A method for evaluating the influence degree of design change of automobile exterior parts based on small-batch trial color spectrum chain search

Abstract

This invention discloses a method for assessing the impact of design changes on automotive exterior parts based on a chain-like search of color spectrum in small-batch trial production, belonging to the field of automotive manufacturing technology. This invention establishes a hierarchical coupling relationship between the substrate and each color version, as well as a coupling relationship between color versions based on coating process dependence. It maps geometric design changes across domains to film thickness drift in each color version, then performs a chain-like search and propagation attenuation along the coupling relationship to calculate color difference exceedances, film thickness exceedances, and rework cost indices for each color version. These are then weighted to synthesize the final change impact, classify risk levels, and output an assessment report. This invention enables quantitative prediction of the coupled impact of design changes on the coating quality of multiple color versions, providing an efficient and quantifiable assessment tool for engineering change decisions in the small-batch trial production stage, significantly reducing trial production verification costs and time.

Description

Technical Field

[0001] This invention relates to the field of automotive manufacturing technology, and more specifically, to a method for evaluating the impact of design changes on automotive exterior parts based on a chain search of color spectrum in small-batch trial production. Background Technology

[0002] In the development of automotive products, the color design of exterior components (including bumpers, wheel arches, door sill protectors, tailgate trim, grille trim, etc.) is a key factor affecting the overall appearance quality and brand recognition of a vehicle. With the increasing demand for personalized automotive exteriors and the continuous evolution of automotive brand family design languages, color schemes for automotive exterior components are becoming increasingly diversified. On the same base (i.e., the same exterior component product), multiple active color versions often exist simultaneously, corresponding to different model configurations, regional markets, or user optional extras.

[0003] During the small-batch trial production phase, the base geometry design of the exterior components is still undergoing continuous iterative optimization. Changes in geometry, such as wall thickness adjustment, draft angle correction, parting line displacement, and modification of the fillet radius of the snap-fit ​​structure, directly alter the surface curvature of the exterior components. These geometric changes, in turn, affect the flow and deposition behavior of the paint during the coating process, causing the film thickness distribution of the same color version to drift in different zones. Changes in film thickness distribution further affect the final color appearance—when the film thickness deviates from the target value, color difference (…) This will exceed the tolerance range, resulting in substandard appearance quality.

[0004] What's more complex is that, because there are multiple active color versions on the exterior components, these different color versions often share some coating types, paint formulations, or spraying production lines. This means that changes to the substrate geometry design will not only affect the coating quality of a single color version, but may also affect other color versions through coupling relationships at the coating process level—such as sharing the undercoat or using the same spraying fixture—creating a chain reaction where "one change affects everything." This cross-color version coupling and propagation effect makes the impact assessment of design changes exceptionally complex.

[0005] Currently, the existing technologies for assessing the impact of changes in the color design of automotive exterior parts mainly fall into the following categories:

[0006] Category 1: Evaluation methods based on color difference data management.

[0007] For example, Chinese patent application CN116051310A discloses a method, apparatus, terminal device, and medium for managing color difference data in passenger vehicles. It determines the color difference threshold range of color difference data for passenger vehicle components and adjusts the component coloring strategy when the color difference data exceeds the threshold, thereby achieving structured management and anomaly warning for color difference data. However, this method is a passive response mechanism after a color difference problem has occurred, and cannot predict it before design changes occur; furthermore, it only processes the color difference data of a single component and does not involve the evaluation of the coupling effects between multiple color versions.

[0008] The second category: structured methods based on color configuration / BOM management.

[0009] For example, Chinese patent application CN115311039A discloses a method, apparatus, device, and storage medium for generating and calculating vehicle color parts. It constructs a color configuration mark library and a color matrix configuration, and correlates color parts (neutral parts + styling parts) with vehicle model color configurations to achieve structured generation of the color part bill of materials (BOM). This type of method achieves structured data organization of color parts at the BOM / configuration management level, but it falls under the scope of bill of materials and configuration management, does not involve the calculation and prediction of physical quantities in the painting process, and cannot assess the impact of substrate geometric changes on painting quality.

[0010] Category 3: Change impact assessment methods based on knowledge graphs.

