are used to connect the wing sections. After the docking is completed, the wing profile is measured and adjusted until the wing profile accuracy and the step difference between the wing components meet the acceptance requirements, thus completing the wing assembly work.

[0141] Furthermore, the above analysis of the assembly process and assembly index requirements of the segmented wing identifies the interface between the bottom of the wing and the fuselage as a key assembly feature, such as Figure 4As shown in the figure. With the interface as the assembly reference, the interface error will have a significant impact on the wing shape accuracy. Let E0 be the manufacturing error of the bottom joint of each wing section, E1 be the matching error between each wing section and the assembly tooling, and E2 be the docking error between the two wings. According to the product assembly accuracy requirements, the wing assembly step difference is decomposed into multiple feature points and surface information, and an error spinor model of the assembly step difference under the joint constraints of dimensional tolerance and form and position tolerance is constructed.

[0142] Furthermore, a spatial coordinate system with the center of the oblong hole at the wing intersection interface as the origin is established, and the x direction is set as the heading direction, the y direction is the wingspan direction, and the z direction is the height direction. Since the position tolerance of the hole is ±0.05mm, the screw model and constraint inequality of the tolerance domain of the oblong hole at the wing bottom interface are derived as follows:

[0143]

[0144] Furthermore, a space coordinate system is established with the center of the spacecraft wing surface as the origin. According to the requirement that the wing shape surface accuracy error is less than 0.8 mm, the error spinor model and constraint inequality of the wing can be derived as follows:

[0145]

[0146] S2: Constructing a key feature pose change model considering matching errors

[0147] Specifically, we first analyze the changes in assembly errors between parts on the wing, and believe that the matching errors between the two wings are caused by manufacturing errors at key features. We select two key features F located on the wing frame. 2i and F2, F 2i Key features of the wing bottom fixture F 1i F2 participates in the cooperation of the next part. The error screw model construction method in S1 is used to construct the key feature F 1i and F 2i The error screw models in their respective local coordinate systems are converted to the global coordinate system, and the matching error of the two wings is calculated to be 0.1 mm.

[0148] Furthermore, the size range of the parts matching area is obtained, and the position change of the wing frame caused by the matching part of the wing bottom fixture and the wing frame is calculated. The position change in the Z direction is 0.1mm, and the verticality change of the wing frame axis is 0.1°. According to the wing frame structure, two different key features F on an assembly are obtained. 2i and F2, and the key feature F2 relative to the key feature F 2iThe pose change matrix of represents the relationship. Combined with the assembly error of the fuselage frame and the bottom fixture, the final cumulative error of the wing frame assembly can be calculated, that is, the pose change of the key feature F2, which is 1.65mm in the Z direction.

[0149] S3: Constructing single assembly error accumulation and multi-assembly coordination error model

[0150] like Figure 5 As shown, specifically, firstly, according to the geometric change analysis of the basic error source in S1, the screw model and constraint inequality of the center of the oblong hole on the wing interface are obtained. Then, according to the posture change transmission model between different key features on the parts caused by assembly errors established in S2, the assembly error caused by the assembly of the wing bottom and the fixture is calculated. The relationship between two different key features on the same part is analyzed through the wing part structure, and the verticality of the wing skeleton axis is calculated and deduced. Then, the position change of the farthest section of the wing from the wing bottom is calculated. The error change of the wing surface in the normal direction is shown as follows: Figure 6 shown.

[0151] Furthermore, the range and tolerance domain of the geometric deviation of each part of the assembly wing I are obtained, the error geometric model of the assembled parts equivalent to the coordinate system is determined, the matching error transmission form between the parts on the assembly wing I is defined, and then the posture change of the subsequent parts assembly transmission is determined. By analyzing the assembly process of each component part on the assembly wing I, the assembly error transmission calculation method of step 2 is used for each component part, and then the final assembly error of the assembly wing I is calculated as:

