Reliability-based robust optimization design of vehicle braking systems under multiple failure modes based on high-precision surrogate models

In the process of enhancing the braking safety and performance stability of the automotive braking system, comprehensively considering the failure modes of the braking system and obtaining an accurate reliability function model are crucial for completing the reliability-based robust optimization design (RBROD) of the braking system. In engineering practice, the modal characteristics and thermal-mechanical coupling reliability robust design method under multiple failure modes encounter problems such as low computational efficiency and difficult problem-solving. To tackle these issues, this paper proposes a new method called Reliability-Based Robust Optimization Design under Multi-Failure Mode (MF-RBROD). In the proposed optimization method, the deep learning surrogate model replaces the experiments and finite element analysis (FEA) in the original adaptive process, greatly saving time and cost. The accuracy of the finite element model is determined by comparing the modal experimental results with the finite element analysis results of the disc brake. A total of 300 training samples and 50 testing samples were obtained. After establishing the reliability performance function through samples, among the four surrogate model methods, the DL-AK method with the smallest average relative error was selected to conduct a reliability sensitivity analysis of the braking system. The error of this method is only 0.01911%, which is significantly better than the other three surrogate models. The MF-RBROD model was formed by weighted transfer method. The MF-RBROD method has effectively improved the reliability of the disc brake. Specifically, the frequency reliability has been significantly enhanced from 0.9414 to 0.9971, while the thermo-mechanical coupling reliability has been raised from 0.9483 to 0.9951. At the same time, the sensitivities of various design parameters affecting the braking system reliability and the total mass of the disc brake were significantly reduced. The optimization results of the disc brake have demonstrated the effectiveness of the MF-RBROD method.