[0011] For example, Chinese patent CN119829389B discloses a method and system for assessing the impact of operational changes based on knowledge graphs. It constructs a directed knowledge graph (nodes + edges + relationships), uses depth-first search to query the impact paths of changes, and outputs the impact assessment results based on historical anomaly data and relational aggregation. While this type of method adopts the technical paradigm of "knowledge graph + change path search," its application domain is IT operations and maintenance (nodes are services / hosts / databases, edges are call / connection relationships), and its propagation weights are abstract correlation degrees or anomaly probabilities, completely neglecting cross-domain mapping of physical quantities. Therefore, it cannot be applied to assessing the physical impact of substrate geometry changes on coating quality in the automotive painting process.

[0012] Category 4: Film thickness analysis methods based on geometry-coating simulation.

[0013] Existing technologies also include simulation analysis methods for coating film thickness. For example, Chinese patent CN115715244B discloses a programming method for spraying equipment and corresponding spraying equipment. This method acquires the geometric data of the component to be sprayed, simulates the robot's spraying path and spraying mode, and determines the spraying mode data representing the layer thickness distribution, thereby achieving offline programming of the spraying robot. Furthermore, Dürr's DXQ3D.onsite module, released in 2020, also provides a digital film thickness calculation solution based on a three-dimensional digital model. Academic research also includes film thickness simulation and control studies for robotic spraying of vehicle body surfaces. However, the purpose of these methods is to plan the spraying robot path and optimize spraying process parameters (i.e., "how to spray"), rather than assessing the impact of design changes on the quality of multi-color version prototypes (i.e., "how the changes affect the product"); and their processing objects are single components or single coating schemes, without involving the evaluation of coupling propagation between multiple color versions.

[0014] In summary, existing technologies lack a method to quantitatively predict the coupled impact of substrate geometry design changes on the coating quality of multiple active color versions during the small-batch trial production stage, before design changes occur. Designers in the small-batch trial production stage often have to rely on experience for qualitative judgments or conduct independent test sprays for each color version, which is inefficient and costly, and fails to meet the industry's increasingly shorter automotive development cycles. Summary of the Invention

[0015] In view of the above-mentioned prior art, the technical problem to be solved by the present invention is: how to quantitatively predict the coupled impact of the design change on the coating quality of multiple active color versions during the small-batch trial production stage, before the change of the substrate geometry design occurs, that is: to transform the "geometric design change" into the "physical impact at the coating process level", and further predict the propagation effect of the impact among different color versions, thereby providing a quantitative evaluation basis for engineering change decisions.

[0016] To address the above problems, this invention provides a method for evaluating the impact of design changes on automotive exterior parts based on a chain search of color spectrum in small-batch trial production, comprising the following steps:

[0017] S1. Establish the coating process coupling relationship between the exterior trim substrate and multiple active color versions.

[0018] Obtain the substrate geometry model of the automotive exterior parts and multiple active color versions corresponding to this substrate during the small-batch trial production stage. Each color version is associated with its own coating layer stack structure, film thickness tolerance window, and color difference tolerance threshold. Using the substrate geometry model as the substrate object and each color version as the color version object, establish the membership coupling relationship between the substrate object and each color version object (representing that the coating system of the color version object is attached to the substrate object), as well as the coupling relationship between each color version object based on coating process dependency (determined according to the number of shared coating types, paint formulation similarity, and whether they share a spraying production line). The coupling relationship is characterized by coupling strength.

[0019] Preferably, the coupling strength is determined in the following manner:

[0020]

[0021] in, For color version With color version The coupling strength between them , These are collections of coating types for two color versions. For coating Jaccard similarity, The similarity of the formulations for the two color versions is obtained by measuring the cosine similarity of the pigment component content vectors or the distance in Lab color space coordinates. A value of 1 indicates that both color versions share the same spraying production line; otherwise, a value of 0 indicates that the color versions share the same spraying production line. , , Preset weighting coefficients and satisfying .

[0022] S2. Map the changes in the substrate geometry design across domains to the coating film thickness drift for each color version.

