Akademik Veri Yönetim Sistemi
Çukurova Üniversitesi Mühendislik Fakültesi dergisi, cilt.40, sa.3, ss.581-592, 2025 (TRDizin)
This study evaluates the effectiveness of machine learning and deep learning-based models in imbalance fault diagnosis, using Fast Fourier Transform (FFT) for frequency-based feature extraction. Imbalance problems in industrial machinery reduce equipment lifetime and increase maintenance costs. Therefore, in order to improve early detection and predictive maintenance processes, vibration data was analysed and frequency components were extracted using FFT and classification was performed using machine learning based models. In the experimental process, Support Vector Machines, Random Forest and Multi-Layer Perceptron (MLP) models were applied and their classification performance was compared using metrics. According to the results, the MLP model performed best with the highest accuracy rate (99%). As it is the model that best learns the frequency components extracted by the FFT, it most accurately captures the distinct patterns of imbalance errors. This study contributes to machinery imbalance fault diagnosis by combining FFT-based feature extraction with machine learning algorithms.