Title:  Advanced Control Design of Multi-Port Autonomous Reconfigurable Solar Power Plant

Committee:

Dr. Maryam Saeedifard, ECE, Chair, Advisor

Dr. Lukas Graber, ECE

Dr. Santiago Grijalva, ECE

Dr. Suman Debnath, Oak Ridge National Lab

Dr. Mahshid Amirabadi, Northeastern

Abstract: The multi-port autonomous reconfigurable solar power plant (MARS), which is an integration of photovoltaic (PV) and energy storage system (ESS) to the transmission ac grid and a high-voltage direct current (HVdc) link, is designed to provide frequency response and reject disturbances in the grid with continued operation and reduced transient instability. The complex architecture of the MARS and the intermittent nature of PV underlies the need for developing simple, efficient, and easily generalizable control methods for a MARS-type system that integrates multiple power sources. The presence of different sources such as PV and ESS in each arm of the MARS causes uneven distribution of active power among different submodules (SMs) present in MARS, thereby leading to unbalanced modules’ capacitor voltages that may impact system stability under various operating conditions. Moreover, in the case of partial shadings, shaded PV SMs will suffer from decreased injected power causing power mismatch between different SMs in the MARS system. An energy balancing control (EBC) method is proposed to balance the capacitor voltages of different types of SMs. Moreover, the machine-learning-based EBC criteria are proposed to improve the system efficiency and reduce the switching frequency. The proposed EBC criteria can disable/enable the EBC depending on the MARS input power dispatch commands with high accuracy. To simplify the design process and acquire a better simulation result, the thesis further proposes a neural network-based power mismatch elimination (NNPME) strategy. The NNPME strategy employs ESS to its maximum capacity and the dc and ac circulating currents to transfer power between the SMs, arms, and legs of the MARS system. The aforementioned controls are data-driven methods that require a large amount of simulation data. A model predictive control (MPC) is proposed for more accurate and efficient control of MARS. It can optimally allocate uneven power of ESS and PV in one arm and counteract capacitor voltage deviations. The proposed advanced controls facilitate the efficient control and energy management of a system with multiple input power sources like MARS to fully utilize its potential with an extended operating region while maintaining high efficiency.