[0023] The geometric design changes applied to the substrate geometry model are obtained. The external surface deviation field before and after the changes is calculated using a parametric 3D model. The surface of the exterior trim is divided into multiple spraying zones, and the local geometric distortion intensity, local curvature, and distance to the nearest edge of each zone are calculated. The geometric distortion intensity of each zone is mapped to the film thickness drift of each color version in each zone through a coating process mapping function. This mapping function takes the geometric distortion intensity as the input variable and the film thickness drift as the output variable, and the mapping relationship of each color version is calibrated differently depending on the number of coating layers.

[0024] Preferably, the coating process mapping function is implemented using a regression model:

[0025]

[0026] in, For the first The color version is in the first The film thickness drift of each spraying zone For local geometric distortion intensity, For local curvature, The distance to the nearest edge, To prevent division by a constant, For the first The regression coefficients of the intercept term corresponding to the paint type of each color version. , , As common regression coefficients, each coefficient is calibrated in groups according to color version coating type. The calibration method is to obtain the coefficients by performing multiple linear regression or nonlinear least squares fitting on the geometric adjustment records and corresponding measured film thickness change records in the historical small batch trial production dataset.

[0027] S3. Perform a chain search along the coupling relationship and calculate the impact of the change node by node.

[0028] Using the base object as the search entry point, the propagation strength of each color version object is initialized along its membership coupling relationship, and a chained search is performed. For the currently accessed color version object, the color difference offset is predicted based on its film thickness drift, and the color difference out-of-bounds index, film thickness out-of-bounds index, and rework cost index are calculated and weighted to synthesize the direct change impact of the color version object. Propagation is then extended to other color version objects along coupling associations where the coupling strength meets a preset threshold condition. The propagation strength is attenuated and pruned based on the cumulative value of the film thickness drift. The direct change impact and the coupling propagation impact are summarized to obtain the final change impact of each color version object.

[0029] Preferably, the predicted color difference offset Determine using the following formula:

[0030]

[0031] in, This is a set of color difference sensitive zones, including at least edge zones and curvature abrupt change zones. The color difference sensitivity coefficient is set according to the color version type: the sensitivity coefficient of metallic paint and pearlescent paint is greater than that of solid color paint.

[0032] Preferably, the chain search adopts priority-driven traversal: color version objects directly reachable by membership coupling are sorted in descending order of direct change impact and placed into the traversal queue. Each time, the highest priority unprocessed object is taken out for processing. Adjacent objects are only explored along coupling associations with coupling strength greater than a preset threshold. Newly enqueued objects are marked with priority according to the degree of coupling propagation impact they are affected by. The traversal terminates when the queue is empty.

[0033] Preferably, the propagation intensity is attenuated based on the cumulative value of the film thickness drift: when propagating to an adjacent color version object, the increment of the propagation intensity is proportional to the coupling intensity and inversely proportional to the cumulative value of the absolute value of the film thickness drift of the current object, and it continuously attenuates along the propagation chain until it is below the pruning threshold.

[0034] Preferably, when the color difference exceeding the limit, film thickness exceeding the limit, and rework cost exceeding the limit are weighted and combined to form the direct change impact, the weights of the three indicators are dynamically adjusted according to the evaluation scenario: in the rapid screening scenario, the weights of the color difference exceeding the limit and the rework cost exceeding the limit are higher than those of the film thickness exceeding the limit; in the process verification scenario, the weight of the film thickness exceeding the limit is higher than the other two; in the comprehensive evaluation scenario, the weights of the three indicators are equal.

[0035] Preferably, the method further includes a trial production iteration calibration step: after the current round of trial sampling is completed, the measured color difference value and measured film thickness value of the color version object are obtained, and the residual between the measured value and the predicted value is calculated; when the absolute value of the residual is greater than the preset calibration threshold, the regression coefficient of the coating process mapping function is corrected with the measured value as feedback, so that the prediction error of subsequent rounds converges; each color version object carries a trial production iteration round mark, and the coupling strength is incrementally updated after each formula revision.

[0036] S4. Output the evaluation report.

[0037] The risk level is divided according to the score range of the final change impact of each color version object, and an assessment report is output including the color version identifier, predicted color difference offset, final change impact and risk level.

