Functional Tree Forest: A Hybrid Bagging-Boosting Ensemble Machine Learning Approach for Battery Fault Diagnosis in Electric Vehicles
- 1Department of Computer Engineering, Dokuz Eylul University, Izmir 35390, Turkey
Abstract
Accurate battery fault detection and diagnosis (FDD) is essential for ensuring the safety, health monitoring, and operational performance of electric vehicles (EVs). Yet, existing research still suffers from issues such as limited classification accuracy due to non-linear feature interactions and insufficient interpretability. To address these limitations, this study proposes the Functional Tree Forest (FTF), a hybrid Bagging-Boosting ensemble machine learning approach that combines multiple functional trees with LogitBoost-based modeling, while preserving model explainability. Unlike conventional decision trees that rely solely on axis-aligned univariate attribute splits, the FTF architecture dynamically integrates multivariate linear regression functions at both internal nodes and terminal leaf nodes, capturing complex feature interactions. The proposed method was evaluated on five publicly available EV battery fault datasets. The experimental results showed that FTF achieved top numerical accuracy across 13 baseline machine learning approaches, including linear models, conventional tree-based classifiers, ensembles, and network-based topologies. Furthermore, when compared with results directly given in the state-of-the-art studies utilizing identical datasets under varying experimental protocols, FTF yielded competitive performance relative to previously reported literature, although the magnitude of the performance difference varied across datasets.
Related
- Conformal Prediction and Multi-Ensemble Gradient Boosting for Subtype-Selective Binding Affinity Estimation Across Human Adenosine Receptors
- A Gradient Boosting-Based Machine Learning Model for the Prediction of the Hardness of Thermoplastic Polymers
- Predicting rate of penetration (ROP) with machine learning models: A physics-validated comparative case study of Gradient Boosting Regressor (GBR), Random Forest (RF), and K-Nearest Neighbours (KNN)
- Canonical-Ensemble Response from Spin-Adapted Time-Dependent Density Functional Theory
- Ensemble Molecular Surface Electrostatic Potential Descriptors as Informative Features for Active Learning
BibTeX
@article{kubilay2026functional,
title = {{Functional Tree Forest: A Hybrid Bagging-Boosting Ensemble Machine Learning Approach for Battery Fault Diagnosis in Electric Vehicles}},
author = {Ibrahim Kubilay and Derya Birant},
journal = {Batteries},
year = {2026},
month = oct,
doi = {10.3390/batteries12110443},
url = {https://www.mdpi.com/2313-0105/12/11/443},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Functional Tree Forest: A Hybrid Bagging-Boosting Ensemble Machine Learning Approach for Battery Fault Diagnosis in Electric Vehicles
AU - Ibrahim Kubilay
AU - Derya Birant
PY - 2026
DA - 2026/10/26
JO - Batteries
DO - 10.3390/batteries12110443
UR - https://www.mdpi.com/2313-0105/12/11/443
AB - Accurate battery fault detection and diagnosis (FDD) is essential for ensuring the safety, health monitoring, and operational performance of electric vehicles (EVs). Yet, existing research still suffers from issues such as limited classification accuracy due to non-linear feature interactions and insufficient interpretability. To address these limitations, this study proposes the Functional Tree Forest (FTF), a hybrid Bagging-Boosting ensemble machine learning approach that combines multiple functional trees with LogitBoost-based modeling, while preserving model explainability. Unlike conventional decision trees that rely solely on axis-aligned univariate attribute splits, the FTF architecture dynamically integrates multivariate linear regression functions at both internal nodes and terminal leaf nodes, capturing complex feature interactions. The proposed method was evaluated on five publicly available EV battery fault datasets. The experimental results showed that FTF achieved top numerical accuracy across 13 baseline machine learning approaches, including linear models, conventional tree-based classifiers, ensembles, and network-based topologies. Furthermore, when compared with results directly given in the state-of-the-art studies utilizing identical datasets under varying experimental protocols, FTF yielded competitive performance relative to previously reported literature, although the magnitude of the performance difference varied across datasets.
SN - 2313-0105
ER -