Renewable generation and energy storage systems are technologies which evoke the future energy paradigm. While these technologies have reached their technological maturity, the way they are integrated and operated in the future smart grids still presents several challenges. Microgrids appear as a key technology to pave the path towards the integration and optimized operation in smart grids. However, the optimization of microgrids considered as a set of subsystems introduces a high degree of complexity in the associated control problem. Model Predictive Control (MPC) is a control methodology which has been satisfactorily applied to solve complex control problems in the industry and also currently it is widely researched and adopted in the research community. This paper reviews the application of MPC to microgrids from the point of view of their main functionalities, describing the design methodology and the main current advances. Finally, challenges and future perspectives of MPC and its applications in microgrids are described and summarized.

In this article, a stochastic programming method is used to control a residential microgrid where possible disturbance realizations over a horizon are shaped into a tree structure with a given probability. This tree is used by model predictive controllers within a hierarchical and distributed scheme. The upper layer is characterized by a smart contract deployed in the blockchain, acting as a fully distributed coordinator that performs control tasks and collects and distributes relevant data. The bottom layer consists of individual agents that solve locally a reduced tree-based model predictive control (TBMPC) problem and interact with the smart contract to iteratively optimize the overall problem in a distributed fashion. The performance of the approach is demonstrated through several simulations in which various power-trading configurations are assessed.

This article focuses on the improvement of the real-time operation of the microgrid based on model predictive control (MPC), taking into account the stochastic nature of renewable resources and demand, as well as fault-tolerant mechanisms. Thus, a fault-tolerant strategy (FTS) module is inserted into the control loop, which is devoted to fault diagnosis and isolation, as well as setting reconfiguration actions that may involve modifications through the model parameters, the objective function, and the constraints used in the optimization problem solved by the MPC. Meanwhile, the MPC controller is responsible for managing the power of the microgrid, where two stochastic approaches are compared: tree-based MPC (TB-MPC) and chance-constrained MPC (CC-MPC). Experiments carried out in a simulated full-scale laboratory microgrid under various conditions demonstrate the effectiveness of the proposed algorithms in dealing with power equipment faults.

Stochastic Model Predictive Control (SMPC) is an effective decision-making method in applications where uncertainties play a significant role. This work introduces a non-linear formulation of SMPC specifically designed for cancer therapy. The proposed method considers the stochastic nature of tumor growth, non-linear dynamics, and a potential side effect of the treatment. Through one-year simulations, the results showcase the effectiveness of this strategy in controlling drug dosing.