Energy load prediction for Open Charge Point Protocol - compliant electric vehicle charging stations

This thesis develops an Open Charge Point Protocol (OCPP)–compliant Charging Station Management System (CSMS) and end-to-end data pipeline to enable short-term energy forecasting for individual EV charging stations. The system captures and processes OCPP transactions in PostgreSQL, deriving per-session energy for consistent forecasting targets. Using this platform, four models—Random Forest, K-Nearest Neighbors (KNN), Long Short-Term Memory (LSTM), and Prophet—are evaluated with expanding-window cross-validation on 30 stations and Adaptive Charging Network (ACN) datasets. Key findings show that data regularity and density strongly influence accuracy: high usage, structured sites achieve higher R2 values, while sparse, irregular stations remain challenging. Aggregating to 2–4 hour intervals improves R2 at busy sites but increases absolute errors elsewhere. The study contributes (i) an operational OCPP-compliant CSMS, (ii) a method for transforming telemetry into forecasting targets, and (iii) comparative guidance on model selection and time resolution for operational energy management.

Name thesis:degree_name

Master of Applied Science (MASc)

Discipline thesis:degree_discipline

Electrical and Computer Engineering

Grantor

University of Ontario Institute of Technology

Year dc:date.issued

2025

Author dc:creator

• Lee, Hyon

Advisor dc:contributor.advisor

• Azim, Akramul

Language dc:language.iso

en

Handle dc:identifier.uri

https://hdl.handle.net/10155/2009

OAI identifier oai:identifier

oai:ontariotechu.scholaris.ca:10155/2009