Finding fault types of BLDC motors within UAVs using machine learning techniques

Area of Science:

• Aerospace Engineering
• Electrical Engineering
• Computer Science

Background:

• Unmanned Aerial Vehicles (UAVs) are increasingly used across diverse sectors like environmental monitoring, healthcare, and transportation.
• Brushless DC (BLDC) motors, powered by advancements in battery technology, are crucial for UAV propulsion.
• BLDC motor failures in UAVs can result in critical control loss and potential accidents, necessitating effective diagnostic methods.

Purpose of the Study:

• To propose and evaluate a novel machine learning-based approach for predicting BLDC motor defects in UAVs.
• To compare the efficacy of three distinct machine learning models—K-Nearest Neighbor (KNN), Support Vector Machine (SVM), and Bayesian Network (BN)—in identifying these motor flaws.

Main Methods:

• The study employs machine learning techniques to analyze data indicative of BLDC motor health.
• Performance evaluation of KNN, SVM, and BN models was conducted for defect prediction accuracy.

Main Results:

• The comparative analysis determined the effectiveness of each machine learning model in predicting BLDC motor defects.
• Results highlight the potential of machine learning for enhancing the reliability and safety of UAV operations.

Conclusions:

• Machine learning offers a promising solution for early detection of BLDC motor failures in UAVs.
• The findings support the development of predictive maintenance strategies to mitigate risks associated with UAV motor malfunctions.