The inaugural OUC-sponsored University of Central Florida (UCF) Meter Data Science Competition provided teams of graduate students the chance to analyze real electrical and water data and apply their findings to help OUC identify either new electric vehicle (EV) charging usage or suspected utility theft.
OUC announced the two winning teams ― each analyzed a different use case ― at UCF’s virtual 2021 Big Data Analytics Symposium on March 18.
Fifteen teams of two to four graduate students initially entered the competition, which began in September 2020. While working on their chosen use case, each team could collaborate with subject matter experts at OUC and tap into their knowledge. Six teams completed their assignments.
In use case No. 1, EV charging detection, teams were tasked with disaggregating customer electric consumption data, using 15-minute intervals to help them identify the addition of EVs in OUC’s service territory. They were to leverage data to help OUC:
- Predict future EV penetration by geographic area (i.e. planning for charging infrastructure deployment)
- Identify electric distribution transformers most heavily impacted by current (and forecasted) EV penetration
- Target market EV rebates and other OUC programs and products (i.e. rooftop solar) to this unique customer segment
- Extend the concept/model to identify other common, household loads/appliances
- Understand how time-of-use and other pricing signals may change customers’ EV charging behavior
OUC awarded $2,500 to the winning team of Daniel Mariano, Ryan Jones, Anna Perdue and Dharitri Rathod.
“The winning team created a comprehensive working solution that incorporated outside details such as public charging stations into the data models to help OUC predict confirmed EV owners,” said Dawn Frye, Manager of Smart Grid Data. “OUC is looking forward to leveraging the models to identify and predict EV charging stations in OUC’s service territory.”
In use case No. 2, utility theft, teams were asked to identify outliers indicative of complex theft and tampering that typically does not trigger a meter alarm. Their findings were to provide OUC with the ability to:
- Calculate and leverage electric-to-water usage rations to identify theft by region, subdivision, etc.
- Identify and incorporate other contributing factors (correlations) that increase the likelihood of theft or tampering at a particular premise, within a region, etc.
- Create visualizations, scoring indexes and other outputs that could allow the Revenue Assurance team to effectively prioritize field investigations and mitigate revenue losses
The team of Jianbin Zhu, Yuan Du and Phuong Pho claimed the top prize of $2,500.
“They created solid models and visualizations that OUC can use to identify outliers in terms of low energy use, especially as it relates to water consumption at the same premise,” said Eddie Fee, Director of Meter Services. “Basically, homes with very low energy use but average or above average water usage may indicate a revenue or theft issue for OUC.”
Click here to hear winning team members comment on the real-world experience they gained from the competition.