Motorcycle digital chassis optimization method based on computer-aided process design

Abstract

The invention relates to the technical field of computer-aided process design, and particularly discloses a motorcycle digital chassis optimization method based on computer-aided process design, comprising the following steps: S1, arranging a geometric measurement network formed by a plurality of measurement points on the surface of an original design model; s2, generating multiple groups of candidate manufacturing process parameters based on one group of initial manufacturing process parameters; s3, inputting a group of candidate manufacturing process parameters into the first agent model to obtain a predicted coordinate value set; s4, inputting the predicted coordinate value set into a second agent model to obtain a predicted value of the performance index; s5, constructing a joint response relationship for describing a mapping relationship among the manufacturing process parameters, the measurement point coordinate value set and the performance indexes; and S6, obtaining optimized manufacturing process parameters based on the joint response relationship, and outputting the optimized manufacturing process parameters and a corresponding predicted coordinate value set. The scientificity and stability of process design are improved.

Description

Technical Field

[0001] This invention relates to the field of computer-aided process design technology, and more specifically to a method for optimizing a motorcycle digital chassis based on computer-aided process design. Background Technology

[0002] In the design and manufacturing process of motorcycle chassis, the geometric accuracy and overall performance of the chassis structure have a direct impact on the handling stability, safety, and durability of the entire vehicle. As chassis structures become increasingly complex, the influence of manufacturing process parameters on chassis geometry and performance indicators exhibits obvious nonlinear and coupled characteristics. Traditional process design methods that rely on experience or single physical simulations are unable to accurately characterize the overall geometric deviations caused by changes in manufacturing process parameters and their comprehensive impact on performance.

[0003] Under current technological conditions, the manufacturing stage often only allows adjustments to local dimensions or single performance indicators, lacking a digital analysis method that can simultaneously correlate the overall geometric deformation distribution of the chassis with multiple performance indicators at the manufacturing process parameter level. This makes it difficult to achieve systematic optimization of chassis performance while meeting geometric tolerance constraints during process parameter optimization, easily leading to conflicts between performance improvement and geometric loss of control. Summary of the Invention

[0004] The purpose of this invention is to provide a method for optimizing the digital chassis of motorcycles based on computer-aided process design, thereby solving the above-mentioned technical problems.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] The method for optimizing a motorcycle's digital chassis based on computer-aided process design includes the following steps:

[0007] S1: Obtain the original design model of the motorcycle chassis and arrange a geometric measurement network consisting of multiple measurement points on the surface of the original design model;

[0008] S2: Within the manufacturing process parameter space, generate multiple sets of candidate manufacturing process parameters based on a set of initial manufacturing process parameters;

[0009] S3: Input a set of candidate manufacturing process parameters into the first surrogate model. The first surrogate model outputs the predicted coordinate values ​​of the measurement points in the geometric measurement network to form a set of predicted coordinate values.

[0010] S4: Input the set of predicted coordinate values ​​into the second proxy model, and the second proxy model outputs the predicted values ​​of one or more performance indicators of the motorcycle chassis.

[0011] S5: Based on multiple sets of candidate manufacturing process parameters, the set of predicted coordinate values ​​corresponding to each set of candidate manufacturing process parameters, and the predicted values ​​of performance indicators, construct a joint response relationship that describes the mapping relationship between manufacturing process parameters, the set of measurement point coordinate values, and performance indicators.

[0012] S6: Obtain and output optimized manufacturing process parameters based on joint response relationships.

[0013] Preferably, arranging the geometric measurement network includes:

[0014] Identify multiple key structural regions in the original design model. Key structural regions are areas where the geometry changes abruptly or where there are connection features.

[0015] In each critical structural region, measurement points are arranged according to a preset density rule, and all measurement points form a geometric measurement network.

[0016] Preferably, generating multiple sets of candidate manufacturing process parameters includes:

[0017] The manufacturing process parameter space is defined by the value range of multiple manufacturing process parameters;

[0018] Using the initial manufacturing process parameters as the center point, multiple sets of candidate manufacturing process parameters are generated within the manufacturing process parameter space using a preset sampling method.

[0019] Preferably, the construction and invocation of the first proxy model includes:

[0020] The first proxy model contains multiple independent prediction sub-models, each of which corresponds to a measurement point in the geometric measurement network.