[0038] Preferably, the risk levels include low-risk, medium-risk, and high-risk zones: if the final impact of the change falls within the first score range, it is considered low-risk, and the change is approved; if it falls within the second score range, it is considered medium-risk, requiring additional verification spraying and color difference retesting, and the change is conditionally approved; if it falls within the third score range, it is considered high-risk, requiring re-sampling or formula adjustment, and the change triggers an engineering change request; the three score ranges increase sequentially and do not overlap.

[0039] Preferably, it also includes a two-stage correlation analysis step between trial production and mass production: establishing a correlation mapping relationship between the impact of changes to each color version object in the trial production stage and the impact of changes to the same type of color version in the mass production stage; when the current evaluation is a change evaluation in the trial production stage, the potential impact prediction on the mass production stage is output synchronously based on the correlation mapping relationship.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1. This invention, by establishing a quantitative mapping function (regression model) between geometric distortion intensity and film thickness drift, achieves for the first time end-to-end quantitative calculation of film thickness distribution for each color version, from changes in substrate geometric design to the overall coating thickness distribution. This overcomes the limitations of traditional methods that treat "color management" and "geometric design" separately. The mapping function is calibrated using historical small-batch trial production data, considers nonlinear corrections for local curvature and edge effects, and possesses clear physical meaning and verifiability.

[0042] 2. This invention constructs a substrate-color version coupling relationship (membership coupling relationship + coupling association based on coating process dependence), quantitatively characterizing the process dependencies such as shared coating type, formulation similarity, and collinear spraying among color versions. A chain search algorithm is then used to assess the propagation of design change impacts along the coupling relationship. The propagation intensity is physically driven by the cumulative value of film thickness drift, making the previously elusive chain effects of multiple color versions calculable, traceable, and predictable, filling a gap in existing technologies in this area.

[0043] 3. First, this invention can quickly output the impact score and risk level of each color version without actual test spraying, significantly reducing the number of invalid sampling rounds and verification test sprays in the trial production stage, thereby significantly reducing trial production costs and shortening the trial production cycle. Second, the trial production iterative calibration steps enable the prediction model to continuously self-optimize as trial production data accumulates, continuously improving prediction accuracy. Furthermore, the two-stage correlation analysis between trial production and mass production allows the evaluation results of the trial production stage to provide predictive references for the mass production stage, realizing knowledge transfer from trial production to mass production. This invention can be widely applied to the design change impact assessment of small-batch trial production of various automotive exterior parts (bumpers, wheel arches, door sill protectors, tailgate trim, grille trim, etc.), possessing cross-model and cross-project versatility and scalability. Attached Figure Description

[0044] Figure 1 This is a schematic diagram of the overall process of the method for evaluating the impact of design changes on automotive exterior parts based on small-batch trial production color spectrum chain search. Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of protection of this invention.

[0046] Please see Figure 1 This embodiment provides a method for evaluating the impact of design changes on automotive exterior parts based on a chain search of color spectrum in small-batch trial production, which includes the following steps S1 to S4.

[0047] Step S1: Establish the coating process coupling relationship between the exterior trim substrate and multiple active color versions.

[0048] Taking the exterior bumper as an example, it had three active color versions during the small-batch trial production stage: solid white ( ), metallic silver ( Pearl Red ).

[0049] First, obtain the base geometry model of the bumper, as well as the paint layer structure, film thickness tolerance window, and color difference tolerance threshold associated with each color version. The parameters for each color version are as follows:

[0050] (Plain white) solid color paint {Primer, Color Paint, Clear Varnish} ±8μm 1.5 (metallic silver) Metallic paint {Primer, Color Paint, Clear Varnish} ±6μm 1.0 (Pearl Red) Pearl paint {Primer, Color Paint, Pearl Finish, Clear Varnish} ±5μm 0.8

[0051] Then, using the base geometry model as the base object Each color version is used as a color version object. , , Establish the following two types of coupling relationships:

[0052] (1) Membership coupling relationship: All three color versions of the coating system are attached to the base object. (That is, the painting process for each color version is implemented based on the base geometry), using directed edges. express.

[0053] (2) Coupling based on coating process dependence: The coupling strength is calculated based on the number of coating types shared between the two color versions, the similarity of the paint formulations, and whether they share the same spraying production line. :

[0054]

[0055] in, For coating Jaccard similarity, Formula similarity (obtained by cosine similarity of pigment component content vectors or distance in Lab color space coordinates). A value of 1 indicates that the same spraying production line is used; otherwise, a value of 0 indicates that the line is used. , , Preset weighting coefficients (preferred values) , , ).