[0152] δ A =1.85mm

[0153] Furthermore, the error transmission direction and transmission path of the assembly are obtained, and the final assembly error δ of the assembly wing I and wing II calculated in S32 is A , δ B , the assembly coordination error transmission accumulation model and assembly dimension chain between assemblies were constructed, and the posture change of wing II in the Z direction of the global coordinate system was solved to be 1.65mm, the verticality β=0.1°, and the assembly step difference XN1=2.54mm between the two wings at the joint. At the same time, the Monte Carlo method was used to analyze the dimension chain, and the probability that a single assembly under the current assembly scheme meets the wing assembly index requirements was calculated. According to the assembly requirement that the forward step difference between the wing components is less than 1mm, the probability that the assembly step difference between the wing section I and the wing section II meets the assembly requirement of 1mm is calculated to be P=16%.

[0154] Furthermore, since the wing shape surface accuracy requirement is ±0.8mm, the position change caused by positioning and assembling based on the center of the intersection oblong hole at the two farthest tooling interfaces at the bottom of the wing is greater than the wing assembly shape surface accuracy, which will have a greater impact on the subsequent docking error between wings. According to the wing shape assembly index requirements, the wing shape surface accuracy must be limited to ±0.8mm. Through calculation, it is found that the assembly position change caused by the intersection oblong hole at the bottom of the wing at the interface must be less than 0.0242mm. Therefore, in order to meet the assembly target, a card with a positioning accuracy of 0.2mm is added to the assembly tooling process to assist the assembly of each section of the wing.

[0155] S4: Build multiple single-objective tolerance optimization models for product performance improvement needs

[0156] Specifically, the manufacturing performance requirements of the two-wing assembly are first analyzed, and the corresponding target optimization model is selected. Considering that the wing assembly faces the problem of difficult to meet the requirements of step difference assembly accuracy and excessive repair workload during the production process, combined with the step difference assembly dimension chain between the two wings, such as Figure 7 As shown in the figure, tolerance function models of manufacturing cost, quality loss and repair cost are established respectively to optimize the tolerance of the spacecraft step-difference assembly dimension chain.

[0157] Furthermore, for the construction of the manufacturing cost-tolerance model, the tolerance-cost data generated in the production process is analyzed and the appropriate tolerance cost model is selected for fitting, and an accurate composite tolerance cost model is established for different tolerance characteristics.

[0158] Specifically, for the outer circle feature, the mathematical expression of the tolerance manufacturing cost is: For internal hole features, the mathematical expression for the tolerance manufacturing cost is: For positioning dimensions, the mathematical expression for tolerance manufacturing cost is: For planar features, the mathematical expression for the tolerance manufacturing cost is: For the position of the shaft (hole), the mathematical expression of the tolerance manufacturing cost is: For coaxiality, the mathematical expression of tolerance manufacturing cost is: C M (T) = 0.0373e -3.08T .

[0159] Furthermore, the tolerance-cost function of each component ring in the assembly dimension chain is sorted out. The step difference assembly dimension chain includes component rings L1, L3, L4, L6, L8, L9, L10, LS1, LS2 and closed ring L0, among which L1 is the thickness dimension of the upper end of the wing skeleton of section II; L3 is half of the internal length dimension of the wing skeleton of section II; L4 is the length dimension from the upper surface of the wing skeleton of section I to the convex end; L6 is the thickness dimension of the convex end of the interface of the wing skeleton of section I; L8 is half of the internal length dimension of the wing skeleton of section I; L9 is the change in the assembly error of the wing shape of section I caused by the positioning of the bottom joint of section I on the tooling; L10 is the change in the assembly error of the wing shape of section II caused by the positioning of the bottom joint of section II on the tooling; LS1 is the skin thickness dimension of section I wing; LS2 is the skin thickness dimension of section II wing; L0 is the assembly step difference between the two wings. At the same time, the above component ring processing features are all plane features, so the tolerance processing cost of each component ring is:

[0160]

[0161] Calculate the processing cost of each tolerance and sum them up to get the total processing cost C of the product assembly tolerance M (T i ):

[0162]

[0163] In the formula, C M (T i ) is the total processing cost of assembly tolerance; n is the number of rings in the assembly dimension chain; T i is the tolerance of the ith component ring; C Mi (T i ) is the manufacturing cost generated by the tolerance of the i-th component ring. So far, the construction of the manufacturing cost-tolerance model has been completed.

