Recently, the research team led by Professor He Wenbin from the College of Mechanical and Electrical Engineering at Zhengzhou University of Light Industry (ZZULI) published a high-quality academic paper titled "Reliability-based Design Optimization of Hinge Sleeve using Adaptive E-SVM" in Reliability Engineering & System Safety (CAS Q1, TOP journal, IF=11), an international prestigious journal in the field of reliability. Associate Professor Li Xiaoke from the College of Mechanical and Electrical Engineering serves as the first author.

In the lightweight design of complex engineering structures, a highly non-linear implicit relationship exists between design parameter uncertainties and performance responses. Therefore, how to efficiently and accurately conduct structural reliability analysis and optimization design with small samples remains a significant challenge in reliability engineering. To address the aforementioned issue, this study proposes an ensemble SVM (E-SVM) reliability-based design optimization (RBDO) method using Bayesian optimization. By adaptively combining polynomial with Gaussian kernel functions via Bayesian optimization, and through the adoption of K-fold cross validation for hyperparameter optimization, this method establishes a high-precision and strongly generalizable E-SVM model. On this basis, an adaptive sampling strategy and a boundary failure probability quantification mechanism are introduced to progressively refine the model and achieve efficient convergence in reliability analysis. Experimental results reveal that the proposed method effectively improves the precision of surrogate models, exhibiting strong effectiveness and robustness.

This research has been supported by programs such as the National Natural Science Foundation of China and the Henan Province Science and Technology Program for Tackling Key Problems.

Journal article link: https://www.sciencedirect.com/science/article/pii/S0951832025006350?getft_integrator=clarivate&pes=vor&utm_source=clarivate