(184j) Robust Model Predictive Control for Smart Grid Integrated with Solar Power and Energy Storage System Under Regular and Abnormal Loads

In this work, a scenario-based robust model predictive control (MPC) integrated with a load estimator is proposed to minimize the power loss and stabilize voltage fluctuation when photovoltaic (PV) cells and energy storage system (ESS) are connected to the grid.

Due to the high level installation of rooftop photovoltaic cells in California, the solar power becomes an important renewable energy source injected into the grid. However, the rapid growth of capacity and penetration of the PV may lead to serious fluctuation of voltage in the grid and thereby degrade the power quality of the distribution line. Moreover, the collection of solar energy is dependent on the weather and solar irradiation. Thus it may not be available when unexpected power demand occurs. In the meanwhile, it is worthwhile to note that ESS such as Lithium battery has progressed significantly in terms of the capacity, charging/discharging rate and life time. The combination of PV and ESS may provide an attractive scheme to meet energy demands and stabilize voltage in a smart grid. Nevertheless, the resulting integrated system contains fast dynamics from the PV inverter and slow dynamics from the Lithium battery. Thus how to coordinate, control and optimize the fast electrical components and slow chemical processes become an important research topic.

In the controller design, the power demand data from Southern California Edison (SCE) and California Independent System Operator (CAISO) is used to build different scenarios for residential/commercial/industrial regions, respectively. Then a scenario-based robust MPC is developed to coordinate PV and ESS, reduce power loss and compensate the abnormal power load. Both battery and dynamic load models are carefully selected to simulate the real process and then integrated into the DistFlow equations [1] to describe the power of the distribution line. Directly solving MPC with such complex multi-scenario model is very time consuming. Therefore, a fast algorithm is proposed to find a high quality solution within reasonable time.

To formulate a closed-loop control scheme, an optimization-based power load estimator is also developed to deliver states information to MPC. This
estimator is able to reconstruct the real/reactive power load according to the noise-corrupted measurement data. The accuracy of this estimator can be quantified and guaranteed especially when high power load happens. Such closed-loop control framework enables the grid to quickly recover from undesired disturbances (i.e. abnormal loads) by effectively changing the real and reactive powers provided by PV cells and ESS.

Reference
[1] M. E. Baran and F. F. Wu, Optimal capacitor placement on radial distribution systems,” IEEE Trans. Power Del., vol. 4, no. 1, 725–734, Jan. 1989.