Keren Dai

Multifrequency and complex mechanical vibrations are widespread in numerous modern engineering fields. Existing elastic metamaterials are limited in their ability to achieve wide-band multifrequency vibration reduction in the low-frequency range, making them ineffective at mitigating complex multifrequency vibrations. For structures composed of a single material, the bandgap characteristics are primarily determined by the designed structural configuration and geometric parameters. To address this issue, this paper innovatively proposes a cosine-ligament local resonance three-dimensional elastic metamaterial capable of achieving multifrequency, wide-band vibration attenuation in the low-frequency range. By investigating the impact of geometric parameters on the bandgap characteristics, the optimal parameter combination for this model was determined, resulting in model optimization. The finite element method was employed to study the bandgap and transmission properties, and the mechanism of bandgap formation was analyzed in conjunction with vibration modes at the bandgap boundaries. The results demonstrate that the cosine-ligament configuration metamaterial exhibits excellent vibration reduction performance in four frequency bands: 151–159 Hz, 429–531 Hz, 568–932 Hz, and 1044–1060 Hz. Vibration table testing and model validation experiments further confirmed the vibration attenuation performance of the 2×3×3 finite array unit of this cosine-ligament configuration metamaterial in practical applications. The results indicate that it can reduce vibration amplitudes by approximately 99% within the frequency range of 414–995 Hz.

Driven by the “More than Moore” law, the miniaturization and multi‐functional integration of micro‐energy units and micro‐sensors are crucial for next‐generation compact microsystems. To address severe signal oscillation and unstable energy supply under continuous extreme overload (>10 000 g ), this study proposes a Sensing‐in‐Energy (SiE) microdevice featuring a MEMS movable inertial structure built into a supercapacitor electrolyte cavity. This architecture leverages high‐ g shock‐driven transient contact between embedded metal microspheres and the electrode to modulate the soft short‐circuit sensing effect. By utilizing electrolyte damping for signal self‐filtering, the device achieves a high‐range (30 000 g ) and large‐amplitude (450 mV) output with weak oscillation (signal adhesion coefficient reduced by 90.84%). To ensure precise development of the SiE microdevice, a multi‐physics‐driven design method coupling transient fluid‐structure interaction (FSI) and micro‐nano scale rough electrical contact theory was established. This reveals the dynamic mapping laws from mechanical excitation to fluid‐structure coupling interface response and electrochemical output, reducing simulation‐experiment error to within 8%. Furthermore, a high‐precision microsphere embedding process was developed to minimize manufacturing randomness, maintaining signal repeatability error below 10%. This work offers a new paradigm for designing SiE microdevices for extreme environments and lays a technical foundation for the development of future high‐performance heterogeneous microsystems.

Anti-collision detection technologies primarily rely on optical, radar, or laser sensors; however, their performance often deteriorates severely under adverse weather conditions (e.g., rain, snow, fog) or in scenarios involving visual occlusion. By contrast, magnetic anomaly detection leverages perturbations in the geomagnetic field induced by target objects (e.g., vehicles, metallic obstacles), exhibiting intrinsic all-weather operability and strong anti-interference capability. Nevertheless, conventional magnetic anomaly detection methods suffer from the limited applicability of the magnetic dipole model, which only affords coarse positioning accuracy and is predominantly suited for long-range targets. To address this limitation, this paper proposes an Extended N-th-Pole Magnetic Dipole (E-NMD) model that improves accuracy by analyzing the Lagrangian cosine term and rigorously constraining truncation errors under specific operational conditions. Experimental results demonstrate that, for steel with a relative permeability of 200, the model achieves a fitting variance of 99.87%. Furthermore, to overcome the inversion difficulties arising when the strength of short-range magnetic anomalies is comparable to sensor noise, an Adaptive Iterative Extended Kalman Filter (AI-EKF) is developed to enable robust noise suppression and precise distance estimation. Results indicate that E-NMD outperforms the traditional N-th-Pole Magnetic Dipole (NMD) model in proximal state estimation, achieving a 39.62% reduction in Root Mean Square Error (RMSE). Finally, in light of parameter uncertainty in magnetic anomaly targets under real-world conditions, a Dual-Mode Pairwise Iterative Extended Kalman Filter (DI-EKF) is introduced to jointly estimate parameters and system states, yielding an 89% reduction in RMSE compared to AI-EKF.

Wearable protective metamaterials are expected to simultaneously provide compliant low-load buffering, stable high-load resistance, curve-conformal deployability, and integrated sensing, which remains challenging for many existing designs. Here, we propose a dual-stage gradient compression–torsion metamaterial (DSGCTM) together with a data-driven framework for multi-objective optimization and inverse design. By combining geometric gradients with compression–torsion coupling, the DSGCTM achieves a controlled soft-to-stiff response, while its architecture enables interlocking arrays that conform to curved surfaces. In drop-ball tests, the proposed structure reduces the transmitted force from a 1190 N impact to 563 N. For engineering-metric–driven applications, a surrogate-assisted optimization strategy is developed to improve the balance between energy absorption and plateau stability, showing over 30% improvement in specific energy absorption and more than 10% enhancement in crushing force efficiency. For response-driven customization, a Best-of-N forward-consistency inverse design network is developed to alleviate the non-uniqueness issue in mapping target mechanical responses to structural geometries, with the inverse-designed samples reproducing key mechanical features within 7% error. In addition, an integrated triboelectric sensing layer enables self-powered recognition of five impact types, with all classification accuracies exceeding 90% on the independent test set. These results establish DSGCTM as a staged protective architecture with conformal deployability and demonstrate a feasible route toward data-driven customization, while further showing the feasibility of integrated triboelectric self-sensing.