High-precision assembly part shape maintaining method based on Monte Carlo large model prediction

By using the Monte Carlo large model prediction method to optimize the assembly tolerance coefficient, the problem of low quality in the assembly coordination of large and complex components is solved, and a high-precision and reliable assembly process is achieved.

CN121765856APending Publication Date: 2026-03-31AVIC XIAN AIRCRAFT IND GRP CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511805061.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-12-03
Publication Date
2026-03-31

AI Technical Summary

Technical Problem

Existing technologies lack a systematic approach to the assembly coordination of large and complex components, resulting in poor assembly quality. Traditional methods rely on experience and cannot efficiently perform dimensional chain calculations and analyses, and the analysis of influencing factors is inaccurate.

Method used

A Monte Carlo large model prediction method is adopted. By constructing a Monte Carlo tree model through the linear relationship between the external dimensions and manufacturing tolerances of multi-level components, random sampling and visualization model verification are carried out to optimize the assembly tolerance coefficient and ensure the accurate allocation of overall product tolerances.

Benefits of technology

It achieves precise allocation of global tolerances, improves the controllability and quality of the assembly process, and ensures high precision and reliability of aircraft assembly.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121765856A_ABST
    Figure CN121765856A_ABST
Patent Text Reader

Abstract

The invention relates to but not limited to the field of large part assembly, in particular to a high-precision assembly part shape maintaining method based on Monte Carlo large model prediction. Comprising the following steps: aiming at multi-stage components of specific product clear group assembly design, extracting the boundary dimension of each stage of component; determining the manufacturing tolerance of each stage of component after forming based on the target boundary dimension in combination with the manufacturing process capability; based on a process instruction of assembling each level of component to the assembled part, segmenting the assembly process and defining the tolerance of each level of assembly process to be in a linear relation with the manufacturing tolerance after the corresponding level of component is formed; constructing a fit tolerance prediction model based on a Monte Carlo method, and determining the contribution degree sequence of the tolerance of each level of assembly process to the tolerance of the whole product; carrying out random sampling on the fit tolerance coefficient to obtain an optimal fit tolerance coefficient; the effectiveness of a prediction result is checked through a visual model, and under the condition that the prediction result is effective, the optimal solution is called as an implementation parameter to complete assembly of the whole part product.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to, but is not limited to, the field of large component assembly, and specifically to a high-precision method for preserving the shape of assembled components based on Monte Carlo large model prediction. Background Technology

[0002] 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 about more efficient production and high-quality manufacturing performance. At the same time, it has also put forward requirements for higher quality, lower cost and more targeted performance requirements for product production. This makes it difficult for traditional tolerance allocation schemes based on engineering experience to guarantee product assembly performance.

[0003] Assembly coordination techniques for large and complex components can only optimize the assembly coordination process from a single dimension and address a single problem. These include techniques such as dimensional chain analysis and tolerance optimization, deformation prediction and control during positioning, manufacturing process accuracy evaluation, and assembly gap and interference compensation. A systematic assembly coordination approach has not been developed, failing to efficiently, effectively, and cost-efficiently solve the engineering problems of precise assembly of large aircraft composite structures. Previously, most process engineers used extreme value methods or statistical analysis to solve engineering problems. These methods cannot efficiently calculate and analyze the dimensions of typical complex spatial dimensional chain assemblies. Furthermore, the use of existing 3D tolerance analysis software in engineering applications heavily relies on engineers with specialized knowledge and extensive experience, requiring repeated verification to avoid errors, resulting in inaccurate analysis of influencing factors and their contribution.

[0004] Therefore, in order to achieve predictable assembly accuracy and avoid the blindness of traditional allocation strategies, it is necessary to develop an assembly tolerance allocation scheme based on multiple iterative algorithms. This would enable precise traceability of the overall machine fit tolerances to the assembly process, improve the controllability of the assembly process, and provide a robust and reliable solution. Summary of the Invention

[0005] Objective: This invention provides a high-precision method for conforming the shape of assembled components based on Monte Carlo large-scale model prediction. Applicable to aircraft assembly processes, it serves as a guiding scheme for precise and non-destructive connection of assembly components. This addresses the problem of low assembly quality caused by blind assembly in existing technologies without a hierarchical tolerance allocation strategy. It enables precise allocation of global tolerances to intermediate processes, effectively ensuring the effectiveness of the assembly process and its dynamic controllability.

[0006] Technical solution: A high-precision method for conforming the shape of assembled parts based on Monte Carlo large model prediction includes: Step 1: For a specific product, identify the multi-level components of the assembly design and extract the external dimensions of each level of components; Step 2: Based on the target external dimensions and manufacturing process capabilities, determine the manufacturing tolerances of each level of components after molding; Step 3: Based on the process instructions for assembling components at all levels into the completed assembly part, the assembly process is divided and the tolerance of each assembly process is defined as having a linear relationship with the manufacturing tolerance of the corresponding component after molding. Step 4: Based on the external dimensions of each level of components and the manufacturing tolerances of each level of components after forming, construct a prediction model for mating tolerances based on the Monte Carlo method, and clarify the contribution ranking of each level of assembly process tolerances to the overall product tolerance. Step 5: Randomly sample the fit tolerance coefficients to obtain the optimal fit tolerance coefficients; Step 6: Verify the validity of the prediction results using a visualization model. If valid, use the optimal solution as the implementation parameter to complete the assembly of the overall component product. If invalid, return to step 5.

[0007] Furthermore, in step 1, the external dimension x of the multi-level component is the thickness of each component along the normal direction of the assembly surface.

[0008] Furthermore, the manufacturing tolerance Y in step 2 is defined as follows: Step 21, the manufacturing tolerance of component 1 is Y1=a1x1; where the external dimension of component 1 is x1; Step 22, the manufacturing tolerance of component 2 is Y2 = a1x2; where the external dimension of component 2 is x2; The coefficients a1 and a2 are constants; if component 1 and component 2 are manufactured using the same process, then a1 = a2.

[0009] Furthermore, step 3 specifically includes: Step 31, the tolerance of the first assembly process Z1 = A1Y1; Step 32, the tolerance of the second assembly process Z2 = A2Y2 Where A and B are constants, and both are ≥1.

[0010] Furthermore, step 4, which involves constructing a prediction model for fit tolerances based on the Monte Carlo method, includes: Step 41: Build a Monte Carlo tree, set the assembly processes covered by the entire assembly process as Monte Carlo tree nodes, and initialize the tolerances of each assembly process. Step 42: Set up the prediction model. The information in each tree node includes the multi-level component dimensions x, the component manufacturing tolerance Y, and the assembly process tolerance Z. Step 43: Within the set total number of training iterations, train the Monte Carlo tree to obtain a trained Monte Carlo tree. Step 44: Based on the tree structure of the trained Monte Carlo tree, set a range for the output of each prediction model, take the range of the prediction model output as the training objective, evaluate the prediction model output that reaches the training objective each time, and obtain the scheme with the highest evaluation index as the optimal prediction model based on the Monte Carlo method under the current training conditions.

[0011] Furthermore, the sampling process in step 5 includes: Step 61, generate random numbers for the tolerance coefficients in assembly tolerance Z: A1=random[1,maxA1]; A2=random[1,maxA2]; ...; A n =random[1,maxA n ]; where n is the assembly level; max represents the upper limit of the standard tolerance; Step 62: Substitute the randomly generated tolerance coefficients A, B, ..., N into the assembly tolerance Z formula, i.e., Z1=A1Y1, Z2=A2Y2...; Step 63, assembly tolerances at all levels must meet the following requirements: If the assembly area involves one assembly component, its assembly tolerance should satisfy Z1≤W; if the assembly area involves two assembly components, its assembly tolerance should satisfy Z1+Z2≤W; and so on; where W is the tolerance index of the finished assembly component. Step 64: Based on a random algorithm, a sequence of Z1Z2…Z can be generated. n The data matrix serves as the database for training the Monte Carlo model.

