Researchers at the University of California, Los Angeles (UCLA) have developed a groundbreaking framework for Structural Health Monitoring (SHM) that leverages artificial intelligence and diffractive optics to enable low-power, high-resolution vibration analysis. Led by Professor Aydogan Ozcan of the Department of Electrical and Computer Engineering, the team has introduced a method that co-optimizes a passive diffractive layer with a shallow neural network. This hybrid approach allows the system to encode time-varying mechanical vibrations into distinct spatiotemporal optical patterns, effectively bypassing the limitations of traditional sensor networks. Current SHM systems are essential for assessing the integrity of civil infrastructure, such as bridges and buildings, particularly after seismic events or natural hazards. However, these systems typically rely on dense arrays of accelerometers and strain gauges. Such conventional setups are not only expensive to install and maintain but also demand significant power and generate massive datasets requiring complex digital signal processing. Furthermore, achieving the high spatial resolution necessary for precise damage localization often necessitates an impractical number of sensors. The UCLA innovation addresses these challenges through physical-digital co-integration. Instead of digitizing raw physical signals at the source, the new system attaches a passive, AI-optimized diffractive layer directly to the structure. As the target oscillates, the diffractive surface moves, modulating an incoming light wave to physically encode structural displacements. This modulated optical signal is then captured by a minimal number of detectors and rapidly decoded by a low-power neural network. Ozcan notes that this paradigm represents a fundamental shift from standard digital sensing. By shifting a portion of the computational burden into the physical domain, the system intelligently pre-encodes complex, multidimensional oscillation information directly into the light. This eliminates the need for heavy on-site processing and reduces energy consumption, as the diffractive layer itself is entirely passive and consumes no power during operation. In collaboration with Professor Ertugrul Taciroglu's lab and the California Geological Survey, the team validated their platform using a laboratory-scale building model on a programmable shake table. Utilizing millimeter-wave illumination, they successfully extracted one-dimensional and two-dimensional vibration spectra under various dynamic excitations, including seismic waveforms from real earthquake datasets. The team also demonstrated the versatility of a wavelength-multiplexed diffractive system, which can simultaneously monitor vibrations at multiple points using light sources with distinct wavelengths. The technology offers significant potential for scalability. Because the diffractive surface operates as a passive encoder, its design can be adapted to different parts of the electromagnetic spectrum. By adjusting the dimensions of the diffractive features in proportion to the illumination wavelength, the same architectural principles can be applied to visible or infrared light, making the system adaptable to various environmental conditions and monitoring requirements. This advancement promises to revolutionize how critical infrastructure is monitored, offering a cost-effective, energy-efficient, and highly scalable solution for ensuring public safety.