Tamara Todic’s research while affiliated with University of Strathclyde and other places

... The effect of imbalanced data in common NILM datasets is also an important issue on NILM tasks. To mitigate this issue, an active learning framework [65], which aims to reduce the amount of sample data needed to train a model, may be employed. Finally, it is also interesting to investigate the performance of the proposed method on identifying the on/off states of the electrical appliances after the load is disaggregated in our future work. ...

Reference:

Sequence-to-point learning based on spatio-temporal attention fusion network for non-intrusive load monitoring

... Energy metering has been used to investigate and describe energy consumption on dairy farms (Rajaniemi et al., 2017, Rajaniemi et al., 2015. Todic et al. (2022) used non-intrusive load monitoring to determine the electricity consumption of some of the main electricity consuming systems on three German dairy farms, such as the vacuum pump and compressors. While Todic et al. (2022) were able to successfully quantify the electricity demand of vacuum pumps, compressors, water pumps, and other electricity consuming systems with an accuracy above 95 %, their method was not always reliable with reported accuracy for a milking robot on one farm as low as 35 %, and transferability across farms was sighted as an issue. ...

Reference:

Farm electricity system simulator (FESS): A platform for simulating electricity utilisation on dairy farms