[0056] In this embodiment, the three color versions share the same spraying production line ( Coating Jaccard similarity: and All , and for , and for Recipe similarity (via Lab coordinate distance transformation): , , Substituting into the formula, the coupling strength is calculated as follows: , , .

[0057] Step S2: Map the changes in substrate geometry across domains to the coating film thickness drift for each color version.

[0058] In this embodiment, the base geometry design is changed so that the draft angle of the lower edge of the front bumper is adjusted from 3° to 5° (draft angle adjustment amount). ).

[0059] First, the external surface deviation field before and after the change is calculated using a parametric 3D model (such as CATIA). The key area for deviation is the lower edge flange area, with a maximum deviation of approximately 0.8mm. The exterior trim surface is divided into 30 painting zones. .

[0060] Then, the local geometric distortion intensity of each partition is calculated:

[0061]

[0062] and local curvature and distance to the nearest edge By partition (For example, the left side of the bottom edge folded over): , , .

[0063] Finally, the geometric distortion intensity is mapped to the film thickness drift of each color version in each zone using a coating process mapping function (implemented using a regression model):

[0064]

[0065] in To prevent division by a constant (0.5mm), regression coefficients were calibrated by color version and coating type. The calibration method was as follows: using at least 50 sets of geometric adjustment records and corresponding measured zone film thickness variation records from the historical small-batch trial production dataset of this exterior part, multiple linear regression or nonlinear least squares fitting was performed. Typical ranges for regression coefficients of each coating type are as follows:

[0066] Solid color paint group 0.5～2.0 0.8～1.5 -0.5～0.1 0.5～2.0 Metallic paint set 0.3～1.5 0.3～0.8 -0.3～0.2 0.3～1.5 Pearl / Triple Coating Set 0.2～1.0 0.2～0.5 -0.2～0.3 0.2～1.0

[0067] The above numerical ranges are empirical values ​​obtained by the inventors through regression calibration in multiple small-batch trial production projects of exterior parts. Actual values ​​should be based on the calibration results of specific projects. Substituting the median values ​​of each group, the calculations for each color version in the respective zones are obtained. Film thickness drift: (plain white), (metallic silver) (Pearl Red).

[0068] Step S3: Perform a chain search along the coupling relationship and calculate the impact of the change node by node.

[0069] With base object As the search entry point, along the membership coupling relationship Initialize the propagation intensity of each color version (Pick For each color version, predict the color difference shift based on the film thickness drift, and calculate the direct change impact; then, along the coupling strength greater than a preset threshold ( The coupling correlation is taken as 0.2 to 0.4) for propagation, and the propagation intensity is attenuated and pruned according to the cumulative value of the film thickness drift.

[0070] (a) Color difference offset prediction

[0071] Color difference offset Determine using the following formula:

[0072]

[0073] in This is a set of color difference-sensitive zones, including at least edge zones and zones with abrupt curvature changes. The criterion is: the distance from the centroid of a zone to its nearest edge. Less than the preset edge distance threshold When it is determined to be an edge-sensitive partition; when the local curvature is... Greater than the preset curvature threshold It was determined to be a region sensitive to curvature abrupt changes. and The value is related to the coating type: solid color paint , Metallic paint , Pearl paint , .

[0074] The color difference sensitivity coefficient is obtained through DOE experimental calibration or calculation using the Kubelka-Munk theoretical model. Specifically, the DOE calibration method involves intentionally introducing different film thickness deviations in each zone (achieved by adjusting the spray flow rate or spray gun speed), measuring the corresponding color difference offset, and then using the linear regression slope as the sensitivity coefficient. Typical sensitivity coefficient ranges for different coating types: solid color paint. Metallic paint Pearl paint .

[0075] In this embodiment, Color difference sensitive zones include (edge ​​flange area) and , (Area of ​​abrupt curvature change in the front face), with a sensitivity coefficient of 1.8, the calculation results are as follows: Exceeding .