[0164] Furthermore, for the construction of the quality loss-tolerance model, the difference between the actual performance and the ideal performance of the product is identified as the quality loss, and the model construction work is carried out based on the Taguchi quality loss function. For the quality loss caused by the assembly step difference, let S(T0) be the target step difference value, S(T) be the actual step difference calculated by the tolerance of each component ring, A represents the loss coefficient of the product assembly function failure caused by the step difference, and x0 is the allowable deviation value of the actual step difference from the target. The tolerance-quality loss model can be constructed:

[0165]

[0166] S(T i )=Σ A,B (T i )

[0167] In the formula, C Q (T) is the quality loss, A is the loss coefficient of product assembly failure due to step difference, which is 4 here; x0 is the allowable deviation of the actual step difference from the target, which is 0.4 here; S(T i ) is the actual step difference, S(T m ) is the ideal step difference, which is taken as 0 here; T i is the tolerance of each component ring; Σ A,B is the assembly step coordination error.

[0168] Furthermore, for the construction of the repair cost-tolerance model, the expected value function of the repair quantity is first calculated with the tolerance as the independent variable. At the same time, the repair quantity, repair area, repair economic cost and repair time cost generated in the product manufacturing process are considered to establish a repair cost model that can comprehensively characterize the benefit loss caused by the repair link. It is expressed as:

[0169]

[0170] In the formula, C R (T i ) is the repair cost; C Area is the area of ​​the repair surface that can be calculated, and the value here is 5.37; C Cost is the repair economic cost coefficient related to the repair surface material and shape characteristics, and the value here is 3; C Time is the time cost coefficient required to complete the repair work under the factory processing efficiency, which is taken as 1.5; S(T i ) is the expected step difference calculated from the tolerance data; R(T) is the maximum step difference that can be accepted to achieve assembly performance assurance, and its value here is 0.1;

[0171] S5: Establish a multi-model weight parameter calculation method that integrates performance requirements and empirical data

[0172] Specifically, first obtain the tolerance data contained in the factory MBD model or the tolerance data set by the engineering staff's production experience, and use it to calculate the manufacturing cost, quality loss, and repair cost values ​​under the tolerance allocation plan. After that, analyze the product production and service performance requirements, identify the key target models that can improve the product production efficiency, set the initial proportion of each model, and complete the calculation of the weight parameters of each model, as shown below: Figure 8 shown.

[0173] Furthermore, the product performance requirements and assembly process are analyzed to roughly determine the importance of each target model and obtain α·C M (T m ), β·C Q (T m ),γ·C R (Tm ) The initial ratio of the three parts is then calculated by reading the tolerance data contained in the factory MBD model or the tolerance experience data given by the engineering technicians. M (T m ), C Q (T m ), C R (T m ) specific value; finally, let α+β+γ=1, and combine it with the above proportional formula to obtain the specific values ​​of α, β, γ used in the tolerance optimization link, and then obtain the comprehensive evaluation index of assembly performance R(T i ).

[0174] Specifically, for the assembly of two wing sections, by analyzing the production and service performance requirements, it is believed that the control of quality loss and repair cost is more important for product performance guarantee. Therefore, the quality loss, manufacturing cost and maximum repair quantity are set in the comprehensive evaluation index R(T m ) are 0.3, 0.3, and 0.4 respectively. At this time, the tolerance data T m The value is known, that is, C M (T m ),C Q (T m ),C R (T m ) are all known quantities, and α, β, γ are unknown quantities. According to the equation:

[0175] [α·C M (T m )]:[β·C Q (T m )]:[γ·C R (T m )]=0.3:0.3:0.4

[0176] α+β+γ=1

[0177] Convert the above ratio into multiple equations to obtain a set of linear equations for solving the specific values ​​of α, β, and γ:

[0178]

[0179] The specific values ​​of α, β, and γ can be obtained. At the same time, in the calculation process, the selection of weight parameters will restrict the excessive model values ​​while also adjusting the model values ​​that are too small to a certain extent. On a macro level, it guides the subsequent algorithm optimization direction to be consistent with the product optimization requirements, and obtains the accurately established assembly performance comprehensive evaluation index R (T i ) to complete the fusion construction of the multi-objective optimization model.