[0012] Furthermore, the verification process of the visualized model in step 6 includes: Step 71, define the indicator function δ(x) i The output value is 0 or 1; if W i ≤W,δ(x i W = 1; i ≥W, δ(x) i )=0; where W i These are the tolerance indicators for the finished assembled parts predicted by the visualization model. Step 72, define the function f(x) i )=W i =Z xi ;x i Let i be the combination of the external dimensions of each component in the i-th test group, where i is a natural number greater than or equal to 1; Step 73, for each iteration process, the indicator function δ(x) must first be determined. i ) Does x satisfy ≠ 0? Calculate the function f(x) i ), output extreme value I=Min[f(x i)- f(x i+1 )-…]; Step 74: Perform calculations with I→0 as the optimization objective, and output the optimal assembly tolerance coefficients for each level.

[0013] Furthermore, in step 6, the validity of the prediction results is verified by examining the accuracy of the fitting of the response surface regression equation through the residual normal plot of the standardized effects, including: To determine whether the residual distribution after model fitting is uniform, if each sample data point satisfies the standard diagonal linear distribution and the distance deviation value is 5%, then it is judged to pass the residual normality test, and the model is effective.

[0014] Beneficial effects: In the field of aircraft assembly tolerance allocation, to improve the accuracy of assembly shape and thus ensure aircraft maneuverability and equipment reliability, this invention proposes a high-precision assembly component shape preservation method based on Monte Carlo large model prediction. This method uses Monte Carlo simulation to perform multi-assembly level and multi-key feature collaborative optimization, accurately and quantitatively determining the contribution of key features, key influencing factors, and their values ​​at each manufacturing stage. This enables the formation of a full-process error control strategy under the constraints of overall aircraft surface quality indicators, providing a theoretical basis for subsequent assembly tolerance allocation and inspection requirements. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A flowchart of a high-precision assembly component shape preservation method based on Monte Carlo large model prediction; Figure 2 Assembly schemes that do not assign tolerances based on Monte Carlo large model predictions; Figure 3 Assembly results for which tolerances were not assigned according to the Monte Carlo large model prediction; Figure 4 The assembly results are based on the tolerance allocation predicted by the Monte Carlo large model. Figure 5 This is a normal residual plot of the standardized effect of the response factor in the prediction model. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, 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 those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] The features and illustrative embodiments of various aspects of the present invention will now be described in detail. Numerous specific details are set forth in the following detailed description to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention may be practiced without requiring some of these specific details. The following description of embodiments is merely intended to provide a better understanding of the invention by illustrating examples of the invention. The invention is by no means limited to any specific setups and methods set forth below, but covers any improvements, substitutions, and modifications to structures, methods, and devices without departing from the spirit of the invention. Well-known structures and techniques are not shown in the drawings and the following description to avoid unnecessarily obscuring the invention.

[0019] In the description of this invention, it should be noted that the directions or positional relationships indicated by terms such as "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer" are based on the directions or positional relationships shown in the accompanying drawings and are only for the convenience of describing and simplifying the invention, and should not be construed as limiting the invention. Furthermore, the use of ordinal numbers (e.g., "first and second," etc.) is for distinguishing objects and is not limited to this order, and should not be construed as indicating or implying relative importance.

[0020] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly, encompassing both direct connection and indirect connection via an intermediate medium. Those skilled in the art can understand the specific meaning of these terms in this invention based on the specific circumstances.

