The Time Series Group in DynSys section is involved in development of new statistical methods and their applications in modelling dynamical systems. Most of the research is based on particular practical problems, which is reflected in the amount of research being done in collaboration with various industrial companies and interdisciplinary projects. Our research is applied within the numerous fields including the following topics:
Methodological Research
- Forecasting: Methodological developments related to forecasting encompass statistical modeling aspects, communication of the forecasts, and subsequent decision-making.
- Modelling of Spatial and Spatio-temporal Processes: Some processes require the development of specific spatial or spatio-temporal approaches. This may involve clustering, spatial smoothing (kriging), and spatio-temporal dynamic modeling
- Nonlinear and Nonparametric Methods in Time Series Analysis: Improving models and methods for time series analysis requires constant developments, which may derive from eg. new regression methods, varying-coefficient models, regime-switching concepts, or mixtures of models.
Applied Research
-
Wind Power Forecasting: Wind power forecasting is a significant area of expertise at DTU Informatics, which research efforts concentrated on forecasting at different time scales, and optimal decision-making (management, trading, maintenance planning) based on forecasts.
-
Solar Power Forecasting: Solar power forecasting complements DTU Informatics research activities on optimal management of renewable energies. Even though developments of solar power capacities are fairly limited compared to those related to wind power, this form of renewable is expected to play a significant role in the future.
-
Electricity Price Forecasting: The research on electricity price forecasting mainly focuses on aspects related to the effects of stochastic generation on electricity prices and competitive bidding.
-