Anomaly Detection in Brushless Fan Systems using integration of wavelet approach and convolutional neural network
Hayder A. Abdulrahem · Thi-Qar University Journal for Engineering Sciences · 2024
In this research, trustworthiness is improved, and the probability of malfunctioning in DC fan systems is decreased was addressed. This research suggests a new method that employs artificial intelligence to detect faults by combining Convolutional Neural Networks (CNN) with wavelet analysis for better anomaly detection. Ten fans were initially used in the experiment; three were faulty, and seven operated correctly. Data for all fans was collected using an Arduino. Wavelet features were employed to identify faults, followed by a CNN classifier (. Specifically, a binary CNN classifier) to determine the presence or absence of faults. The results were very promising, demonstrating significant potential for improving the reliability of DC fan systems.