AI-driven Inverse Design of Band-Tunable Mechanical Metastructures for Tailored Vibration Mitigation

TL;DR

The study addresses the challenge of designing mechanical metastructures with tunable bandgaps for vibration mitigation by introducing an AI-driven inverse design framework. It combines a FEM-inspired forward model with multi-head spatial attention and a multiscale Gaussian self-attention inverse model to map between metastructure geometry and transmissibility spectra, demonstrated on 720 FE-generated samples and validated experimentally with 3D-printed PLA lattices and mass inserts. Key contributions include the forward model architecture (12 GFE blocks, 8 FSA heads), the MGSA-based inverse design with Gaussian spectrum encoding, and successful design of interlaced honeycomb metastructures with tailored bandgaps, achieving similarity scores near 0.991 between predicted and original geometries. The work enables efficient, data-driven design of complex multiscale metastructures for practical vibration control and points toward extensions to 3D interlaced layouts and 4D, smart-material-enabled metamaterials.

Abstract

On-demand vibration mitigation in a mechanical system needs the suitable design of multiscale metastructures, involving complex unit cells. In this study, immersing in the world of patterns and examining the structural details of some interesting motifs are extracted from the mechanical metastructure perspective. Nine interlaced metastructures are fabricated using additive manufacturing, and corresponding vibration characteristics are studied experimentally and numerically. Further, the band-gap modulation with metallic inserts in the honeycomb interlaced metastructures is also studied. AI-driven inverse design of such complex metastructures with a desired vibration mitigation profile can pave the way for addressing engineering challenges in high-precision manufacturing. The current inverse design methodologies are limited to designing simple periodic structures based on limited variants of unit cells. Therefore, a novel forward analysis model with multi-head FEM-inspired spatial attention (FSA) is proposed to learn the complex geometry of the metastructures and predict corresponding transmissibility. Subsequently, a multiscale Gaussian self-attention (MGSA) based inverse design model with Gaussian function for 1D spectrum position encoding is developed to produce a suitable metastructure for the desired vibration transmittance. The proposed AI framework demonstrated outstanding performance corresponding to the expected locally resonant bandgaps in a targeted frequency range.

Figures (14)

• Figure 1: Different types of metastructure unit cells can be categorized based on their lattice structures. (a) Five popular unit cells are used as primary lattices. (b) A complex array of primary lattices can be used to create different possible interesting metastructure unit cells. (c) Multi-unit patterns represent the arrangement of unit cells to form complete 2D and 3D metastructures.
• Figure 2: (a)The Tomb of Itmad-ud-Daula is a mausoleum in the city of Agra in the Indian state of Uttar Pradesh (27.1929 $^{\circ}$ N, 78.0310 $^{\circ}$ E). The architecture of this monument is unique in terms of crafting which uses a combination of pietra-dura and calligraphy-based decorative art techniques. (b to d). Consequently, the latticework is depicted through the carved panels of calligraphic design. The walls, floors, and niches are decorated with beautiful geometrical patterns. (e and f). Here, we have specifically focused on the Jali screens at the centre of the niches all around the tomb. They were used extensively for elegance and their ability to facilitate the entry of natural light and air inflow. Often it is also used as part of a passive cooling system. (g) Shows the corresponding metastructure models (LS1-LS9) inspired from these architectural elements.
• Figure 3: Schematic diagram illustrates latticed mechanical metastructures featuring different interlaced lattice structures (LS1-LS9). Dynamic analysis of the metastructures is performed and mode shapes of lattices at a frequency corresponding to minimum force transmissibility for in-plane excitation are shown. In mode shapes, red indicates maximum deformation and blue indicates minimum deformation. The attenuation corresponding to each geometry is obtained by using the transmissibility plot for each structure. We have highlighted distinct attenuation zones within the transmissibility response of metastructures through green-shaded patches.
• Figure 4: Conducting a comparative analysis, this study investigates the impact of scaling (x, 1.5x, 2x) of the unit cell on the bandgap, with the resulting BG ratio illustrated in a bar diagram. The basic building blocks of unit and sub-unit cells are represented in the middle column of the figure. Subunit cells are used as an interlaced component for the primary structure from which each configuration is constructed. The patterns of latticed mechanical metastructures are 3D Printed using PLA having different primary lattice structures namely (a) Honeycomb-honeycomb interlaced (LS1) (b) Honeycomb interlaced with a honeycomb pattern (LS2) (c) Honeycomb interlaced with a star (LS3) (d) Chiral interlaced structure (LS4) (e) Ring interlaced structure (LS5) (f) Fishtail structure (LS6) (g) Star-honeycomb structure (LS7) (h) Honeycomb-auxetic structure (LS8) (i) Honeycomb-auxetic interlaced structure (LS9).
• Figure 5: (a) A comparison of bandgap at different frequency levels of different structures interlaced in the honeycomb structure. (b), (d) The comparison of static performance is obtained through analytical computations of different structures. The displacement control mechanism is set as 5 mm compression of the initial height of samples (c) Comparison of bandgap in different order (n=0, 1, 2) of the fractal structure.