Research Projects
Development of realistic pricing models for liberalized power markets
The goal of the project is the development of classes of models, statistical models and fundamental models, to describe and predict the behavior of the liberalized power markets. The main aspects of the project are the modeling of Price Forward Curves (PFC), load and price prediction and Monte Carlo simulation of hourly spot and long term future markets.
The stochastic models use time series analysis and data mining approaches, like Kalman filters, clustering methods, regression and support vector machines (SVM) to identify seasonality and statistical behavior of load and price movements in market and grid data. Because of the high amount of jumps, structural correlation aspects and the possibility of negative prices, an important aspect lies in the robustness of the system identification and data mining algorithms.
The fundamental models build a physical model by modeling the market using models of the market participants (producers, consumers, grid, etc) and the impact of seasonality, holidays and weather on load and demand. The prices can be modeled and simulated with marginal cost curves and given load predictions. Hence fundamental models can be arbitrarily complex; the identification of the important elements plays an important role.
