Machine Learning-Based Battery State-of-Health Prediction for Unmanned Aerial Vehicles Predictive Maintenance

Jul 16, 2026

Upload by Ezra

Battery state-of-health (SoH) prediction estimates remaining capacity by modeling lithium battery degradation across its life cycle. Machine learning models can infer remaining capacity from operational measurements such as voltage, current, and temperature, enabling efficient predictive maintenance for unmanned aerial vehicles (UAVs). UAV-specific SoH modeling remains challenging due to scarce data, strong variability across battery types, and limited onboard sensing. This research builds an ML pipeline using knowledge transfer and ResNet-50 feature extraction from image-transformed time series, trained on labeled flight experiments for two LiPo capacities.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages6

Reading Time15 min

Introduction and Background

Battery Degradation and SoH Prediction

Predictive Maintenance for UAVs

Data Collection Challenges

Proposed ML Pipeline and Knowledge Transfer

Machine Learning-Based Battery State-of-health Prediction for Unmanned Aerial Vehicles Predictive Maintenance Jiarui Xie Department of Mechanical Engineering McGill University Montreal, Canada [jiarui.xie@mail.mcgill.ca](jiarui.xie@mail.mcgill.ca) Lingchen Kong Department of Mechanical Engineering McGill University Montreal, Canada [lingchen.kong@mail.mcgill.ca](lingchen.kong@mail.mcgill.ca) Elaine Mosconi Department of Information Systems and Quantitative Management Methods Université de Sherbrooke Sherbrooke, Canada [elaine.mosconi@usherbrooke.ca](elaine.mosconi@usherbrooke.ca) Mohamed Rami Latreche Department of Information Systems and Quantitative Management Methods Université de Sherbrooke Sherbrooke, Canada [mohamed.rami.latreche@usherbrooke.ca](mohamed.rami.latreche@usherbrooke.ca) Yaoyao Fiona Zhao Department of Mechanical Engineering McGill University Montreal, Canada [yaoyao.zhao@mcgill.ca](yaoyao.zhao@mcgill.ca) Sean Smith VOZWIN Inc. Pointe-Claire, Canada [smith.sean@vozwin.co](smith.sean@vozwin.co)m Abstract— Battery state-of-health (SoH) prediction aims to estimate the remaining capacity by modeling battery degradation through its life cycle. Machine learning (ML)-based SoH models can accurately predict the battery remaining capacity based on voltage, current, and temperature. Battery SoH prediction for unmanned aerial vehicles (UAVs) is a crucial yet overlooked domain with data scarcity and high variability. Accurate battery SoH information contributes to efficient predictive maintenance, enhancing UAV profitability and flight safety. However, UAVs are compatible with a variety of batteries and the available data for each type of battery are scarce. Furthermore, the available input features from UAV batteries are limited to the built-in sensors because of the lightweight requirements. This research aims to develop an ML pipeline for UAV battery SoH prediction while mitigating data scarcity using knowledge transfer. 342 and 289 flight experiments have been conducted to collect operational data from lithium polymer batteries of 2200 mAh and 1100 mAh, respectively. Voltage, current, and throttle are selected as the input features of the ML model according to the existing literature and sensor availability. The remaining capacity is measured at every 10th experiment to label the dataset. To address data scarcity, the time-series data acquired from the experiments are transformed into images to utilize the image feature extraction capability of a pretrained ResNet-50. Finally, accurate SoH prediction models were obtained using transfer learning between two battery types. Keywords—Machine learning, predictive maintenance, battery, state-of-health, unmanned aerial vehicle I. INTRODUCTION AND BACKGROUND Lithium batteries, which feature high energy density, long life span, and cost-efficiency, have become a cornerstone in the field of energy storage technology and electromobility, particularly revolutionizing the electric vehicles (EVs) and Mathematics of Information Technology and Complex Systems (MITACS) Accelerate program [grant number IT13369] and Réseau SDG Innovation Network-Collaborative R&D projects in digital, intelligent and sustainable transformation.ere. If none, delete this text box. unmanned aerial vehicles (UAVs) [1, 2] . However, the degradation of lithium batteries along the charging and discharging cycles increases the risk of vehicle accidents [3] . Predictive Maintenance (PdM) is widely implemented for UAVs and EVs to enhance their reliability and extend the overall lifespan. Different from traditional maintenance technologies, PdM can monitor the system state of health (SoH) and predict potential faults before they occur. PdM continuously monitors and processes the machine operational data to estimate the health indicators (HIs), SoH, and remaining useful life (RUL), which are the most critical parameters of lithium batteries [4-6] . Machine Learning (ML) is a prom

FAQs

Why is battery state-of-health prediction important for UAVs?

It helps estimate remaining capacity and supports predictive maintenance, improving flight safety and enabling more efficient UAV operation.

What input features are used for the machine learning model?

Voltage, current, and throttle are selected based on literature and the availability of onboard sensors.

How does the research mitigate data scarcity across different battery types?

It uses knowledge transfer between two lithium polymer battery types and applies transfer learning so the model can learn from limited data.

Jul 16, 2026

Upload by Ezra