[0021] It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other, and the various embodiments can be referenced and cited in each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0022] The present invention will be further described in detail below with reference to the embodiments and accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0023] A high-precision method for conforming the shape of assembled parts based on Monte Carlo large model prediction includes: Step 1: For a specific product, identify the multi-level components of the assembly design and extract the external dimensions of each level of components; Step 2: Based on the target external dimensions and manufacturing process capabilities, determine the manufacturing tolerances of each level of components after molding; Step 3: Based on the process instructions for assembling components at all levels into the completed assembly part, the assembly process is divided and the tolerance of each assembly process is defined as having a linear relationship with the manufacturing tolerance of the corresponding component after molding. Step 4: Based on the external dimensions of each level of components and the manufacturing tolerances of each level of components after forming, construct a prediction model for mating tolerances based on the Monte Carlo method, and clarify the contribution ranking of each level of assembly process tolerances to the overall product tolerance. Step 5: Randomly sample the fit tolerance coefficients to obtain the optimal fit tolerance coefficients; Step 6: Verify the validity of the prediction results using a visualization model. If valid, use the optimal solution as the implementation parameter to complete the assembly of the overall component product. If invalid, return to step 5.

[0024] Furthermore, in step 1, the external dimension x of the multi-level component is the thickness of each component along the normal direction of the assembly surface.

[0025] Furthermore, the manufacturing tolerance Y in step 2 is defined as follows: Step 21, the manufacturing tolerance of component 1 is Y1=a1x1; where the external dimension of component 1 is x1; Step 22, the manufacturing tolerance of component 2 is Y2 = a1x2; where the external dimension of component 2 is x2; The coefficients a1 and a2 are constants; if component 1 and component 2 are manufactured using the same process, then a1 = a2.

[0026] Furthermore, step 3 specifically includes: Step 31, the tolerance of the first assembly process Z1 = A1Y1; Step 32, the tolerance of the second assembly process Z2 = A2Y2 Where A and B are constants, and both are ≥1.

[0027] Furthermore, step 4, which involves constructing a prediction model for fit tolerances based on the Monte Carlo method, includes: Step 41: Build a Monte Carlo tree, set the assembly processes covered by the entire assembly process as Monte Carlo tree nodes, and initialize the tolerances of each assembly process. Step 42: Set up the prediction model. The information in each tree node includes the multi-level component dimensions x, the component manufacturing tolerance Y, and the assembly process tolerance Z. Step 43: Within the set total number of training iterations, train the Monte Carlo tree to obtain a trained Monte Carlo tree. Step 44: Based on the tree structure of the trained Monte Carlo tree, set a range for the output of each prediction model, take the range of the prediction model output as the training objective, evaluate the prediction model output that reaches the training objective each time, and obtain the scheme with the highest evaluation index as the optimal prediction model based on the Monte Carlo method under the current training conditions.

[0028] Furthermore, the sampling process in step 5 includes: Step 61, generate random numbers for the tolerance coefficients in assembly tolerance Z: A1=random[1,maxA1]; A2=random[1,maxA2]; ...; A n =random[1,maxA n ]; where n is the assembly level; max represents the upper limit of the standard tolerance; Step 62: Substitute the randomly generated tolerance coefficients A, B, ..., N into the assembly tolerance Z formula, i.e., Z1=A1Y1, Z2=A2Y2...; Step 63, assembly tolerances at all levels must meet the following requirements: If the assembly area involves one assembly component, its assembly tolerance should satisfy Z1≤W; if the assembly area involves two assembly components, its assembly tolerance should satisfy Z1+Z2≤W; and so on; where W is the tolerance index of the finished assembly component. Step 64: Based on a random algorithm, a sequence of Z1Z2…Z can be generated. n The data matrix serves as the database for training the Monte Carlo model.

[0029] Furthermore, the verification process of the visualized model in step 6 includes: Step 71, define the indicator function δ(x) i The output value is 0 or 1; if W i ≤W,δ(x i W = 1; i ≥W, δ(x) i )=0; where W i These are the tolerance indicators for the finished assembled parts predicted by the visualization model. Step 72, define the function f(x) i )=W i =Z xi ;x i Let i be the combination of the external dimensions of each component in the i-th test group, where i is a natural number greater than or equal to 1; Step 73, for each iteration process, the indicator function δ(x) must first be determined. i) Does x satisfy ≠ 0? Calculate the function f(x) i ), output extreme value I=Min[f(x i )- f(x i+1 )-…]; Step 74: Perform calculations with I→0 as the optimization objective, and output the optimal assembly tolerance coefficients for each level.

