On incorporating forecasts into linear state space model Markov decision processes.

Area of Science:

• Energy Systems Engineering
• Control Theory
• Applied Mathematics

Background:

• Weather forecast data is increasingly vital for optimizing energy system operations.
• Integrating dynamic forecast information into control models presents significant challenges.

Purpose of the Study:

• To develop a computationally tractable framework for incorporating dynamically revealed weather forecasts into energy system control.
• To ensure consistency between evolving forecasts and the energy system's state.

Main Methods:

• An augmented state space model with linear dynamics was formulated.
• The Martingale Model for Forecast Evolution (MMFE) was employed to enforce forecast-state consistency.
• The model generates jointly Markovian dynamics, enabling the creation of tractable Markov Decision Processes (MDPs).

Main Results:

• The proposed formulation successfully integrates dynamically revealed forecast information into an MDP framework.
• The MMFE consistency requirements are enforced within a computationally tractable MDP.
• This represents the first instance of enforcing MMFE consistency within a tractable MDP for energy systems.

Conclusions:

• The developed augmented state space model provides a robust and computationally tractable method for utilizing weather forecasts in energy system control.
• This approach enhances the ability to manage energy systems dynamically in response to evolving weather predictions.
• The work contributes to the mathematical foundations of energy systems, as part of the theme issue 'The mathematics of energy systems'.