[0180] S6: Fusion construction of tolerance allocation multi-objective optimization model to characterize comprehensive product performance

[0181] Specifically, first obtain the manufacturing cost model C established in S4 M (T i ), quality loss model C Q (T i ) and repair cost model C R (T i ), the total manufacturing cost of the component ring tolerance in the two-segment wing assembly step difference dimension chain is obtained as:

[0182]

[0183] The quality loss caused by the step difference between the two wing sections on product performance is:

[0184]

[0185] In the wing assembly and repair process, the repair cost caused by ensuring that the wing assembly step difference meets the production requirements is:

[0186]

[0187] Furthermore, the weight parameter values ​​of each model obtained in S5 are obtained, and the weighted sum of each model is obtained to obtain the evaluation index R (T i ), the expression is:

[0188] R(T i )=α·C M (T i )+βC Q (T i )+γC R (T i )

[0189] Furthermore, taking manufacturing cost, quality loss and repair cost as optimization targets, relying on product comprehensive performance evaluation index, the optimization is transformed into a single-objective optimization problem, and the constraint space and parameter requirements of the model are further improved. Finally, a single-objective model and comprehensive evaluation index R(T i ), the optimization model of constraint conditions and weight parameter settings is as follows:

[0190]

[0191] In the formula, C M (T i ) is the manufacturing cost, C Q (T i ) is the quality loss cost, C R (Ti ) is the repair cost, T i is the ring tolerance, α, β, γ are the weight parameters of manufacturing cost, quality loss and repair cost respectively, S(T i ) is the actual step difference, S(T m ) is the ideal step difference, (T i ) min is the minimum tolerance that can be achieved by the i-th component ring under the current processing capability, (T i ) max is the maximum tolerance that can be achieved by the i-th component ring under the current processing capability. i ) is the comprehensive performance evaluation index of the assembly, and the obtained R(T i ) is smaller, the better the overall performance of the product is. The algorithm is optimized to achieve R(T i ) is minimized as the goal.

[0192] S7: Iterative solution of tolerance allocation multi-objective optimization model based on accelerated particle swarm optimization

[0193] Specifically, MATLAB software is first selected as the platform for constructing the above mathematical model and implementing the algorithm. The optimization program is based on the mathematical model established above, and the optimization objective function and tolerance range constraints are written into a format that the program can recognize. Then, the objective function, constraint function file and each sub-function file are called in the accelerated particle swarm algorithm file, and the various parameters of the algorithm are set, as shown in Table 1.

[0194] Table 1

[0195]

[0196]

[0197] Specifically, in order to improve the efficiency of model input and program writing, the optimization algorithm program is divided into four parts: target model module, constraint imposition module, algorithm implementation module and sub-function module. For the writing of the objective function file, based on the above tolerance allocation multi-objective optimization model, the function models of manufacturing cost, quality loss and repair cost are written in MATLAB software, and the weighted sum of the values ​​of each model is obtained according to the calculated weight parameters to obtain the final objective function; the constraint imposition file is written, mainly based on the establishment of inequalities to establish the constraint space for subsequent optimization;

[0198] Furthermore, the algorithm implementation module mainly reads the objective function file and constraint file, sets the algorithm parameters such as variable upper and lower limits, particle population size, number of iterations, and calls the APSO algorithm code to achieve the optimization solution of the objective function; for the sub-function module, the operation of the algorithm implementation module requires calling various sub-function files, including accelerated particle swarm code, particle function initialization, penalty function establishment and other files.

[0199] Specifically, the particle acceleration coefficient value is adjusted in the accelerated particle swarm code file.