[0030] Furthermore, in step 6, the validity of the prediction results is verified by examining the accuracy of the fitting of the response surface regression equation through the residual normal plot of the standardized effects, including: To determine whether the residual distribution after model fitting is uniform, if each sample data point satisfies the standard diagonal linear distribution and the distance deviation value is 5%, then it is judged to pass the residual normality test, and the model is effective.

[0031] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and should not be construed as limiting the scope of the invention.

[0032] Figure 1 This is a flowchart illustrating a high-precision assembly component shape preservation method based on Monte Carlo large model prediction, provided in this embodiment. A product is considered as the assembly object, consisting of multiple assembly components. The external dimension of each component is defined as x, specifically referring to the thickness along the normal direction. The manufacturing tolerance of this component is typically highly correlated with its external thickness x; therefore, the manufacturing tolerance Y = ax is defined, where the coefficient a is related to the manufacturing process and molding material. For example, if components 1 and 2 are both shot-peened from 7050-T7451 aluminum alloy, then the manufacturing tolerance coefficient a of component 1 is equal to the manufacturing tolerance coefficient b of component 2. Components at each level are connected one by one according to the assembly steps in the process instructions. To ensure the effectiveness of the connection process, a fitting clearance is reserved based on the component manufacturing process. The cumulative fitting clearance tolerance Z at each level must meet the tolerance index W requirements of the finished product.

[0033] For example, an analysis of the pre-designed clearance between the wing panel and the wing rib is conducted. Figure 2This is an assembly scheme without tolerances assigned based on Monte Carlo large-scale model predictions. The wall panel body thickness is 11.09 mm, i.e., x1 = 11.09 mm; the stringer thickness is 7.85 mm, i.e., x2 = 7.85 mm. Since the wall panel and stringer are bonded together, the total thickness of the stringer assembly area is x3 = x1 + x2, i.e., x3 = 18.94 mm. Because both the wall panel and stringer are thermosetting laminated composites, the manufacturing tolerances are Y1 = ax1 and Y3 = ax3, where a = 3%, i.e., Y1 = 0.33 mm and Y3 = 0.56 mm. According to the assembly documents, the connecting component between the wall panel assembly and the stringer assembly is the rib flange. Since the rib assembly is manufactured using room temperature forming of metal, the forming precision of the rib is high, and it is usually used as a process reference during the assembly process. Therefore, the assembly process uses the rib as a standard part, connecting the stringer assembly and the wall panel assembly to the rib in the standard position. At this time, the forming tolerance of the rib can be ignored. Therefore, the fit tolerance of the wall panel is Z1=AY1; the fit tolerance of the stringer assembly is Z3=BY3. According to assembly experience, with A=1.2 and B=1.2, then Z1=0.396mm and Z3=0.5682mm.

[0034] Figure 3 The assembly results are based on the above scheme. Since this type of assembly unit uses a convex-concave meshing joint, after the wall panel is connected and fitted with the standard rib, there will always be a gap of not less than Z3-Z1=0.1722mm between the stringer area and the rib flange. When the assembly gap is greater than 0.15mm, it is easy for the stringer and skin to detach around the fasteners. Therefore, in this assembly state, there is a certain probability that some connection areas will experience detachment.

