Cloud-based control of industrial cyber-physical systems

This paper presents an implementation of a control algorithm to a cloud system. The motivation is that cloud implementations of low-level systems in the production industry are gradually becoming more common. Microsoft Azure platform is utilized for the cloud-based control and the case is tested using a customized laboratory model, which can be presented as an agent in a typical production system. The model offers the regulation of a ball on an inclined surface and uses two asynchronous motors connected to frequency converters to control the position of the ball. These frequency converters are controlled by a Programmable Logic Controller (PLC). Windows Communication Foundation (WCF) services and Azure IoT Hub were selected to be used with the cloud-based control system. Experimental results have shown our solution can control the system with sampling period equal or higher than 100ms. The latency of WCF service is at around 100ms and latency of Azure IoT Hub is at around 1000ms, so the prediction algorithms could be implemented in the cloud for the latter. This research also shows the feasibility of migrating machine learning algorithms that demand high computing power to the cloud to reduce the computing burden on the local control units.