[0200] nc1=c1×(1-λ)

[0201] nc2=c2×(1-λ)

[0202]

[0203]

[0204] Among them, χ is the probability of mutation; ω and φ are coefficient factors; f max 、f min 、f avg are the maximum, minimum and average fitness of the particles respectively.

[0205] Furthermore, through the acceleration coefficients nc1 and nc2 obtained by the above calculation, the particle velocity calculation formula is:

[0206]

[0207] Where a1 and a2 are calculated by comparing the particle with the minimum fitness value with the pbest particle value. If the two particle values ​​are the same, the values ​​of a1 and a2 are both 0, otherwise the values ​​are both 1.

[0208] Furthermore, after completing the input of the optimization objective function and the construction of the accelerated particle swarm algorithm program, the algorithm implementation module is run to obtain the optimized tolerance values ​​of each component ring.

[0209] Further, two assembly process solutions in S3 are obtained, namely, the assembly process before improvement (i.e., the wing bottom joint is used as the assembly reference) and the assembly process after improvement (i.e., the positioning method using the outer shape card plate, such as Fig. 9 As shown in Table 2, two tolerance allocation optimization models are established according to the two assembly process schemes. The tolerance allocation schemes under the two assembly process schemes are optimized respectively. At the same time, the data of each model before and after the optimization are recorded and analyzed. The optimization results and effects are shown in Table 2.

[0210] Table 2

[0211]

[0212] The above optimization effects are as follows Fig.10 and Fig.11 As shown in the figure, the manufacturing cost, quality loss, repair cost and comprehensive evaluation index are all three lines, and the three lines represent the percentage of improvement, after optimization and before optimization from top to bottom. Fig.10 Taking the manufacturing cost as an example, the corresponding improvement percentages are 10.64%, 36.28 after optimization, and 40.6% before optimization. By analyzing the two assembly process schemes, it can be found that the optimization range of each model in the assembly process with added pallets is significantly smaller than the optimization range of the assembly process without pallets, which fully shows that the assembly step difference is mainly caused by the bottom interface benchmark. By adding pallets during the assembly process, the influence of the bottom interface benchmark error of each section of the wing on the step difference can be effectively avoided. At the same time, relying on the multi-model weight parameter value method, the improvement of mass loss and repair cost is also significantly greater than the manufacturing cost, indicating that more optimization space is left for mass loss and repair cost, which is consistent with the expected optimization goals and performance requirements, effectively avoiding problems such as imbalance of target model data and optimization deviation from the demand direction, and effectively ensuring the coordination quality of assembly step difference.

[0213] In the present invention, by analyzing the error coupling relationship between the matching features of complex assemblies, an assembly feature geometric error spinor model based on the error source tolerance domain is constructed, and according to the matching relationship between the key features of each component, the transmission path and network of the assembly coordination error are clarified, and the actual change relationship model between different features on the same assembly part is established, and the error accumulation calculation model of a single assembly and the coordination error model between multiple assemblies are obtained. At the same time, in view of the problem that the weight value of each tolerance target model is difficult to determine, a multi-model weight parameter value method supported by performance requirements and empirical data is proposed to achieve efficient and accurate construction of multi-objective models. First, the key objectives of tolerance optimization are identified based on the assembly performance requirements, and then the initial proportions of each target model are set; secondly, the factory MBD digital model tolerance data or tolerance engineering experience data are introduced to provide data support for subsequent weight parameter calculations; finally, based on the initial proportions of each model and the tolerance experience data, the weight parameter values ​​of each target are obtained to complete the fusion construction of the multi-objective optimization model. Finally, taking the assembly of segmented wings of a certain type of spacecraft as an example, the tolerance optimization models of the two assembly processes before and after optimization were analyzed and established, and the accelerated particle swarm algorithm was used to solve the two models respectively. The results show that the introduction of the pallet in the assembly process significantly reduces the optimization range of the tolerance model. Fig.11As shown in the figure, the reductions in manufacturing cost, quality loss and repair cost are 0.78%, 11.21% and 8.50% respectively, which fully shows that the main error of assembly step difference comes from the change of bottom interface datum. Through the coordination between assembly process improvement and tolerance optimization, the improved assembly process can effectively alleviate the pressure of tolerance optimization and further improve the assembly quality and economic benefits of products.

