Publications

How innovation management and digital entrepreneurship models can turn the technical use cases of the I-GReta smart grid project (battery storage, building energy management, planning energy communities and bidirectional charging) into viable business cases.

Every building labels its monitoring data its own way, so analyses like anomaly detection have to be set up again for each one. This paper sorts and labels raw building automation data automatically: it tests which known operating rules each data stream follows, from its statistical features, and finds the streams that belong together from their conditional probabilities. On unlabelled data from a real building, it reliably identified the data types of certain sensors and actuators.

A three-step, simulation-based method that tests closed-loop building controllers with little user input: it runs the controller against a model of the system it controls across scenarios, scores it with KPIs, and compares it with an estimate of optimal control. Applied to two air handling unit controllers, the improvements it pointed to raised their control quality (ITAE) by 85% and 63%.

An open-source platform, built on FIWARE, that brings high-rate measurements (one reading a second) from a site's energy equipment, such as its battery system and smart meters, into one place: stored, visualised, and open to other applications.

Errors in control logic are a common cause of wasted energy in buildings. For a real building in operation, this paper evaluates its control from monitored data alone: KPIs assess its performance, the causes of control errors are found automatically in the cloud, and the control code is improved with components from the open-source AixOCAT library.

A hardware controller and its digital twin, tested side by side in real time on a cloud platform. Their output time series differ, but their performance follows the same patterns, with KPIs within 3 to 8% of each other, so testing a virtual controller is a good indicator of how the hardware will behave.

My master's thesis in Software Systems Engineering (graded 1.3): a simulation of offshore wind farms, and an optimisation of where their turbines go.