SBIR Phase I: Generative Physics-Informed AI for Computational Physics and Model-Based Engineering Development

The broader impact of this Small Business Innovation Research (SBIR) Phase I project will be the democratization and enhancement of physics-based simulation models in product engineering. By developing a Generative Bayesian Physics-Informed Classifier (B-PIC) network, this project aims to make advanced simulation tools more accessible, reducing the need for specialized analysts. This innovation has the potential to significantly lower development costs and time, enabling earlier and more frequent simulations in the product design process. The resulting sustainable engineering practices will lead to longer-lasting, higher-performing products, benefiting various industries and contributing to economic growth. Additionally, this technology will foster broader scientific and technological understanding by integrating recent advances in generative artificial intelligence into physical sciences, paralleling the impact seen in computer vision and natural language processing. This Small Business Innovation Research (SBIR) Phase I project proposes to address the challenges of mastering and setting up analyst-caliber physics simulations. The current process is complex, time-consuming, and requires extensive training. By incorporating strategies from Physics-Informed Gaussian Process (PIGP) and Bayesian Physics-Informed Neural Network (BPINN) architectures, the B-PIC network will integrate physics into its architecture, loss, and error functions. This approach aims to minimize the need for package-specific expertise and promote efficient, accurate simulations. The research objectives include developing the B-PIC network, optimizing the setup process for partial differential equations (PDEs), and demonstrating the system's effectiveness in reducing simulation time and cost. The anticipated technical results will showcase the network's ability to transform physics simulation from a validation tool to a crucial development driver in product engineering. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.