[0021] When constructing the first proxy model, each prediction sub-model is trained using the historical manufacturing dataset. Each record in the historical manufacturing dataset contains a set of manufacturing process parameters and the actual measured coordinates of the motorcycle chassis manufactured under those parameters at the corresponding measurement points.

[0022] When the first proxy model is invoked, a set of candidate manufacturing process parameters are input. Each prediction sub-model outputs a predicted coordinate value of a corresponding measurement point. The outputs of all prediction sub-models constitute the set of predicted coordinate values ​​corresponding to the set of candidate manufacturing process parameters.

[0023] Preferably, the construction and invocation of the second proxy model includes:

[0024] The second agent model is a single machine learning model;

[0025] When constructing the second proxy model, the historical performance dataset is used for training. Each record in the historical performance dataset contains a set of coordinate values ​​of a motorcycle chassis sample at all measurement points in the geometric measurement network, as well as the values ​​of one or more performance indicators of the motorcycle chassis sample obtained through physical testing.

[0026] When the second proxy model is invoked, a set of predicted coordinate values ​​is input, and the second proxy model outputs the predicted values ​​of one or more performance indicators.

[0027] Preferably, constructing the joint response relationship includes:

[0028] The joint response relationship comprises multiple polynomial mathematical models;

[0029] The first type of polynomial mathematical model describes the mapping relationship between manufacturing process parameters and performance indicators, while the second type of polynomial mathematical model describes the mapping relationship between manufacturing process parameters and a coordinate component of a single measurement point in a geometric measurement network.

[0030] Based on multiple sets of candidate manufacturing process parameters, the set of predicted coordinate values ​​corresponding to each set of candidate manufacturing process parameters, and performance indicators, regression analysis is used to determine the coefficients of each term in the first type of polynomial mathematical model and the second type of polynomial mathematical model.

[0031] Preferably, obtaining optimized manufacturing process parameters includes:

[0032] Define the direction for performance indicator optimization;

[0033] A geometric tolerance condition is set, wherein the deviation between the predicted coordinate value of each measurement point in the geometric measurement network and its initial coordinate value in the original design model does not exceed a preset allowable range.

[0034] Within the manufacturing process parameter space, optimization algorithms are used to iteratively solve the joint response relationship;

[0035] In each iteration, based on the current combination of manufacturing process parameters, the corresponding performance index values ​​and the coordinate deviations of all measurement points are calculated through the joint response relationship.

[0036] When a set of manufacturing process parameters is obtained through iterative solution, and the corresponding performance index values ​​satisfy the optimization direction, and the coordinate deviation of all measurement points satisfies the geometric tolerance condition, the set of manufacturing process parameters is determined as the optimized manufacturing process parameters.

[0037] The beneficial effects of this invention compared to the prior art are as follows:

[0038] This invention constructs a digital optimization process that establishes a correlation between manufacturing process parameters, chassis geometry, and performance indicators. This enables the collaborative completion of motorcycle chassis process design and performance optimization within a unified computer-aided environment, effectively improving the predictability and controllability of manufacturing results and performance during the chassis design phase. This invention allows for early assessment of the combined impact of different combinations of manufacturing process parameters on chassis geometry and performance indicators during the design phase, avoiding unpredictable geometric deviations and performance fluctuations caused by process adjustments, thereby reducing trial production and rework. Simultaneously, this invention considers both chassis performance targets and geometric accuracy constraints during optimization, ensuring chassis geometric consistency and assembly reliability while meeting structural performance requirements, reducing quality risks caused by geometric deviations. This invention enables systematic optimization of manufacturing process parameters, allowing chassis design to move beyond reliance on single-experience judgments and instead make decisions based on digital analysis results, thus improving the scientific rigor and stability of process design. Furthermore, this invention helps shorten the chassis development cycle, improves the collaborative efficiency between design and manufacturing, and provides effective technical support for high-quality, digital, and refined design of motorcycle chassis. Attached Figure Description

[0039] The invention will now be further described with reference to the accompanying drawings.

[0040] Figure 1 This is a flowchart illustrating the motorcycle digital chassis optimization method based on computer-aided process design, as described in this invention. Detailed Implementation

[0041] 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, and 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.

