Railway Track Fault Detection Using Selective MFCC Features from Acoustic Data

Railway track faults may lead to railway accidents and cause human financial loss. Spatial, temporal, weather elements, wear tear, ballast, loose nuts, misalignment, cracks leading accidents. Manual inspection of such defects is time-consuming prone errors. Automatic provides a fast, reliable, unbiased solution. However, highly accurate fault detection challenging due the lack public datasets, noisy data, inefficient models, etc. To obtain better performance, this study presents novel approach that relies on mel frequency cepstral coefficient features from acoustic data. The primary objective increase performance. As well as designing an ensemble model, we utilize selective using chi-square(chi2) have high importance with respect target class. Extensive experiments were carried out analyze efficiency proposed approach. experimental results suggest 60 features, 40 original 20 chi2 produces optimal both regarding accuracy computational complexity. A mean score 0.99 was obtained machine learning models collected Moreover, performance significantly than existing approaches; however, vary in real-world settings.