Route-Based Online Energy Management of a PHEV and Sensitivity to Trip Prediction

Karbowski, Dominik; Kim, Namwook; Rousseau, Aymeric · 2014 · OpenAlex-citations

DOI: 10.1109/vppc.2014.7007126

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of optimizing energy management in plug-in hybrid electric vehicles (PHEVs) under realistic driving conditions where future speed profiles are uncertain. While optimal control methods like Dynamic Programming require full knowledge of future trips, real-world driving is stochastic. The authors propose a route-based online energy management strategy using the Pontryagin Minimum Principle (PMP), which is more compatible with real-time implementation than global optimization techniques. The study specifically evaluates the sensitivity of this PMP controller to trip prediction errors, aiming to determine its robustness when the predicted speed profile differs from the actual driving behavior. The methodology involves modeling a Prius-like PHEV with a medium all-electric range using the Autonomie forward-looking simulation environment. The baseline control strategy employs an "EV+CS" approach, depleting the battery until a threshold is reached before switching to charge-sustaining mode. The proposed PMP controller optimizes battery power demand to minimize fuel consumption by minimizing a Hamiltonian function, utilizing an equivalence factor to balance fuel and electric energy costs. To simulate real-world uncertainty, the authors generated ten stochastic speed profiles for a 36-km itinerary in Munich using a GIS-assisted process that combines deterministic route attributes (e.g., speed limits, grades) with Markov chains trained on recorded vehicle data. The PMP controller was tested against these profiles, with equivalence factors tuned to target a final state of charge of 30%. The results demonstrate that the PMP controller achieves fuel savings in all tested scenarios compared to the baseline strategy. When the equivalence factor was optimally chosen for each specific trip, the controller reduced fuel consumption by an average of 4.6%, with maximum savings of 5.8% on individual trips. The PMP strategy improved efficiency by keeping the engine off longer, operating it more efficiently when active, and reducing energy recirculation losses, albeit with more frequent engine starts. Crucially, the study analyzed the sensitivity to sub-optimal equivalence factors, simulating a scenario where the prediction used for tuning differed from the actual trip. Even with a fixed, non-adaptive equivalence factor, the controller maintained an average fuel saving of 3.3%. However, the results showed high sensitivity for specific trips, where a fixed factor could yield significant savings on one trip but increase consumption on another. The significance of this work lies in demonstrating that PMP-based energy management is viable for real-world applications despite imperfect trip predictions. The findings suggest that while a static equivalence factor provides consistent benefits, an adaptive algorithm that periodically adjusts the factor during the trip is necessary to maximize savings and mitigate penalties caused by prediction discrepancies. The study validates the use of GIS-assisted stochastic trip prediction as a practical method for generating inputs for online energy management controllers, bridging the gap between theoretical optimal control and implementable vehicle strategies.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• telematics crash prediction