Machine learning to deal with increasing uncertainties in power systems
Mrs Laurine DUCHESNE will publicly defend his thesis entitled "Machine learning to deal with increasing uncertainties in power systems"
With the energy transition, the share of renewable energy sources is growing rapidly. However, renewable generations such as solar and wind power generations are weather-driven, intermittent and difficult to predict. Uncertainties in power systems are therefore increasing, complicating the tasks of transmission system operators. As a result, the field of electric power systems is in the search for novel methods for power systems planning and operation.
The focus of this thesis is the operation planning context. In this context, the transmission system operator takes decisions under uncertainty, from several days ahead to several hours ahead, to ensure that adequate resources will be available in real-time operation to meet the electricity demand. For instance, the operator may decide to postpone planned maintenance or redispatch generating unit outputs. To tackle the increasing uncertainties in power systems, we propose in this thesis new decision-making tools leveraging machine learning. In particular, we propose to exploit simplified models, called proxies, of the behavior of the operator in response to realization of uncertainties (e.g. demand and renewable generation) over the future target horizon considered. These proxies must be fast and yet accurate, in order to replace traditional heavy models of the behavior of the operator and allow one to anticipate the impact of a look-ahead decision over a very large number of possible future scenarios (e.g. plausible realizations of demand and renewable generation) in a short amount of time. In this thesis, we propose a methodology to build these proxies with machine learning and we illustrate how they can be used in operation planning.