[0214] The present invention aims at the problems faced by the fusion construction of multi-objective optimization models in the assembly tolerance optimization process of complex mechanical products, such as inconsistent data connotation, insufficient effective data support and difficulty in coordination between multiple objective models. At the same time, considering the lack of effective theoretical guidance from performance demand analysis and assembly process improvement in the prior art, the low-quality tolerance optimization data obtained thereby is difficult to meet the assembly performance requirements of modern products. Firstly, considering the posture changes between different key features on parts caused by fitting errors, an assembly error transmission accumulation model and a coordinated dimension chain are constructed, and the improvement of the assembly process scheme is achieved by analyzing the error prediction data; secondly, in combination with the error prediction data, a manufacturing cost, quality loss and repair cost model is established, and a multi-model fusion weight calculation method that integrates performance requirements and empirical data is proposed to achieve the comprehensive construction of a multi-objective optimization model; finally, an accelerated particle swarm algorithm is used to solve the tolerance allocation multi-objective optimization model to obtain an allocation scheme for improving assembly performance.

[0215] The above is a detailed introduction to a tolerance allocation multi-objective model parameter fusion optimization method and system provided by the embodiment of the present application. The description of the above embodiment is only used to help understand the method and its core idea of ​​the present application; at the same time, for those of ordinary skill in the art, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0216] For example, certain words are used in the specification and claims to refer to specific components. Those skilled in the art should understand that hardware manufacturers may use different nouns to refer to the same component. This specification and claims do not use differences in names as a way to distinguish components, but use differences in the functions of components as the criteria for distinction. As mentioned throughout the specification and claims, "including" and "comprising" are open-ended terms, so they should be interpreted as "including / including but not limited to". "Approximately" means that within an acceptable error range, those skilled in the art can solve the technical problem within a certain error range and basically achieve the technical effect. The subsequent description of the specification is a preferred embodiment of the present application, but the description is for the purpose of illustrating the general principles of the present application, and is not used to limit the scope of the present application. The scope of protection of the present application shall be determined by the definition of the attached claims.

[0217] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a product or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such a product or system. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the product or system including the elements.

[0218] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0219] The above description shows and describes several preferred embodiments of the present application, but as mentioned above, it should be understood that the present application is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the application concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art do not deviate from the spirit and scope of the present application, and should be within the scope of protection of the claims attached to the present application.

Claims

1. A tolerance allocation multi-objective model parameter fusion optimization method, which is used to improve the performance requirements and assembly coordination process of segmented wing structural parts during the production and manufacturing process, characterized in that: The fusion optimization method comprises the following steps: S1: According to the structural characteristics and assembly process of the parts to be produced, a geometric error screw model of the basic error source characteristic variation is constructed; S2: Construct a key feature pose change model considering the matching error based on the geometric error screw model; S3: Based on the geometric error screw model and the key feature pose change model, a single assembly error accumulation model and a multi-assembly coordination error model are constructed; S4: Based on the product requirements of the parts to be produced, the single assembly error accumulation model and the multi-assembly coordination error model, multiple single-objective tolerance optimization models are constructed to improve performance according to product requirements; S5: Establish a multi-model weight parameter calculation method that integrates performance requirements and empirical data; S6: Through the multi-model weight parameter calculation method, multiple single-objective tolerance optimization models in S4 are integrated to construct a tolerance allocation multi-objective optimization model that characterizes the comprehensive performance of the product; S7: Accelerated particle swarm optimization is used to iteratively solve the tolerance allocation multi-objective optimization model; The method for calculating the multi-model weight parameters in S5 is specifically as follows: S51: According to the factory MBD digital model or the tolerance data set by the engineering staff’s production experience, the dimension values, tolerance values ​​and processing feature information of each component ring of the assembly are obtained, and the specific values ​​of the manufacturing cost function model, quality loss function model and repair cost model under the current tolerance data are preliminarily calculated, which are expressed as C M (T m ), C Q (T m ) and C R (T m ), and the weight parameters of each model are set to be α, β and γ respectively; S52: Combined with factory production and manufacturing analysis, determine the key links of the assembly process and the performance requirements of product production, obtain the key target model that has a significant impact on the required performance improvement in the tolerance optimization process, determine the importance of each target model to the final target performance improvement, and obtain the initial ratio of the weight parameter of each target model to the product of the model value, that is, α·C M (T m ), β·C Q (T m ) and γ·C R (T m ) S53: Decomposing the above initial proportional equation into multiple inequivalent equations, each weight parameter in each equation is an unknown quantity, and the remaining values ​​are all known, and at the same time, setting α+β+γ=1, combining multiple inequivalent equations, a set of linear equations in which all unknown quantities are weight parameters is obtained; S54: Solve the linear equations to obtain specific values ​​of each weight parameter, and complete the construction of a multi-model weight parameter calculation method that integrates performance requirements and empirical data, where the linear equations are: In the formula, C M (T m ) is the manufacturing cost value under empirical data, C Q (T m ) is the mass loss value under empirical data, C R (T m ) is the value of the repair cost loss under empirical data, c1 and c2 are constants that can be obtained from the proportional relationship, and α, β, and γ are the weight parameters of each model to be determined.