[0035] Figure 4 This is the assembly result based on the Monte Carlo large-scale model's hierarchical prediction of tolerance allocation. Using the same assembly background as before, since the component manufacturing process and molding dimensions remain unchanged, the wall panel thickness is 11.09 mm (x1 = 11.09 mm), and the stringer thickness is 7.85 mm (x2 = 7.85 mm). Because the wall panel and stringer are bonded together, the total thickness of the stringer assembly area is x3 = x1 + x2, i.e., x3 = 18.94 mm. Since both the wall panel and stringer are thermosetting laminated composites, the manufacturing tolerances are Y1 = ax1 and Y3 = ax3, where a = 3%, i.e., Y1 = 0.33 mm and Y3 = 0.56 mm. To avoid the risk of debonding between the stringer area and the wall panel, Z3 - Z1 < 0.15 mm, ideally approaching 0. Given Z1=AY1=0.33A and Z3=BY3=0.56B, the goal is to get 0.56B-0.33A→0. Generate random numbers with parameters A and B:

[0036] Each trial group Xi contains one parameter A and one parameter B, and the indicator function δ(x) is calculated.i ), determine the indicator function δ(x) i Does δ(x) satisfy ≠ 0? i If δ(x) = 0, then exclude this set of experimental parameters. i If x ≠ 0, calculate the function f(x). i )= Z xi =0.56B xi ;ϵ(x i )= Z xi =0.33A xi Output extreme value I=Min[f(x)] i )- ϵ(x i The calculation was performed with I→0 as the optimization objective. The results showed that A=0.91mm and B=0.53mm were a set of optimal solutions. At this point, a 0.3mm gap was reserved between the stringer and the flange, and a 0.3mm gap was also reserved between the wall panel and the flange. In this area, the gaps between the stringer, wall panel, and flange were the same. If the stringer thickened due to manufacturing tolerances, a gap would appear between the wall panel and the flange. When a gap appears at this point, the stringer is under pressure and will not cause debonding defects. Actual assembly site measurements showed that the gap at the wingtip side often resulted in good adhesion at the stringer, and no non-destructive defects were observed.

[0037] Figure 5 To predict the normal distribution of the standardized effect of the response factor in the model, the residual normal probability plot can be used to verify whether the residuals are normally distributed. The distribution of each data point in the normal probability plot of the residuals is a straight line, which shows that the model fitting error is extremely small, indicating the effectiveness, accuracy and reliability of the linear regression model.

[0038] According to the prediction results, under the optimal conditions, the fluctuation range of the actual assembly gap after three parallel tests was 0 ± 0.0051 mm, which is less than the required range of 0.15 mm, and the deviation from the target value was within 5%. This confirms the good correlation between the predicted value and the experimental value, indicating that the optimal processing and assembly process obtained by this method is reasonable.

[0039] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A high-precision method for conforming the shape of assembled parts based on Monte Carlo large model prediction, characterized in that, include: Step 1: For a specific product, identify the multi-level components of the assembly design and extract the external dimensions of each level of components; Step 2: Based on the target external dimensions and manufacturing process capabilities, determine the manufacturing tolerances of each level of components after molding; Step 3: Based on the process instructions for assembling components at all levels into the completed assembly part, the assembly process is divided and the tolerance of each assembly process is defined as having a linear relationship with the manufacturing tolerance of the corresponding component after molding. Step 4: Based on the external dimensions of each level of components and the manufacturing tolerances of each level of components after forming, construct a prediction model for mating tolerances based on the Monte Carlo method, and clarify the contribution ranking of each level of assembly process tolerances to the overall product tolerance. Step 5: Randomly sample the fit tolerance coefficients to obtain the optimal fit tolerance coefficients; Step 6: Verify the validity of the prediction results using a visualization model. If valid, use the optimal solution as the implementation parameter to complete the assembly of the overall component product. If invalid, return to step 5.

2. The high-precision assembly component shape preservation method based on Monte Carlo large model prediction according to claim 1, characterized in that, In step 1, the external dimension x of the multi-level components is the thickness of each component along the normal direction of the assembly surface.

