In a new study, researchers describe their achievement in designing PMUTs with all the desired properties. They did it all with the ground-breaking creative power of MultiphysicsAI. This cutting-edge solution combines the power of cloud-based finite element method (FEM) simulations with neural surrogates. Thus, it heralds the dawn of biomed ultrasonic applications optimization. With the help of MultiphysicsAI, engineers can traverse the complex trade-offs between sensitivity and bandwidth orders of magnitude more efficiently than they could using conventional approaches.
With MultiphysicsAI, PMUT designs can be quickly explored and optimized, resulting in validated performance improvements in as little as minutes. This stands in sharp contrast to the often long days required for traditional design iterations. Utilizing standard cloud infrastructure, MultiphysicsAI efficiently processes data and simulations, making it accessible to a broader range of researchers and developers.
Advanced Optimization Techniques
The key power of MultiphysicsAI comes from being trained on more than 10,000 randomized geometries. This deep training allows the platform to create AI surrogates with a mean error of just 1%. The sub-millisecond inference time enables unprecedented increases in key performance indicators such as transmit sensitivity, center frequency, and fractional bandwidth, as well as electrical impedance. This improvement speeds up the design process and increases precision.
MultiphysicsAI rigorously optimizes four geometric parameters that govern PMUT performance. Using its optimized algorithms, it produces Pareto front optimization which significantly boosts fractional bandwidth from 65% to 100%. This enhancement holds great importance for applications that demand high precision and resolution such as ultrasonic imaging and therapy.
Performance Enhancements
Beyond bandwidth enhancements, MultiphysicsAI’s optimization techniques result in significant sensitivity improvements. Researchers have seen remarkable increases in sensitivity up to 2-3 dB. Further, they did this while maintaining a stable center frequency of 12 MHz to within ±0.2%. These innovations are crucial in addressing the serious requirements of biomedical applications like tissue engineering and drug delivery where accuracy and dependability are critical.
Its unique advantage of simultaneous balancing of many performance metrics at the same time makes MultiphysicsAI a major disruptor in the PMUT design field. The new platform provides ODOT’s engineers with the tools they need to understand impacts and make better decisions on design trade-offs. This empowerment promotes creativity and accelerates the evolution of next-generation ultrasonic devices.