Interactive graphics on large datasets drives remote condition monitoring on a cloud

Abstract

We demonstrate a new system for condition monitoring using the cloud. The system combines state of the art pattern search capability with youShare, a platform that allows people to run compute-intensive research in an ordered manner over the internet. Data from sensors distributed across one or more assets at one or more sites are uploaded to the cloud compute resource. The uploading triggers the deployment of a range of pattern search services, and is capable of rapidly detecting novel patterns in the data. The outputs of these processes are archived as a matter of course, but are also sent to a further service which processes the data for remote visualisation on a web browser. The system is built in Java, using GWT and RaphaelGWT for graphics rendering. The design of these systems must satisfy conflicting requirements of data currency and data throughput. We present an evaluation of our system that involves processing data at a range of frequencies and bandwidths that are commensurate with commercial requirements. We show that our system has the potential to satisfy a range of processing requirements with minimal latency, and that the user experience is easily sufficient for rapid interpretation of complex condition monitoring data.