How AI Accelerates PMUT Design for Biomedical Ultrasonic Applications

The white paper explores how AI accelerates the design of piezoelectric micromachined ultrasonic transducers (PMUTs) for biomedical applications. It introduces a MultiphysicsAI workflow that combines cloud-based FEM simulation with neural surrogates to optimize PMUT design efficiently. By training on 10,000 randomized geometries, AI surrogates achieve 1% mean error and sub-millisecond inference for key performance indicators. The approach enables Pareto front optimization, increasing fractional bandwidth and sensitivity while maintaining a 12 MHz center frequency. This innovative workflow streamlines the design process, offering rapid performance improvements and transparent, data-driven exploration.