Model
Digital Document
Publisher
Florida Atlantic University
Description
Microgrid is an essential part of the nation’s smart grid deployment plan, recognized especially for improving efficiency, reliability, flexibility, and resiliency of the electricity system. Since microgrid consists of different distributed generation units, microgrid scheduling and real-time dispatch play a crucial role in maintaining economic, reliable, and resilient operation. The control and optimization performances of the existing online approaches degrade significantly in microgrid applications with missing forecast information, large state space, and multiple probabilistic events. This dissertation focuses on these challenges and proposes efficient online learning and optimization-based approaches.
For addressing the missing forecast challenges on online microgrid operations, a new fitted rolling horizon control (fitted-RHC) approach is proposed in Chapter 2. The proposed fitted-RHC approach is designed with a regression algorithm that utilizes the empirical knowledge obtain from the day-ahead forecast to make microgrid real-time decisions whenever the intra-day forecast data is unavailable. Simulation results show that the proposed fitted-RHC approach can achieve the optimal policy for the deterministic case study and perform efficiently with the uncertain environment in the stochastic case study.
For addressing the missing forecast challenges on online microgrid operations, a new fitted rolling horizon control (fitted-RHC) approach is proposed in Chapter 2. The proposed fitted-RHC approach is designed with a regression algorithm that utilizes the empirical knowledge obtain from the day-ahead forecast to make microgrid real-time decisions whenever the intra-day forecast data is unavailable. Simulation results show that the proposed fitted-RHC approach can achieve the optimal policy for the deterministic case study and perform efficiently with the uncertain environment in the stochastic case study.
Member of