[0042] Please see Figure 1 As shown, this invention is a method for optimizing a motorcycle's digital chassis based on computer-aided process design, comprising the following steps:

[0043] S1: Obtain the original design model of the motorcycle chassis and arrange a geometric measurement network consisting of multiple measurement points on the surface of the original design model;

[0044] In a preferred embodiment of the present invention, arranging a geometric measurement network includes:

[0045] After acquiring the original design model, the geometric information of the model surface is scanned and analyzed point by point. Based on the changes in the surface normals at various locations in the model and the geometric continuity between adjacent regions, the model surface is divided into zones. During this process, the trend of normal deflection between adjacent mesh units or surfaces on the model surface is judged. When there is a significant discontinuity in the normal change between adjacent regions or the change amplitude is significantly higher than the overall average, the corresponding location is marked as a candidate region where the geometric shape has abruptly changed. Simultaneously, combined with the topological connection information of each component in the original design model, welded parts, assembly joints, and component intersections in the model are identified. By reading the positional relationships of component boundaries, connecting lines, or connecting surfaces in the model, regions with direct connections between different components are determined, and these regions are identified as candidate regions with connection characteristics. Subsequently, the candidate regions obtained based on geometric continuity analysis are integrated with the candidate regions identified based on component connection relationships. Regions that overlap or are adjacent in spatial position are merged, thereby forming multiple key structural regions. These key structural regions are used to characterize locations in the original design model where geometric changes are concentrated or structural connections are complex, providing clear spatial boundaries for subsequent placement of measurement points within the corresponding areas.

[0046] After determining the key structural regions, measurement points are arranged within the surface area of ​​each key region according to a pre-defined density rule. This density rule constrains the distribution of measurement points within the region, ensuring that the measurement points cover the overall geometric features of the key structural region and reflect local morphological changes. The arrangement of measurement points follows the geometric boundaries of the region, distributing them uniformly or asymptotically within the region, maintaining a reasonable distance between adjacent measurement points, thereby guaranteeing a complete representation of the region's geometry.

[0047] S2: Within the manufacturing process parameter space, generate multiple sets of candidate manufacturing process parameters based on a set of initial manufacturing process parameters;

[0048] In a preferred embodiment of the present invention, generating multiple sets of candidate manufacturing process parameters includes:

[0049] Based on the determined process variables related to motorcycle chassis manufacturing, the acceptable ranges of each manufacturing process parameter are uniformly constrained, thus forming a manufacturing process parameter space. This space consists of multiple manufacturing process parameters involved in the chassis manufacturing process. Each parameter corresponds to an allowable value range, which is set based on existing process specifications, equipment capacity limitations, and the parameter ranges actually used in historical manufacturing records. By defining the upper and lower limits of each manufacturing process parameter, different combinations of process parameters are ensured to remain within the practically executable manufacturing conditions.

[0050] After defining the manufacturing process parameter space, a set of initial manufacturing process parameters characterizing the current or baseline manufacturing state are mapped into this space, serving as a central reference point for parameter combination generation. Subsequently, within this space, the parameter space is discretized according to a pre-defined sampling method, around the initial manufacturing process parameters. By selecting parameter combinations of different directions and amplitudes within the neighborhood of the initial manufacturing process parameters, multiple distinct candidate manufacturing process parameters are generated. During generation, each candidate manufacturing process parameter set maintains the value range constraints of its corresponding manufacturing process parameter, ensuring that the resulting candidate manufacturing process parameters cover parameter variations near the initial manufacturing process parameters. This provides a foundation for subsequent analysis of the impact of different manufacturing process parameter combinations.

[0051] S3: Input a set of candidate manufacturing process parameters into the first surrogate model. The first surrogate model outputs the predicted coordinate values ​​of the measurement points in the geometric measurement network to form a set of predicted coordinate values.

[0052] In another preferred embodiment of the present invention, the construction and invocation of the first proxy model includes:

[0053] The first surrogate model is used to establish the correspondence between manufacturing process parameters and the local geometric state of the chassis. Each measurement point in the geometric measurement network is treated as an independent geometric description object. Therefore, a prediction sub-model is constructed for each measurement point. The prediction sub-models are structurally independent but maintain a consistent input format.

[0054] Each predictive sub-model includes an input layer for receiving manufacturing process parameters, an intermediate computation layer for characterizing the nonlinear relationship between the manufacturing process parameters and the coordinates of the measurement point, and an output layer for outputting the coordinates of the measurement point. The intermediate computation layer achieves a step-by-step mapping of the manufacturing process parameter features through a multi-layer connection structure.