2. The fusion optimization method according to claim 1, characterized in that: The S1 specifically includes: S11: Analyze the assembly product structure and assembly performance requirements, obtain specific product assembly accuracy requirements, and convert the product assembly accuracy requirements into the form of feature points and surfaces according to the assembly accuracy characteristics, and obtain the assembly dimension chain accuracy index requirements expressed by the screw model; S12: Taking the assembly dimension chain accuracy index requirement as the source and the assembly datum as the end point, find the error transmission path, identify each dimension link on the transmission path as the error source, and obtain the key measurement points and geometric feature information of each error source; S13: The position change data of the geometric features of the assembly are obtained according to the key measurement points and geometric feature information of each error source, and the position change data are converted into an error spinor model in a matrix format using kinematic theory and small displacement spinor method, and the error spinor model of the position change is mapped to the tolerance domain.

3. The fusion optimization method according to claim 1, characterized in that: The S2 specifically includes: S21: construct the global coordinate system of the assembly and the local coordinate systems of each component of the assembly, and set the homogeneous transformation matrix of each local coordinate system relative to the global coordinate system, and obtain the key features on each assembly part; S22: The representation relationship of the key features of each part in its own local coordinate system is obtained through the error screw model in S1, and the homogeneous coordinate transformation matrix is ​​obtained according to the transformation relationship between each local coordinate system and the global coordinate system, and the fitting error between each part is calculated in the global coordinate system; S23: According to the matching errors between the parts, the size and tolerance data of the matching area of ​​the parts are obtained, and based on the size and tolerance data, the part posture change data caused by the matching between the parts are calculated; S24: Analyze and construct the transformation relationship matrix of two key features on the same part and the part posture change matrix caused by the matching errors of the two parts, and then multiply the two to obtain the assembly cumulative error matrix after the parts are matched; S25: Convert the assembly cumulative error matrix into a vector form and ignore the influence of high-order micro-variations to obtain the component form of the cumulative error of part features in each direction.

4. The fusion optimization method according to claim 3, characterized in that: The S3 specifically includes: S31: Analyze the range and tolerance domain of geometric deviation in S1, determine the error geometric model of the assembled parts equivalent to the coordinate system, define the form of matching error transmission between the parts on the assembly, and then determine the posture change of the subsequent parts in the assembly transmission; S32: In combination with the assembly process of each component part of the assembly, the error transfer model is performed for each component part, and according to the calculation method in S23, the error of the previous part is introduced into the next part to determine the final assembly error of the entire assembly and the single assembly error accumulation model; S33: Obtain the error transmission direction and transmission path of the assembly, and use the final assembly error of the entire assembly calculated in S32 as data support, and construct an assembly coordination error transmission accumulation model between assemblies through the absolute value of the difference between the final cumulative error values ​​of the two assemblies.