[0076] (ii) Calculation of the impact of direct changes

[0077] The direct impact of the change is weighted and synthesized by the color difference exceeding the limit index, the film thickness exceeding the limit index, and the rework cost index:

[0078]

[0079] The weights are dynamically adjusted based on the evaluation scenario: in a rapid screening scenario , , In the process verification scenario , , ; Comprehensive evaluation scenario .

[0080] Color difference exceeding the limit:

[0081]

[0082] when The maximum value is 100.

[0083] Film thickness exceeding limits:

[0084]

[0085] in This represents the upper limit of film thickness tolerance.

[0086] Rework cost indicator:

[0087]

[0088] in The cost of re-producing a sample for this color version. This is the normalized benchmark cost.

[0089] In this embodiment of , , (Comprehensive evaluation scenario), weighted .

[0090] (iii) Chain propagation

[0091] Chain search uses priority-driven traversal: color version objects directly reachable by membership coupling are sorted in descending order of direct change impact and placed into the traversal queue. Each time, the highest priority unprocessed object is taken out for processing. Adjacent objects are only explored along coupling associations with coupling strength greater than a preset threshold. Newly enqueued objects are marked with priority according to the degree of coupling propagation impact they are affected by. Traversal terminates when the queue is empty.

[0092] The propagation intensity decays based on the cumulative value of the film thickness drift: when propagating to an adjacent color version object, the increment of the propagation intensity is proportional to the coupling strength and inversely proportional to the cumulative value of the absolute value of the current object's film thickness drift. Specifically, from the node... Along the coupling correlation direction During propagation, the increment is calculated using the following formula:

[0093]

[0094] in The attenuation coefficient (take) ), This represents the average absolute value of the film thickness drift. When the propagation intensity is below the pruning threshold... (Pick Prune the branch when the number of branches reaches 5% to 10%, and terminate the branch. Maximum search depth Take 5% to 10%.

[0095] In this embodiment, from Towards spread:

[0096]

[0097] from Towards spread:

[0098]

[0099] The final impact of the changes at each node is summarized below:

[0100] (Plain white) 75.2 68.6 143.8 High risk (metallic silver) 81.1 0 (Direct Matrix Activation) 81.1 High risk (Pearl Red) 88.5 59.1 147.6 High risk

[0101] Step S4: Output the evaluation report

[0102] Risk levels are categorized based on the final impact score range of each color version object: Low-risk zone. (Change of approval); Medium-risk area (Additional verification spraying and color difference retesting are required; changes are conditionally permitted); High-risk areas (A new sample must be produced or the formula adjusted, triggering an engineering change request). The above interval division is based on the inventors' experience data from multiple small-batch trial production projects of exterior parts—when At that time, the color difference pass rate verified by actual test spraying was ≥95%; when At that time, the pass rate was 60%–95%; At that time, the pass rate was less than 60%.

[0103] The evaluation report output in this embodiment is as follows:

[0104] (Plain white) solid color paint 1.72 12.5 143.8 High risk A new sample must be produced or the formula adjusted, triggering an engineering change request. (metallic silver) Metallic paint 1.59 13.8 81.1 High risk A new sample must be produced or the formula adjusted, triggering an engineering change request. (Pearl Red) Pearl paint 0.91 15.2 147.6 High risk A new sample must be produced or the formula adjusted, triggering an engineering change request.

[0105] The key impact path is: ( (maximum coupling strength), and ( In this embodiment, all three color versions are identified as high-risk. It is recommended to reduce the draft angle adjustment from 2° to 1°, or to make targeted adjustments to the formulas for each color version.

[0106] Supplementary Implementation Method 1: Trial Production and Iterative Calibration

[0107] After the current round of trial sampling is completed, obtain the measured color difference value of the color version object. and measured film thickness Calculate the residual between the measured value and the predicted value. When the absolute value of the residual is greater than the preset calibration threshold (Plain color paint) Metallic paint Pearl lacquer , Calibration is triggered when the color difference tolerance threshold is reached.

[0108] The recursive least squares (RLS) method is used to correct the regression coefficients online. Let the first... The round of obtaining new observational data ,in:

[0109]

[0110]

[0111] Prediction error:

[0112]

[0113] Gain vector:

[0114]

[0115] Regression coefficient update:

[0116]

[0117] Covariance matrix update:

[0118]

[0119] in The forgetting factor is set to 0.95. Each color version object carries a prototype iteration number marker. The formula similarity was recalculated after the formula was revised. Coupling strength Incremental update.

