The problem of controlling energy systems (generation, transmission, storage, investment) introduces a number of optimization problems that need to be solved in the presence of different types of uncertainty. I highlight several of these applications, using a simple energy storage problem as a case application. Using this setting, I describe a modeling framework that is based on five fundamental dimensions and that is more natural than the standard canonical form widely used in the reinforcement learning community. The framework focuses on finding the best policy, where I identify four fundamental classes of policies consisting of policy function approximations (PFAs), cost function approximations (CFAs), policies based on value function approximations (VFAs), and look-ahead policies. This organization unifies a number of competing strategies under a common umbrella.
All Science Journal Classification (ASJC) codes
- Artificial Intelligence