Development of Publicly Available Online Microgrid Database with AI/ML Big Data Analytics Driven by Data Collection Strategies and Providing Quantitative Insightful Information

JouleOne Inc., in partnership with the University of California San Diego (UCSD), proposes an SBIR project to develop a publicly accessible online database of microgrid (MG) installations. This database will utilize artificial intelligence (AI) and machine learning (ML) driven analytics to provide an interactive, semantic-based user experience that delivers technical and commercially relevant insights. This database will complement the existing DOE Microgrid Installation Database hosted by ICF International, which currently offers limited demographic and technology-related data. The modernized platform will become the primary destination for MG quantitative information and analytics, enabling decision-making around planning, design, operation, and optimization of MGs. Industry and government stakeholders will have access to tools for assessing the merits of potential investments in MGs, which should accelerate deployments. Additionally, MG operators, end-users, consultants, researchers, policymakers, and utility companies will benefit from the structured data compilation, analytical tools, and decision-making capabilities. While UCSD has developed core data elements, challenges remain in collecting, transforming, and analyzing diverse MG data in a shared multi-tenant database and making insights publicly accessible. The project will be executed in two phases to design a database solution and bring it to a commercial scale. Phase I will focus on planning and developing a database application using Amazon Web Services (AWS) tools. The key objectives include: (i) investigating a representative sample of MGs data fields and developing a data architecture, (ii) identifying AWS tools including AI algorithms, for the online application, (iii) deploying the prototype database on AWS, and (iv) developing a preliminary user interface and marketing strategies. JouleOne will leverage the UCSD MG and DERConnect database schemas and incorporate relevant data elements from the Common Information Model (CIM). The database will ingest both time-series and static data, capturing a range of data elements such as location, capacity, voltage, load profiles, and DER capacities, along with data necessary for metrics of reliability, resiliency, and operational costs. Phase II will (i) scale the database and train the ML model by using datasets from DERConnect simulations and third-party sources, (ii) iteratively improve the application’s capabilities, including its AI/ML algorithms and interactive user interface, and (iii) develop and test the user interface. The platform will feature incentives such as benchmarking services, community networking capabilities, and a marketplace for suppliers to advertise MG-related products, while also allowing users to monetize excess energy through aggregated energy trading. The approach is anticipated to lead to an on-prem database for analytics and facilitating web data sharing. The AI algorithms will be optimized for data discovery and decision-making, having use outside of the MG domain. When commercialized, the platform will bridge the MG information gap, enhance energy reliability and resilience by promoting MG deployments, and support grid modernization.