Back to search

Cornell University

From Range Anxiety to Driver Advisory: A Physics-Informed Reinforcement Learning Framework for Electric Vehicle Energy Management

Abstract

dc:description.abstract

The global transition to electric vehicles brings a fundamental challenge: drivers remain uncertain about whether their vehicle has enough energy to complete a journey. This range anxiety stems from three interconnected problems: unpredictable charging times, inaccurate energy consumption estimates, and the lack of intelligent systems that jointly optimize routing and charging while accounting for long-term battery health. This thesis addresses these challenges progressively. We first develop a hybrid physics-based and reinforcement learning framework for charging time prediction, capturing the nonlinear CC-CV charging dynamics and how they evolve as a battery ages. We then develop a physics-informed residual learning model for energy consumption prediction, revealing that driving behavior alone accounts for up to 79% variation for identical hardware. Finally, we integrate both models into a unified joint routing and charging optimization framework with an explicit driver-advisory output layer, producing three actionable per-trip recommendations: a route with charging waypoints, a per-stop charging target calibrated to the battery's current health and charger power, and a driving-mode recommendation. Two non-intuitive findings emerge. First, a closed-form detour threshold showing when routing to a non-nearest charger is simultaneously faster, cheaper, and less damaging to the battery. Second, the optimal charging ceiling is a non-linear function of both battery health and charger power, making the common 80% heuristic suboptimal for any battery below 95% of its original capacity. The result is a physically grounded, data-efficient system that treats EV energy management as a single interconnected challenge rather than a collection of isolated problems.

Degree

thesis:*
Name thesis:degree_name
M.S., Mechanical Engineering
Level thesis:degree_level
Master of Science
Discipline thesis:degree_discipline
Mechanical Engineering
Grantor
Cornell University
Year dc:date.issued
2026

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Aryasomayajula, Praharshitha
Chair dc:contributor.chair
  • Malikopoulos, Andreas
Committee member dc:contributor.committeemember
  • Gao, Huaizhu

Subjects

dc:subject × 2

Rights

Language dc:language.iso
en

Identifiers

dc:identifier.*
Dc Identifier Other
Aryasomayajula_cornell_0058O_12690
https://www.proquest.com/LegacyDocView/DISSNUM/32676112

Chain of custody

source
Harvested from
Cornell University
Base URL
ecommons.cornell.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Aryasomayajula, Praharshitha. From Range Anxiety to Driver Advisory: A Physics-Informed Reinforcement Learning Framework for Electric Vehicle Energy Management. Master of Science thesis, Cornell University, 2026. https://hdl.handle.net/1813/126303