Modeling the Driving Behavior of Electric Vehicles Using Smartphones and Neural Networks
DOI: 10.1109/mits.2014.2322651
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 modeling eco-driving behaviors for electric vehicles (EVs), a field previously dominated by research on internal combustion engines. The authors identify a critical gap: existing naturalistic driving studies rely on expensive, instrumented vehicles, limiting the scale and impact of data collection. Furthermore, for EVs, accurately estimating remaining battery range is crucial due to limited driving distances and sparse charging infrastructure. The study proposes a system that uses smartphone sensors to capture driving behavior and neural networks to estimate battery consumption, thereby offering a low-cost, scalable alternative to specialized vehicle instrumentation. The experimental design involved 10 drivers navigating a fixed route combining highway and urban segments, including stop-and-go traffic. Data was collected simultaneously using an instrumented Mitsubishi i-MiEV (connected to the CAN bus and high-accuracy GPS) and an Apple iPhone 4. The smartphone logged variables such as speed, acceleration, and jerk at 1 Hz. The researchers first validated the smartphone data against the vehicle’s onboard instrumentation, finding a Pearson correlation coefficient of 99–100%, confirming the smartphone’s sufficiency for data logging. They then extracted relevant features—mean and variance of speed, acceleration, and jerk—partitioning acceleration into positive and negative values and jerk into four specific maneuver types (starting movement, cruising track, starting brake, ending brake). These 14 features formed the input vector for the neural network, while the vehicle’s recorded battery consumption served as the target output. The study employed the SALMON application to train various neural network architectures, including single-layer and multi-layer perceptrons, using bootstrapping to estimate error given the small dataset. Results indicated that a single-layer perceptron with 14 inputs performed best, achieving a test RMS error of 0.0457 (0.731%), which translates to a prediction capability for expected battery consumption higher than 95%. Sensitivity analysis revealed that acceleration processes, particularly the mean value of ending brake jerk and the variance of starting movement jerk, had the highest influence on consumption, while negative acceleration had significantly less impact. The analysis also confirmed a linear relationship between the driving profile features and energy consumption, suggesting that complex non-linear models were unnecessary for this specific dataset. The significance of this work lies in demonstrating that smartphones are a viable, high-fidelity tool for capturing driving data and estimating EV energy consumption without expensive hardware. This approach facilitates mass data collection from volunteer drivers and enables real-time feedback systems to promote eco-driving. The findings suggest that driver behavior, specifically the smoothness of acceleration and braking maneuvers, is a primary determinant of EV efficiency. The authors conclude that this system can effectively inform drivers of their remaining range and efficiency, paving the way for broader applications in intelligent transportation systems and driver assistance technologies.
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.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | OpenAlex-citations | — | — | 1 | 2026-06-19 |
| archive | success | unpaywall | — | — | 2 | 2026-06-26 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-19 |
| chunk | success | chunk | — | — | 1 | 2026-06-19 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-19 |
| promote | success | — | — | — | 1 | 2026-06-19 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-19 |
| verify | success | — | — | — | 1 | 2026-06-26 |
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.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Theoretical Contribution: computational model