SBIR Phase I: No-code electric grid analytics platform for predictive maintenance planning and emergency response

Sync, Inc., Birmingham AL

Investigators

Abstract

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase I project is in helping in the design of a cost-effective resiliency strategy for defense against the impacts of climate change. Today, the U.S. experiences significantly more weather events imposing substantially more financial burden compared to 40 years ago. As climate change accelerates, communities will be forced to endure increased financial burdens protecting against and mitigating its impacts. Research carried out in this SBIR project is intended to help reduce these expenses. Utilities, cities, insurance companies, and municipalities are stakeholders in these resiliency efforts and will be looking for new tools to help mitigate the effects and reduce the costs of climate change. The research in this project will enable utilities to reduce resilience-related costs and reduce impact on businesses and local economies due to power outages. Such actions may benefit all communities, particularly poorer and marginal communities that often endure the worst impacts of climate change. This Small Business Innovation Research Phase I project aims to develop predictive analytics for tropical storms and wildfires and to integrate this functionality into a power grid analytics software platform. Three artificial intelligence/machine learning (AI/ML) tools will be implemented, qualitatively expanding on early prototypes: (1) in the satellite imagery (SI) domain, an optimized combination of deep-learning neural network (DLNN) techniques will be trained on large-scale satellite images, resulting in the world's first tree growth tracking and species identification tool; (2) a "Virtual Wind Tunnel" (VWT) will be augmented with computational fluid dynamics (CFD) and empirical physics modeling to estimate the probability of trees damaging power transmission assets during weather events forecast; and (3) towards a no-code user interface, existing natural language processing (NLP) will be expanded, with the goal of processing queries from engineers unfamiliar with AI/ML. Key questions addressed by the research include whether the software platform will be able to adapt to new utility customers and service areas without sacrificing performance, whether increased data resolution can be effectively leveraged to better predictive power, and whether the platform can continuously improve event prediction over time by learning from historical grid data. 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.