[0055] During the construction phase, for each measurement point in the geometric measurement network, a subset of data corresponding to that measurement point is extracted from the historical manufacturing dataset. Each record in the historical manufacturing dataset contains a set of actual manufacturing process parameters and the actual measured coordinates of the motorcycle chassis manufactured under those parameters at the corresponding measurement point. The manufacturing process parameters are used as input data to the prediction sub-model, and the actual measured coordinates of the corresponding measurement point are used as supervision signals to train the prediction sub-model. By iteratively updating the model's internal parameters, the model learns the mapping relationship between changes in manufacturing process parameters and changes in the geometric position of the measurement point.

[0056] After training all the prediction sub-models is completed, the multiple prediction sub-models together constitute the first surrogate model. During the invocation phase, a set of candidate manufacturing process parameters are simultaneously input into each prediction sub-model. Each prediction sub-model outputs the predicted coordinate values ​​of its corresponding measurement points according to the learned mapping relationship. The output results of all prediction sub-models are aggregated according to the correspondence of the measurement points in the geometric measurement network, thereby forming a complete set of predicted coordinate values ​​corresponding to the set of candidate manufacturing process parameters.

[0057] S4: Input the set of predicted coordinate values ​​into the second proxy model, and the second proxy model outputs the predicted values ​​of one or more performance indicators of the motorcycle chassis.

[0058] In another preferred embodiment of the present invention, the construction and invocation of the second proxy model includes:

[0059] The second proxy model is used to establish a mapping relationship between the chassis geometry represented by the geometric measurement network and the chassis performance indicators. It is a single machine learning model and uses the set of coordinate values ​​of all measurement points in the geometric measurement network as a unified input.

[0060] During the construction phase, training is conducted based on a historical performance dataset. Each record in the historical performance dataset corresponds to a motorcycle chassis sample that has been manufactured and tested. The record contains the set of coordinate values ​​of the sample at all measurement points in the geometric measurement network, as well as one or more performance index values ​​obtained through physical testing. To ensure the inherent consistency of the input data, the coordinate value sets of each sample are arranged in the order of the same measurement point number, and the coordinate components of each measurement point are sequentially concatenated to form a feature vector of fixed dimensions, so that the model input can fully express the geometric morphology information of the chassis in the global scope.

[0061] The second proxy model includes an input layer for receiving the feature vector, an intermediate hidden layer for learning multi-layer representations of global geometric features, and an output layer for outputting performance index prediction results. The output dimension of the output layer matches the number of performance indices to be predicted. During training, the feature vector formed by the set of coordinate values ​​is used as the model input, and the performance index values ​​of the corresponding samples are used as supervision signals. An iterative training method is used to update the model's internal parameters. In each iteration, the loss is calculated based on the error between the prediction result and the supervision signal, and backpropagation is used to adjust the connection weights of each layer, enabling the model to gradually learn the influence relationship between changes in chassis geometry and changes in performance indices.

[0062] During the invocation phase, a set of predicted coordinate values ​​obtained from the first proxy model is constructed into an input feature vector according to the measurement point order and coordinate organization method consistent with the training data. The second proxy model outputs the predicted values ​​of one or more corresponding performance indicators based on the learned mapping relationship, which are used to characterize the chassis performance under this set of geometric states.

[0063] S5: Based on multiple sets of candidate manufacturing process parameters, the set of predicted coordinate values ​​corresponding to each set of candidate manufacturing process parameters, and the predicted values ​​of performance indicators, construct a joint response relationship that describes the mapping relationship between manufacturing process parameters, the set of measurement point coordinate values, and performance indicators.

[0064] In a preferred embodiment of the present invention, constructing the joint response relationship includes:

[0065] After obtaining multiple sets of candidate manufacturing process parameters and obtaining the corresponding sets of predicted coordinate values ​​and predicted values ​​of performance indicators through the preceding steps, a joint response relationship is constructed based on these data to describe the relationship between the manufacturing process parameters and the output results. The joint response relationship is composed of a first-type polynomial mathematical model and a second-type polynomial mathematical model.

[0066] The first type of polynomial mathematical model is used to characterize the overall impact of changes in manufacturing process parameters on changes in performance indicators. Since different performance indicators have different response characteristics to manufacturing process parameters, a corresponding polynomial model is established for each performance indicator. Each model takes the manufacturing process parameters as input and a single performance indicator as output, thus forming a clear mapping from manufacturing process parameters to performance indicators.

