Answers are grounded in this paper and cite their sources.
For wide-threshold impact sensing, Sense-in-Energy (SiE) microdevices must simultaneously achieve reliable triggering, identifiable voltage output, and low energy loss under different loading conditions. However, a strong nonlinear relationship exists between cantilever-beam topology and the coupled electrochemical-fluid-solid response, making on-demand inverse generation difficult using conventional empirical design and parameter iteration. Here, we propose a physics-informed conditional variational autoencoder (PI-CVAE) framework for inverse cantilever-beam topology design. The cantilever topology is represented by an 11 x 18 binary matrix, and the loss function incorporates a BFS connectivity criterion, a span penalty, and a target-stiffness matching mechanism to reduce topology fracture, insufficient geometric span, and physical-property mismatch. In this study, the structural generation success rate is defined as the physically valid structural generation rate, namely the proportion of generated structures that simultaneously satisfy main-body connectivity, absence of isolated material cells, and the equivalent-stiffness interval required for the corresponding low-g, medium-g, or high-g scenario. In the low-g scenario, PI-CVAE improves this rate by approximately 18 percentage points compared with an ordinary CVAE without physics-informed components. After the equivalent-stiffness triggering intervals for low-g, medium-g, and high-g scenarios are established, the model can directly generate cantilever-beam topologies according to target threshold, voltage-amplitude, and energy-loss conditions. Simulation results show that the optimized structures increase voltage amplitude by up to 47% and reduce energy loss by up to 42%. Multi-g impact experiments show that the measured voltage waveforms of the optimized devices are consistent with simulation predictions in their main trends, with a maximum voltage-amplitude error of approximately 9%, and that ineffective energy consumption is reduced. This study provides an efficient intelligent design tool for the development of SiE devices and demonstrates the generalization capability and engineering potential of this intelligent design method under diverse operating conditions.
No takes yet. Share an insight, caveat, or question.
Liu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: