Hybrid optimal deep learning with IoT based smart based monitoring and maintenance system for axial flow fan using feature optimization

Abstract

Axial flow fan is a mechanical fan that generates airflow in the same direction as its rotational axis. These fans find widespread use in various applications, including ventilation, cooling, and air circulation across industrial, commercial, and residential settings. However, designing these fans can be challenging in the fan manufacturing industry due to the need to accommodate diverse operating conditions. This complexity arises from the fact that multiple design parameters significantly influence fan performance, requiring careful consideration and optimization to ensure efficient operation across various scenarios. In this paper, we present a technique for monitoring and maintenance of axial flow fans using hybrid optimal deep learning with IoT system. Our method leverages the pre-trained U-Net architecture to extract hidden features effectively from the dataset. Furthermore, we introduce an improved triple tree-seed optimization (IT2SO) algorithm for feature optimization, which identifies the most optimal features among the extracted ones. To make informed decisions about axial flow fan process monitoring, we propose the deep boosted hybrid learning (DBHL) technique as the decision model to maintain the proper operation. To validate the effectiveness of proposed IT2SO+DBHL technique, we have conducted experiments using the air movement and control association international (AMCA) dataset. The results demonstrate the superior performance of our monitoring approach compared to existing techniques across various evaluation measures.