Model Predictive Control For Energy Management Of Microgrids

As a main part of smart grid, microgrids have great importance in renewable energy sources promoting, energy conserving, and carbon emission reducing. The main objective of microgrid energy management is to provide the loads demand energy accurately with minimum operating costs and in consideration of the technical conditions and physical equipment constraints. As a result, not only the balance between energy supply and demand is ensured in the short term, but also energy optimization scheduling and economic operation are achieved in the long term. However, due to the intermittent of renewable energy, system complexity, control behavior of binding and many other characteristics, the traditional energy management strategy is difficult to meet the requirements in practical. Therefore, to ensure microgrids operate safely, reliably, and economically and to realize the application of microgrids, it is necessary to study the energy management problems of microgrids.Model predictive control can not only deal with multivariable processes, constraints, and multi-objective optimization, but also achieve good dynamic control performance under the receding horizon optimization scheme, which provides a new research approach for the optimization of microgrid energy management.In this thesis, the main research results include:(1) For energy management of microgrid with multiple smart loads, a scaled time-hierarchal model predictive control strategy is proposed to optimize energy management problem of a microgrid. According to the power flow among energy modules, a hierarchical system model and a scaled time-hierarchal energy optimization management problem are established. The upper layer centralized controller is to optimize the charge/discharge time and power of storage, controllable supply power adjustment and dispatch for the aggregators. The optimization problem in aggregators is to meet loads demand in real time. Meanwhile, in order to improve the disturbances caused by the randomness of renewable energy and variant of loads, a multi-scaled optimization scheme is applied. At a slow scale, the upper optimization problem is solved, and the optimized energy dispatching in the long term can be achieved; at a fast scale, the energy balance between supply and demand of smart users can be realized in a short term.(2) For a multi-microgrid system which consists of multiple microgrids, energy optimization method based on distributed predictive control (DMPC) is proposed. Based on energy flow among microgrids, model description is established. By solving the distributed predictive control optimization problems which are based on the imbalance between supply and demand, energy efficiency of renewable energy is improved and energy demand is met. As a result, the amount of computation for solving optimization problems is reduced under the premise of real time requirements.(3) Energy allocation strategy based on real pricing factor is introduced to optimize the energy distribution in multi-microgrid system from a market perspective. Firstly, based on the power flow among microgrids, the descriptions of multi-microgrids system model and the energy management optimization problem are given. Then, a dual decomposition approach is imposed to decompose the optimization problem into two layers and a Lagrangian multiplier is introduced to achieve the optimal solutions by iterating and coordination. This method not only achieves an optimal energy allocation for multi-microgrid system, but also provides a reference factor for real time pricing making in multi-microgrid system.