EnableDigitalDH – Data science for better data quality: missing link for successful data-driven digitalisation of district heating

In digitalised district heating networks, customers or smart meters continuously send their measurement data to the operator via a data link. The general control system is connected to energy producers and heat consumers (“demand side”), so that measurement data of the entire system are continuously available. To exploit the potential of this data, analysis methods based on machine learning and dynamic modelling have been tested for years. However, the performance and precision of these methods depends heavily on the quality of the input data used. Therefore, the availability of large datasets with good data quality are essential for all of these methods.

Unfortunately, the data quality of real measurement data is often poor: gaps, measurement noise, outliers, transmission errors, etc. are often the limiting factor for applying analysis methods in real systems and thus actually exploiting the optimization potential. To address these issues this project focuses on improving the data quality of measurement data of district eating systems. To do this a novel approach combining physical models with data driven approaches will be explored.

As a result of the project, a comprehensive evaluation and characterization of data-driven analysis methods with special focus on their data quality requirements and of data science methods to improve the data quality of measurement data will be available, based on real measurement data. The project results have the potential to increase the impact of data-driven analysis methods for modern energy systems. The findings and results of the project will be scientifically disseminated; selected data sets, code modules and documentation will be made available in an open access repository.