Multi-Channel IoT-Based Ensemble-Features Fault Diagnosis for Machine Condition Monitoring

This paper proposes a multi-channel internet of things (IoT)-based industrial wireless sensor network (IWSN) with ensemble-features fault diagnosis for machine condition monitoring and fault diagnosis. In this paper, the rolling bearing is taken as an example of monitored industrial equipment due to its wide use in industrial processes. The rolling bearing vibration signals are measured for further processing and analysis. On-sensor node ensemble feature extraction and fault diagnosis using Back Propagation network are then investigated to address the tension between the higher system requirements of IWSNs and the resource-constrained characteristics of sensor nodes. A two-step classifier fusion approach using Dempster-Shafer theory is also explored to increase diagnosis result quality. Four rolling bearing operating in cage fracture, rolling ball spalling, inner ring spalling and outer ring spalling are monitored to evaluate the proposed system. The final fault diagnosis results using the proposed classifier fusion approach give a result certainty of at least 94.21%, proving the feasibility of the proposed method to identify the bearing-fault patterns. This paper is conducted to provide new insights into how a high-accuracy IoT-based ensemble-features fault diagnosis algorithm is designed and further giving advisable reference to more IWSNs scenarios.