I-Corps: Analyzing Customer Behavior for Energy Usage Moderation

Cornell University, Ithaca NY

Investigators

Abstract

The broader impact/commercial potential of this I-Corps project is the potential development of an optimization algorithm that could identify energy consumption patterns that may enable moderation and management of electricity consumption. This technology could help in infrastructure planning, enhance customer relations by understanding the customer's energy behavior, and provide individualized or customized feedback to avoid circuit overload, power quality issues, and power outages (blackouts) which are major problems for the electricity utility and microgrid companies. There is an unmet need to understand customer energy behavior patterns in different geographies and demographics to help create informed decision-making on infrastructure planning and deployment. In addition to application in electricity utility companies, this technology can potentially be adapted for deployment in other applications that require utility infrastructure planning and consumer behavior change, such as gas and water distribution. This I-Corps project is based on the development of a proprietary automatic multi-parametric optimization and machine learning clustering algorithm that could forecast upcoming load on the electrical grid. This is done by analyzing overall and personalized energy consumption by consumers based on variables reflecting localized geography, weather conditions, seasonal conditions (winter vs summer), and time and day electricity consumption as well as demographics. Using multi-parametric variables, a regularization-based optimization algorithm is developed that decouples the contribution of each influencing factor in load consumption to avoid circuit overload and power quality issues. To avoid blackouts, the output of the multi-parametric optimization is combined with a proprietary machine-learning clustering technique that analyzes the load consumption behavior/ pattern in different demographics and maps behavior with the population density to improve load forecast for better accuracy to avoid power outages. The combination of multi-parametric optimization and clustering helps to identify customer load patterns and hence could help in identifying highly accurate load forecast for load balancing which will eliminate blackouts in a region. 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.