In Denmark, district heating is very common.
DTU collects sensor data of three of these district heating systems, including measurements of pressure, flow and temperature within those systems.
However, due to a variety of reasons such as transmission or sensor failures, data points from these time series measurements are missing.

Such a lack of data points within one or multiple time series can negatively impact downstream applications.
Hence, it is advisable to reconstruct these missing data points beforehand.

This thesis addresses the questions resulting from that task.
We will investigate and implement approaches to recover such missing data points.
To this end, it is also necessary to examine whether there are also anomalies in the existing data, and how we can clean or even correct those.
While sensor failures affect a single sensor for a longer period of time, transmission failures may affect multiple sensors at the same time.
We examine, which differences in the recovery approach this necessitates.

Moreover, the suitable recovery mechanism also depends on whether a real time recovery is necessary, or whether we can perform the recovery on a periodic time scale.
In this thesis, we disregard the former and concentrate on the latter part.

Naturally, we also conduct an evaluation of our recovery methods.
We examine, whether there is a single method that is clearly favorable, or whether we should choose different methods for different failure or parameter types.