5. The fusion optimization method according to claim 4, characterized in that: The S3 further includes: S34: The Monte Carlo method is used to analyze the assembly coordination error dimension chain, obtain the accumulation law of coordination error and calculate the error mean; S35: Analyze the assembly errors of each component and the coordination error data of the assembly obtained in S33, identify the key assembly elements and key assembly processes in the assembly process of the mechanical product, and optimize the assembly process links that have a greater impact on the coordination error of the assembly based on the accumulation law of the coordination error and the statistical results of the error mean.

6. The fusion optimization method according to claim 5, characterized in that: The S4 specifically includes: S41: Constructing a manufacturing cost-tolerance model: Analyze the tolerance-cost data generated in the manufacturing process and select a suitable tolerance cost model for fitting to obtain an accurately established tolerance-cost function; secondly, sort out the tolerance-cost functions of each component link in the assembly dimension chain, calculate the processing cost of each tolerance and sum them up to obtain the total processing cost of the product assembly tolerance, and complete the construction of the manufacturing cost-tolerance model; S42: Constructing a quality loss-tolerance model: The deviation between the actual performance and the ideal performance of the product is identified as quality loss; the tolerance of each component ring is used to represent the product assembly accuracy index, and the ideal performance value of the assembly and the actual performance calculated by the component ring tolerance are calculated respectively. The Taguchi quality loss cost function is introduced to complete the construction of the quality loss-tolerance model; S43: Construct a repair cost-tolerance model: verify whether the current assembly accuracy can meet the performance requirements and determine whether repair work is needed; then calculate the expected value of the repair quantity based on the tolerance data of each component ring, and combine the repair economic cost, repair time and repair area size data generated in the mechanical product manufacturing process to obtain the assembly repair cost function represented by the tolerance, and complete the construction of the repair cost-tolerance model.

7. The fusion optimization method according to claim 6, characterized in that: The tolerance allocation multi-objective optimization model that is integrated and constructed in S6 to characterize the comprehensive performance of the product specifically includes: S61: Obtain the manufacturing cost C established in S4 M , quality loss C Q and Repair Cost Model C R The function model is uniformly adjusted to a tolerance T i is the format of the independent variable; S62: Obtain the numerical data of the weight parameters of each model obtained in the multi-model weight parameter calculation method, and use this to weight the sum of each target model to obtain the comprehensive performance index of the product Transform tolerance optimization from a multi-objective optimization problem into a single-objective function optimization process; S63: According to the production constraints of tolerance processing capacity, corresponding constraints are imposed on the optimization model, and the models in S61 are integrated to obtain a mathematical model including the product comprehensive performance index function, each target optimization model function, tolerance range constraints and weight parameters, and complete the fusion construction of the tolerance allocation multi-objective optimization model.

8. The fusion optimization method according to claim 7, characterized in that: The S7 specifically includes: S71: Define the objective function file and obtain the product comprehensive performance index model R (T i ), Manufacturing Cost Model C M (T i ), quality loss model C Q (T i ) and repair cost model C R (T i ) and then convert the above model into a format that can be recognized by the algorithm program, import it into the objective function file, and complete the construction of the objective model module; S72: define a constraint function file, obtain the constraint space data set in S6, and convert the constraint space data into the form of inequalities, and then further write them into a format recognizable by the algorithm program and import them into the constraint function file one by one, so as to complete the construction of the constraint application module; S73: Define and write the accelerated example swarm algorithm program, and at the same time build multiple sub-function files that need to be called for algorithm operation, and then call the target model module, constraint application module and sub-function module in the algorithm implementation program. At the same time, define the population size, number of iterations, and particle random attenuation factor parameters of the accelerated example swarm, and run the accelerated particle swarm algorithm program to obtain the optimized tolerance allocation plan.

9. A tolerance allocation multi-objective model parameter fusion optimization system, characterized in that: The optimization system includes a memory and a processor, the memory is connected to the processor, and the memory stores the fusion optimization method as described in any one of claims 1 to 8.

Images