Fan Fault Diagnosis Using Acoustic Emission and Deep Learning Methods

The modern conception of industrial production recognizes the increasingly crucial role maintenance. Currently, maintenance is thought as a service that aims to maintain efficiency equipment and systems while also taking quality, energy efficiency, safety requirements into consideration. In this study, new methodology for automating fan procedures was developed. An approach based on recording acoustic emission failure diagnosis using deep learning evaluated detection dust deposits blades an axial fan. Two operating conditions have been foreseen: No-Fault, Fault. No-Fault condition, are perfectly clean in Fault material artificially created. Utilizing pre-trained network (SqueezeNet) built ImageNet dataset, acquired data were used build algorithm convolutional neural networks (CNN). transfer applied images spectrograms extracted from recordings fan, two conditions, returned excellent results (accuracy = 0.95), confirming performance methodology.