[0120] Supplementary Implementation Method Two: Correlation Analysis of Trial Production and Mass Production Stages

[0121] Establish a correlation mapping between the impact of changes to each color version during the trial production phase and the impact of changes to the same type of color version during the mass production phase. Collect paired data (trial production phase evaluation values) from historical projects. and actual value in mass production stage (at least 30 groups) For input, Establish a linear regression model for the output:

[0122]

[0123] In this context, #imgpt203# represents the maximum value of the final change impact for each color version, and #imgpt204# represents the value. Parameters #imgpt205# and #imgpt206# (typical range: #imgpt207#, #imgpt208#) are estimated using the least squares method. When the current assessment is a change assessment for the trial production stage, the potential impact prediction for the mass production stage is simultaneously output using a mapping model, #imgpt209#.

[0124] Verification results

[0125] After the above assessment was completed, the front bumper underwent small-batch trial production for verification. The actual test results are compared with the predicted results as follows:

[0126] #imgpt212# (Plain White) 12.5 11.8 +0.7 #imgpt213# (Metallic Silver) 13.8 14.2 -0.4 #imgpt214# (Pearl Red) 15.2 16.1 -0.9

[0127] The prediction errors were all within #imgpt215#, with a maximum error of 0.9, verifying the effectiveness of the invention. In contrast, if the traditional method of testing each color version is used, each of the three color versions requires at least two rounds of testing (a total of six rounds), with each round taking about three days, for a total of about 18 days and a total cost of about 300,000 yuan. However, the method of this invention only requires one CAD model calculation and algorithm run (taking about 2 hours) to output the prediction results for all color versions, significantly shortening the trial production cycle and reducing the trial production cost.

[0128] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Various changes made within the scope of knowledge possessed by those skilled in the art without departing from the concept of the present invention still fall within the scope of protection of the present invention.

Claims

1. A method for evaluating the impact of design changes on automotive exterior parts based on a chain search of color spectrum in small-batch trial production, characterized in that, Includes the following steps: S1. Establish the coating process coupling relationship between the exterior trim base and multiple active color versions: Obtain the substrate geometry model of the automotive exterior parts and multiple active color versions of the substrate during the small-batch trial production stage. Each color version is associated with its respective coating layer structure, film thickness tolerance window and color difference tolerance threshold. Using the base geometry model as the base object and each color version as the color version object, establish the membership coupling relationship between the base object and each color version object, as well as the coupling association between each color version object based on the coating process dependency, and the coupling association is characterized by the coupling strength. S2. Map the changes in substrate geometry across domains to the coating thickness drift for each color version: Obtain the geometric design change amount acting on the substrate geometric model, calculate the outer surface deviation field before and after the change from the parametric three-dimensional model, divide the surface of the exterior trim into multiple spraying zones, and calculate the local geometric distortion intensity of each zone. The geometric distortion intensity of each zone is mapped to the film thickness drift of each color version in each zone through the coating process mapping function. The coating process mapping function takes the geometric distortion intensity as the input variable and the film thickness drift as the output variable. S3. Perform a chain search along the coupling relationship and calculate the impact of the change node by node: Using the base object as the search entry point, the propagation intensity of each color version object is initialized along the membership coupling relationship, and a chain search is performed; For the currently accessed color version object, predict the color difference offset based on its film thickness drift, calculate the color difference out-of-bounds index, film thickness out-of-bounds index and rework cost index, and weight the synthesis of the direct change impact of the color version object. Propagate along the coupling associations that meet the preset threshold conditions to other color version objects. The propagation strength is attenuated and pruned based on the cumulative value of the film thickness drift. By summing the direct change impact and the coupled propagation impact, the final change impact of each color version object is obtained; the coupled propagation impact is the impact component that propagates to the color version object through coupling association. S4. Output Evaluation Report: The risk level is divided according to the score range of the final change impact of each color version object, and an assessment report is output including the color version identifier, predicted color difference offset, final change impact and risk level.

