An optimization approach for slotted Belleville springs in magnetorheological quasi-zero stiffness isolator

Abstract Magnetorheological quasi-zero stiffness isolators (MQZSI) comprise disc springs and magnetorheological dampers (MRD) as core components. The limited quasi-zero stiffness (QZS) region of disc springs adversely affects the controllable characteristics of magnetorheological dampers. To ensure optimal vibration isolation performance of MQZSI systems, an accurate and computationally efficient optimization methodology for disc springs is critically required. This study presents an optimization approach, integrating the finite element method (FEM) surrogate model based on sLSTM-CNN dual-stream architecture with the dung beetle optimizer (DBO) algorithm. Parametric modeling and systematic scanning techniques were employed to generate comprehensive physical field simulation data, which served as the training dataset for the neural network development. Subsequently, the sLSTM-CNN surrogate model predicts the mechanical properties of disc springs, achieving an optimal balance between computational efficiency and prediction accuracy for slotted disc spring mechanical performance estimation. The structural parameters of disc springs are optimized through integration of the sLSTM-CNN surrogate model with the DBO optimization algorithm. Experimental validation confirmed that disc springs optimized through the proposed methodology exhibited only 1.1% error between the working load f 0 and target design specifications. Comparative analysis reveals that the optimized disc spring achieves significantly improved mechanical properties over the control specimen. These results verify the effectiveness and reliability of the proposed optimization method for enhancing MQZSI performance in practical vibration isolation applications.