An efficient embedded detection scheme for bearing condition monitoring

Abstract

Rolling element bearings play an essential role in rotating machines. It is estimated that more than 90% of rotating machines contain bearings, and therefore their failure may lead to a catastrophic failure of the entire machine. Traditional monitoring systems rely on wired sensors to collect measurements from machines and then transmit said measurements to a central station for condition analysis. However, a wired connection leads to higher installation and maintenance costs for each sensor. Wireless Sensor Nodes (WSNs) have become increasingly popular in the field of condition monitoring, since their installation is easier, and the cost is lower. Sensor nodes have limited battery capacity, thus energy conservation is vital in WSNs. Vibration measurements are an effective condition monitoring technique for detecting defects at an early stage. Conventionally, the measured raw data are transmitted to a central station for analysis; this reduces the lifetime of WSNs due to the energy cost of communication. On the other hand, local processing (on-node) has the potential to reduce communication power consumption by processing raw data instead of transmitting, hence conserving battery life. The signal-based methods used to extract the signal features are as follows: time domain, frequency domain, and time-frequency domain. This research focuses mainly on diagnosing the bearing conditions with processing the signal in embedded wireless node. Introduce a cost-effective wireless condition monitoring system for the detection of bearing defects. The system will minimise the consumption of the node’s power, which will make it possible to implement optimised signal processing techniques. The node will also be capable of transmitting information wirelessly to a central station. The proposed algorithm is based on evaluating the bearing conditions to determine if more analysis is required through a two-stage process. First, time-domain statistical indicators are implemented to evaluate the conditions of the signal. Second, if there are abnormal conditions detected, then an envelope analysis will be applied to determine the location of the fault. The main contribution of this research is to implement an on-node autonomous and online condition monitoring system for detecting bearing defects in real applications.

More information

Identifiers

ORCID for Neil White: orcid.org/0000-0003-1532-6452

Catalogue record

Export record