[0067] The second type of polynomial mathematical model is used to characterize the influence of changes in manufacturing process parameters on the coordinate changes of measurement points in a geometric measurement network. Considering that the geometric measurement network contains multiple measurement points and that the change patterns of each measurement point are not the same in different coordinate directions, a corresponding polynomial model is established for each coordinate component of each measurement point in the geometric measurement network, so that the mapping relationship between manufacturing process parameters and single coordinate components can be described separately.

[0068] Each polynomial model structurally includes a constant term characterizing the baseline state, a first-order term describing the independent effect of a single manufacturing process parameter, and an interaction term reflecting the coupling effect between different manufacturing process parameters. This allows for continuous approximation of the response behavior within the range of variations in manufacturing process parameters. When determining the specific coefficients of each polynomial model, multiple sets of candidate manufacturing process parameters are used as independent variable samples. The predicted performance index values ​​corresponding to each set of candidate manufacturing process parameters are used as response samples for the first type of polynomial model. Simultaneously, the coordinate components of each measurement point in the corresponding set of predicted coordinate values ​​are used as response samples for the second type of polynomial model. Regression analysis is used to solve for the model parameters, ensuring that the polynomial model as closely as possible to the changing trends reflected by the sample data.

[0069] The first type of polynomial mathematical model is as follows:

[0070] ;

[0071] P represents the predicted value of the performance indicator, X i X represents the value of the i-th manufacturing process parameter. j This represents the value of the j-th manufacturing process parameter, n represents the total number of manufacturing process parameters, a0 is the coefficient of the constant term, and a i Let b be the coefficient of the first-order term of the i-th manufacturing process parameter. ij Let be the coefficient of the quadratic interaction term between the i-th and j-th manufacturing process parameters, and satisfy b ij =b ji And i and j are the serial numbers of the manufacturing process parameters;

[0072] The mathematical model for a polynomial of the second kind is:

[0073] ;

[0074] C represents the predicted coordinates of a measurement point in the X, Y, or Z direction, where c0 is the coefficient of the constant term, and c i Let d be the coefficient of the first-order term of the i-th manufacturing process parameter. ij Let be the coefficient of the quadratic interaction term between the i-th and j-th manufacturing process parameters, and satisfy d ij =d ji ;

[0075] This invention employs a quadratic polynomial form to construct the first and second types of mathematical models. Its core principle lies in using an analytically computable mathematical structure to efficiently approximate the complex nonlinear physical relationships between manufacturing process parameters, geometric deformation, and final performance. In motorcycle chassis manufacturing, the influence of process parameters (such as welding parameters and assembly sequence) on the geometric coordinates of components, and the influence of geometric coordinates on overall performance (such as stiffness and strength), is usually not a simple linear superposition, but rather involves interactive and saturation effects between parameters. The quadratic polynomial form (containing linear terms, quadratic square terms, and cross-product terms) mathematically provides a unified framework to simultaneously describe the independent influence of individual parameters, the nonlinear influence of the parameters themselves, and the coupled interactive influence between any two parameters—something linear models cannot achieve. By using historical manufacturing data and performance test data, the least squares regression method is employed to determine the coefficients of the polynomial. Essentially, this is a data-driven fitting process aimed at finding a smooth function that statistically best reflects the patterns in the existing data. Once determined, this function constitutes a "surrogate model" that can be computed instantly, replacing time-consuming and expensive physical simulations or pilot experiments. It directly predicts the changes in measurement point coordinates and final performance indicators caused by any given set of process parameters. Therefore, the principle behind this mathematical model is to use parameterized mathematical surfaces to fit and encapsulate complex physical causal relationships. Its form achieves a balance between expressive power and computational complexity in engineering approximations, enabling subsequent automated optimization searches to be performed on explicit and differentiable mathematical relationships. This allows for the systematic search for the optimal combination of process parameters that satisfies both geometric tolerance and performance objectives.

[0076] S6: Obtain and output optimized manufacturing process parameters based on joint response relationships.

[0077] In another preferred embodiment of the present invention, obtaining optimized manufacturing process parameters includes:

[0078] The joint response relationship is composed of two types of polynomial mathematical models, both of which use manufacturing process parameters as independent variables. The first type of polynomial mathematical model is used to directly calculate the predicted values ​​of performance indicators from the manufacturing process parameters, while the second type of polynomial mathematical model is used to calculate the predicted coordinate values ​​of each measurement point in the geometric measurement network in different coordinate directions from the manufacturing process parameters. Therefore, in actual construction, the first type of model is not a single formula but a polynomial model is built for each performance indicator, and the second type of model is also not a single formula but a polynomial model is built for each coordinate component of each measurement point. This allows us to obtain a set of predicted performance indicators and a set of predicted coordinate values ​​covering all measurement points when given a set of manufacturing process parameters.

