Machine condition monitoring with IoT sensor (vibration signal analysis)

Abstract

Due to internal deterioration characteristics, monitored signals from vibrating machines exhibit highly non-stationary behavior and are highly transient. The research project for this thesis is conducted in cooperation with Treon Oy for product improvement of a vibration sensor which contains descriptive information on how the scalar indicators of the vibration signal feature in an IoT condition monitoring sensor works and verifies that the scalar key performance indicators from the sensor are giving the value as it should. The testing process includes acquiring, calculating, and comparing data with sensor-produced scalar data through external calculation. This IoT-based vibration analysis model uses an IoT gateway to collect data from vibration-producing test setups and stream vibration signals. It uses an Industrial node sensor with a frequency range up to 6kHz. The industrial node sensor targeted application acquires vibration from a vibration shaker machine, communicates the data as events, and sends it to clouds. The service running in the cloud named data parser receives the data from node sensor through MQTT protocol. It keeps the data records for performing analysis and monitoring on any platform. The retrieved data is essential for the calculations of the features. Different messages can be sent to industrial node through the MQTT broker to get different kinds of data. Still, the main goal for this research project is to get the raw data from the vibration signal, which is then calculated for comparison with scalar key performance indicators.