2. The method according to claim 1, characterized in that, The coupling strength mentioned in step S1 Determine as follows: in, For color version With color version The coupling strength between them , These are collections of coating types for two color versions. For coating Jaccard similarity, The similarity of the formulations for the two color versions is obtained by measuring the cosine similarity of the pigment component content vectors or the distance in Lab color space coordinates. A value of 1 indicates that both color versions share the same spraying production line; otherwise, a value of 0 indicates that the color versions share the same spraying production line. , , Preset weighting coefficients and satisfying .

3. The method according to claim 1, characterized in that, The coating process mapping function described in step S2 is implemented using a regression model: in, For the first The color version is in the first The film thickness drift of each spraying zone For local geometric distortion intensity, For local curvature, The distance to the nearest edge, To prevent division by a constant, For the first The regression coefficients of the intercept term corresponding to the paint type of each color version. , , As common regression coefficients, each coefficient is calibrated in groups according to color version coating type. The calibration method is to obtain the coefficients by performing multiple linear regression or nonlinear least squares fitting on the geometric adjustment records and corresponding measured film thickness change records in the historical small batch trial production dataset.

4. The method according to claim 1, characterized in that, The predicted color difference offset Determine by the following formula: in, This is a set of color difference sensitive zones, including at least edge zones and curvature abrupt change zones. The color difference sensitivity coefficient is set according to the color version type: the sensitivity coefficient of metallic paint and pearlescent paint is greater than that of solid color paint.

5. The method according to claim 1, characterized in that, The chain search described in step S3 uses priority-driven traversal: color version objects directly reachable by membership coupling are sorted in descending order of direct change impact and placed into the traversal queue. Each time, the highest priority unprocessed object is taken out for processing. Adjacent objects are only explored along coupling associations with coupling strength greater than a preset threshold. Newly enqueued objects are marked with priority according to the coupling propagation impact they receive. The traversal terminates when the queue is empty.

6. The method according to claim 1, characterized in that, The propagation intensity described in step S3 is attenuated based on the cumulative value of the film thickness drift: when propagating to an adjacent color version object, the increment of the propagation intensity is proportional to the coupling intensity and inversely proportional to the cumulative value of the absolute value of the film thickness drift of the current object, and it is continuously attenuated along the propagation chain until it is lower than the pruning threshold.

7. The method according to claim 1, characterized in that, It also includes a trial production iterative calibration step: after the current round of trial sampling is completed, the measured color difference value and measured film thickness value of the color version object are obtained, and the residual between the measured value and the predicted value is calculated; when the absolute value of the residual is greater than the preset calibration threshold, the regression coefficient of the coating process mapping function is corrected with the measured value as feedback, so that the prediction error of subsequent rounds converges. Each color version object carries a prototype iteration round marker, and the coupling strength is updated incrementally after each recipe revision.

8. The method according to claim 1, characterized in that, It also includes a two-stage correlation analysis step between trial production and mass production: establishing a correlation mapping relationship between the impact of changes to each color version object in the trial production stage and the impact of changes to the same type of color version in the mass production stage; when the current evaluation is a change evaluation in the trial production stage, the potential impact prediction on the mass production stage is output synchronously based on the correlation mapping relationship.

9. The method according to claim 1, characterized in that, When the color difference exceeding the limit, film thickness exceeding the limit, and rework cost exceeding the limit are weighted and synthesized into a direct change impact in step S3, the weights of the three indicators are dynamically adjusted according to the evaluation scenario: in the rapid screening scenario, the weights of the color difference exceeding the limit and the rework cost exceeding the limit are higher than those of the film thickness exceeding the limit; in the process verification scenario, the weight of the film thickness exceeding the limit is higher than the other two; in the comprehensive evaluation scenario, the weights of the three indicators are equal.

10. The method according to claim 1, characterized in that, The risk levels mentioned in step S4 include low-risk, medium-risk, and high-risk zones: if the final impact of the change falls within the first score range, it is considered low-risk and the change is approved; if it falls within the second score range, it is considered medium-risk and additional verification spraying and color difference retesting are required, and the change is conditionally approved. When the score falls within the third range, it is considered high-risk, requiring a re-testing or formula adjustment, which triggers an engineering change request. The three score ranges increase sequentially and do not overlap.

Images