[0079] Upon entering the optimization phase, the optimization direction is first set for the performance indicators based on the chassis design and usage requirements. This optimization direction characterizes the expected trend of the performance indicators in terms of numerical changes, thus providing a basis for subsequent judgment on whether the combination of manufacturing process parameters meets the performance requirements. Simultaneously, based on the initial coordinate values ​​of each measurement point in the original design model, geometric tolerance conditions are set. By limiting the allowable offset range of each measurement point in the geometric measurement network, the acceptable boundaries of chassis geometry changes during manufacturing are clarified, ensuring that the optimization process is carried out under the premise of controllable geometric accuracy.

[0080] Subsequently, a set of initial manufacturing process parameters is selected in the manufacturing process parameter space as the starting point for iteration, and the optimization algorithm is called to repeatedly solve the joint response relationship. In each iteration, the current combination of manufacturing process parameters is substituted into the first type of polynomial model set to obtain the predicted value of the current performance index. At the same time, the current combination of manufacturing process parameters is substituted into the second type of polynomial model set to obtain the predicted values ​​of each coordinate component of all measurement points in the geometric measurement network, and a set of predicted coordinate values ​​is formed accordingly. Then, based on the set of predicted coordinate values, the coordinate deviation of each measurement point relative to the initial coordinate value is calculated to verify the geometric tolerance condition.

[0081] During the iterative update process, the combination of manufacturing process parameters is adjusted through optimization algorithms to make the predicted values ​​of performance indicators change toward the set optimization direction. Geometric tolerance conditions are always used as feasibility constraints for screening. When the predicted values ​​of performance indicators corresponding to the combination of manufacturing process parameters obtained in a certain round of iteration meet the optimization direction and the coordinate deviations of all measurement points calculated by the second type of polynomial model fall within the preset allowable range, the combination of manufacturing process parameters is determined as the optimized manufacturing process parameters, and the optimized manufacturing process parameters and the corresponding set of predicted coordinate values ​​are output.

[0082] The overall concept of this invention is based on the fundamental understanding that the geometric state and performance of a motorcycle chassis during actual manufacturing are essentially the result of the combined effects of manufacturing process parameters. While this relationship is complex, it possesses continuity and learnability within a controlled range of process parameters. Therefore, this invention does not attempt to repeatedly test process parameters directly at the actual manufacturing or physical testing level. Instead, it explicitly and model the intrinsic connections between manufacturing process parameters, geometric changes, and performance results through digital means. Its core principle lies in treating manufacturing process parameters as adjustable independent variables, and the coordinate changes of key geometric positions of the chassis and performance indicators as response results driven by these independent variables. By establishing mapping relationships layer by layer, the causal relationships originally implicit in the manufacturing process are transformed into calculable functional relationships. First, at the geometric level, the complex chassis structure is compressed into a geometric description composed of key measurement points, allowing manufacturing-induced deformations to be expressed by data of limited dimensions. Subsequently, proxy relationships are established between process parameters and geometry, and between geometry and performance, allowing the impact of manufacturing process parameters on performance to be transmitted and characterized through the intermediate state of geometric changes. Building upon this foundation, these relationships are then integrated into an analytical response model, allowing performance objectives and geometric constraints to be simultaneously expressed as functional conditions of manufacturing process parameters. This transforms the process adjustment problem, which originally relied on experience and trial production, into a constrained solution problem within the parameter space. It is precisely this holistic modeling approach—starting from process parameters, transmitting through geometric states, and ultimately reflecting in performance results—that enables systematic optimization of manufacturing process parameters in a digital environment without departing from actual manufacturing logic. This principle underpins the inherent logical connections between each step.

