Search Results

Computer modeling and simulation have significantly facilitated the efficiency of product design and development in modern engineering, especially in the automotive industry. For the design and optimization of car models, optimization algorithms usually work better if the initial searching points are within or close to a feasible domain. Therefore, finding a feasible design domain in advance is beneficial. A data mining technique, Iterative Dichotomizer 3 (ID3), is exploited in this paper to identify sets of reduced feasible design domains from the original design space. Within the reduced feasible domains, optimal designs can be efficiently obtained while releasing computational burden in iterations. A mathematical example is used to illustrate the proposed method. Then an industrial application about automotive structural optimization is employed to demonstrate the proposed methodology. The results show the proposed method’s potential in practical engineering.

To reduce the computational time of the iterations in robust design, meta-models are frequently utilized to approximate time-consuming computer aided engineering models. However, the bias of meta-model uncertainty largely affects the robustness of the prediction results, this uncertainty need to be addressed before design optimization. In this paper, an efficient uncertainty quantification method considering both model and parameter uncertainties is proposed. Firstly, the uncertainty of parameters are characterized by statistical distributions. The Bayesian inference is then performed to improve the predictive capabilities of the surrogate models, meanwhile, the model uncertainty can also be quantified in the form of variance. Monte Carlo sampling is finally utilized to quantify the compound uncertainties of model and parameter. Furthermore, the proposed uncertainty quantification method is used for robust design.

Simulation models based design has become the common practice in automotive product development. Before applying these models in practice, model validation needs to be conducted to assess the validity of the models by comparing model predictions with experimental observations. In the validation process, it is vital to develop appropriate validation metrics for intended applications. When dealing with multivariate systems, comparisons between model predictions and test data with multiple responses would lead to conflicting decisions. To address this issue, this paper proposed a Bayesian classifier based validation method. With the consideration of both error rate and confidence in hypothesis testing, Bayesian classifier is developed for decision making. The process of validation is implemented on a real-world vehicle design case. The results show the proposed method’s potential in practical application.

In automobile industry, computational models built to predict the performances of the prototype vehicles are on the rise. To assess the validity or predictive capability of the model for its intended usage, validation activities are conducted to compare computational model outputs with test measurements. Validation becomes difficult when dealing with dynamic systems which often involve multiple functional responses, and the complex characteristics need to be appropriately considered. Many promising data analysis tools and metrics were previously developed to handle data correlation and evaluate the errors in magnitude, phase shift, and shape. However, these methods show their limitations when dealing with nonlinear multivariate dynamic systems. In this paper, kernel function based projection is employed to transform the nonlinear data into linear space, followed by the regular principal component analysis (PCA) based data processing.