Efficient Exploration of Long Data Series: A Data Event-driven HMI Concept

Today’s easy access to data, low cost sensors and data transmission infrastructure leads to an abundance of data about complex systems in many domains like industrial process control, network intrusion detection or maritime surveillance. Analyzing this data can take a lot of effort and often cannot be fully automated. As it is hard to fully automate such analysis tasks, we present an HMI framework that supports an analyst in exploring and navigating through multiple time series of data. It is a semi-automatic approach that uses algorithms for automatically labelling low-level events in the data, but leaves the task of evaluation and interpretation to the human operator. These events are highlighted on specific time bars in the HMI framework. It enables the analyst to 1) summarize the main features of the data series, 2) filter it depending on the analysis objective, 3) identify and prioritize relevant section in the data and 4) directly jump to these sections. We present the theoretical concept of the HMI framework and demonstrate it on a process control application for hybrid energy systems.