[0083] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A method for optimizing a motorcycle's digital chassis based on computer-aided process design, characterized in that: Includes the following steps: S1: Obtain the original design model of the motorcycle chassis and arrange a geometric measurement network consisting of multiple measurement points on the surface of the original design model; S2: Within the manufacturing process parameter space, generate multiple sets of candidate manufacturing process parameters based on a set of initial manufacturing process parameters; S3: Input a set of candidate manufacturing process parameters into the first surrogate model. The first surrogate model outputs the predicted coordinate values ​​of the measurement points in the geometric measurement network to form a set of predicted coordinate values. S4: Input the set of predicted coordinate values ​​into the second proxy model, and the second proxy model outputs the predicted values ​​of one or more performance indicators of the motorcycle chassis. S5: Based on multiple sets of candidate manufacturing process parameters, the set of predicted coordinate values ​​corresponding to each set of candidate manufacturing process parameters, and the predicted values ​​of performance indicators, construct a joint response relationship that describes the mapping relationship between manufacturing process parameters, the set of measurement point coordinate values, and performance indicators. S6: Obtain and output optimized manufacturing process parameters based on joint response relationships.

2. The motorcycle digital chassis optimization method based on computer-aided process design according to claim 1, characterized in that, Setting up a geometric measurement network includes: Identify multiple key structural regions in the original design model. Key structural regions are areas where the geometry changes abruptly or where there are connection features. In each critical structural region, measurement points are arranged according to a preset density rule, and all measurement points form a geometric measurement network.

3. The motorcycle digital chassis optimization method based on computer-aided process design according to claim 1, characterized in that, The generation of multiple sets of candidate manufacturing process parameters includes: The manufacturing process parameter space is defined by the value range of multiple manufacturing process parameters; Using the initial manufacturing process parameters as the center point, multiple sets of candidate manufacturing process parameters are generated within the manufacturing process parameter space using a preset sampling method.

4. The motorcycle digital chassis optimization method based on computer-aided process design according to claim 1, characterized in that, The construction and invocation of the first proxy model includes: The first proxy model contains multiple independent prediction sub-models, each of which corresponds to a measurement point in the geometric measurement network. When constructing the first proxy model, each prediction sub-model is trained using the historical manufacturing dataset. Each record in the historical manufacturing dataset contains a set of manufacturing process parameters and the actual measured coordinates of the motorcycle chassis manufactured under those parameters at the corresponding measurement points. When the first proxy model is invoked, a set of candidate manufacturing process parameters are input. Each prediction sub-model outputs a predicted coordinate value of a corresponding measurement point. The outputs of all prediction sub-models constitute the set of predicted coordinate values ​​corresponding to the set of candidate manufacturing process parameters.

5. The motorcycle digital chassis optimization method based on computer-aided process design according to claim 1, characterized in that, The construction and invocation of the second proxy model includes: The second agent model is a single machine learning model; When constructing the second proxy model, the historical performance dataset is used for training. Each record in the historical performance dataset contains a set of coordinate values ​​of a motorcycle chassis sample at all measurement points in the geometric measurement network, as well as the values ​​of one or more performance indicators of the motorcycle chassis sample obtained through physical testing. When the second proxy model is invoked, a set of predicted coordinate values ​​is input, and the second proxy model outputs the predicted values ​​of one or more performance indicators.

6. The motorcycle digital chassis optimization method based on computer-aided process design according to claim 1, characterized in that, Building a joint response relationship includes: The joint response relationship comprises multiple polynomial mathematical models; The first type of polynomial mathematical model describes the mapping relationship between manufacturing process parameters and performance indicators, while the second type of polynomial mathematical model describes the mapping relationship between manufacturing process parameters and a coordinate component of a single measurement point in a geometric measurement network. Based on multiple sets of candidate manufacturing process parameters, the set of predicted coordinate values ​​corresponding to each set of candidate manufacturing process parameters, and performance indicators, regression analysis is used to determine the coefficients of each term in the first type of polynomial mathematical model and the second type of polynomial mathematical model.

7. The motorcycle digital chassis optimization method based on computer-aided process design according to claim 1, characterized in that, Obtaining optimized manufacturing process parameters includes: Define the direction for performance indicator optimization; A geometric tolerance condition is set, wherein the deviation between the predicted coordinate value of each measurement point in the geometric measurement network and its initial coordinate value in the original design model does not exceed a preset allowable range. Within the manufacturing process parameter space, optimization algorithms are used to iteratively solve the joint response relationship; In each iteration, based on the current combination of manufacturing process parameters, the corresponding performance index values ​​and the coordinate deviations of all measurement points are calculated through the joint response relationship. When a set of manufacturing process parameters is obtained through iterative solution, and the corresponding performance index values ​​satisfy the optimization direction, and the coordinate deviation of all measurement points satisfies the geometric tolerance condition, the set of manufacturing process parameters is determined as the optimized manufacturing process parameters.

Images