3. The high-precision assembly component shape preservation method based on Monte Carlo large model prediction according to claim 1, characterized in that, The manufacturing tolerance Y in step 2 is defined as follows: Step 21, the manufacturing tolerance of component 1 is Y1=a1x1; where the external dimension of component 1 is x1; Step 22, the manufacturing tolerance of component 2 is Y2 = a1x2; where the external dimension of component 2 is x2; The coefficients a1 and a2 are constants; if component 1 and component 2 are manufactured using the same process, then a1 = a2.

4. The high-precision assembly component shape preservation method based on Monte Carlo large model prediction according to claim 1, characterized in that, Step 3 specifically includes: Step 31, the tolerance of the first assembly process Z1 = A1Y1; Step 32, the tolerance of the second assembly process Z2 = A2Y2 Where A and B are constants, and both are ≥1.

5. The high-precision assembly component shape preservation method based on Monte Carlo large model prediction according to claim 1, characterized in that, Step 4, which involves constructing a prediction model for fit tolerances based on the Monte Carlo method, includes: Step 41: Build a Monte Carlo tree, set the assembly processes covered by the entire assembly process as Monte Carlo tree nodes, and initialize the tolerances of each assembly process. Step 42: Set up the prediction model. The information in each tree node includes the multi-level component dimensions x, the component manufacturing tolerance Y, and the assembly process tolerance Z. Step 43: Within the set total number of training iterations, train the Monte Carlo tree to obtain a trained Monte Carlo tree. Step 44: Based on the tree structure of the trained Monte Carlo tree, set a range for the output of each prediction model, take the range of the prediction model output as the training objective, evaluate the prediction model output that reaches the training objective each time, and obtain the scheme with the highest evaluation index as the optimal prediction model based on the Monte Carlo method under the current training conditions.

6. The high-precision assembly component shape preservation method based on Monte Carlo large model prediction according to claim 1, characterized in that, The sampling process in step 5 includes: Step 61, generate random numbers for the tolerance coefficients in assembly tolerance Z: A1=random[1,maxA1]; A2=random[1,maxA2]; ...; A n =random[1,maxAn]; where n is the assembly level; max represents the upper limit of the standard tolerance; Step 62: Substitute the randomly generated tolerance coefficients A, B, ..., N into the assembly tolerance Z formula, i.e., Z1=A1Y1, Z2=A2Y2...; Step 63, assembly tolerances at all levels must meet the following requirements: If the assembly area involves one assembly component, its assembly tolerance should satisfy Z1≤W; if the assembly area involves two assembly components, its assembly tolerance should satisfy Z1+Z2≤W; and so on; where W is the tolerance index of the finished assembly component. Step 64: Based on a random algorithm, a sequence of Z1Z2…Z can be generated. n The data matrix serves as the database for training the Monte Carlo model.

7. The high-precision assembly component shape preservation method based on Monte Carlo large model prediction according to claim 1, characterized in that, The verification process of the visualized model in step 6 includes: Step 71, define the indicator function δ(x) i The output value is 0 or 1; if W i ≤W,δ(x i W = 1; i ≥W, δ(x) i )=0; where W i These are the tolerance indicators for the finished assembled parts predicted by the visualization model. Step 72, define the function f(x) i )=W i =Z xi ;x i Let i be the combination of the external dimensions of each component in the i-th test group, where i is a natural number greater than or equal to 1; Step 73, for each iteration process, the indicator function δ(x) must first be determined. i ) Does x satisfy ≠ 0? Calculate the function f(x) i ), output extreme value I=Min[f(x i )- f(x i+1 )-…]; Step 74: Perform calculations with I→0 as the optimization objective, and output the optimal assembly tolerance coefficients for each level.

8. The high-precision assembly component shape preservation method based on Monte Carlo large model prediction according to claim 1, characterized in that, Step 6, which verifies the validity of the prediction results, uses the residual normal plot of the standardized effects to test the accuracy of the fit of the response surface regression equation, including: To determine whether the residual distribution after model fitting is uniform, if each sample data point satisfies the standard diagonal linear distribution and the distance deviation value is 5%, then it is judged to pass the residual normality test, and the model is effective.