(504f) Exploring the Role of System Knowledge in Data Analytics Using an Iiot-Enabled Smart Manufacturing Testbed

Authors: 
Shah, D., Auburn University
Wang, J., Auburn University
He, Q. P., Auburn University
The emergence of the industrial Internet of Things (IoT) and ever advancing computing and communication technologies have fueled a new industrial revolution. This revolution is happening worldwide to make current manufacturing systems smarter, safer, and more efficient. Although many general frameworks have been proposed for IoT enabled systems and their potential in industrial applications, there is limited literature on demonstrations or testbeds of such systems. Simulation is a powerful tool but the fidelity of the simulated system is limited by the understanding on the system; that is, the model that describes the system. With industrial IoT still in its infancy, there is not sufficient understanding on the property, capacity and performance of IoT sensors to enable accurate simulation. In addition, there is a lack of systematic study on the characterization and representation of data generated by industrial IoT sensors and data analytics challenges associated with industrial IoT sensor data. These are the major motivations for us to build a smart manufacturing testbed equipped with IoT vibration sensors, and to investigate the capabilities of these IoT vibration sensors for process monitoring.

In this work, we will discuss the details of the testbed setup and the characteristics of the data collected from the IoT vibration sensors. We investigate strategies to address IoT sensors’ data challenges and their impacts on data analytics. We show how data visualization and exploration can help identify underlying process patterns and correlations. We demonstrate how data analytics performance and reliability, as well as results interpretability can be significantly enhanced by incorporating system knowledge with data analytics techniques.