The goal of this project is to help City of Gainesville site and optimize E-bus charging stations, as the local transit agency will realize the goal of future bus electrification. This project analyzes real-time operational data, predicts energy needs, and designs an optimal model that can balance cost, convenience, and service coverage.
The real-time bus GPS data (Oct 2022 - Mar 2023) is extracted from Public APIs (acquired by the City of Gainesville) with Python. Then stored in the Postgres database on AWS RDS.
Code: ingest_realtime_data.ipynb.
-
Energy Consumption Predictive Modeling: we developed a predictive model to estimate electric energy consumption by route, time of day, and geographic patterns, helping forecast charging demands accurately. Code:
energy consump.ipynb. -
Optimization Modeling of E-bus charging stations: We applied weighted K-means clustering and scenario-based modeling to identify optimal charging station locations, maximizing coverage and operational efficiency. Hyperparameter tuning improved the model performance, achieving 95% service coverage with minimal cost trade-offs. Code:
KMeans.ipynb.
The findings are presented in the conference:
We are now designing a dashboard using JavaScript, HTML/CSS to show real-time bus activity, predicted coverage, and optimized charging station sites: Link to the Github