A Model Predictive Control Approach for Energy Management in Micro-Grid Systems

MG systems combine RES, storage devices, and the traditional electrical grid to supply power to building loads. However, management and control approaches are required for seamless integration of these systems. A model predictive control (MPC) approach is introduced for efficient MG operation. This approach minimizes the usage of the traditional electric grid by utilizing as much as possible the power generated by RES while optimizing storage device operations. The MPC strategy controls the charge/discharge current of the battery depending on RES production and load consumption variability. The proposed control strategy uses forecasted parameters as inputs, including power production, batteries state-of-charge (SoC), and electricity consumption with daily costs. Real measurements are used as forecasted inputs to the MPC strategy, collected from different sensors such as voltage and current sensors. An IoT/Big-data platform is deployed for data gathering, processing, and predictive analytics using recent machine learning algorithms. These algorithms are used to forecast the MPC input parameters, such as SoC, to determine the right decision according to actual situations. The MPC allows solving an optimization problem by determining the optimal control actions. A Demand/Response control strategy is also developed and integrated into the platform for testing and performance evaluation. The control strategy uses forecasted values of input parameters to predict future actions. The main contributions of this paper are the deployment of a platform for data monitoring and real-time processing that integrates IoT/Big-data technologies, and a predictive control strategy for balancing power flows between PV, batteries, and the traditional electrical grid. Simulations have been conducted using real data from a deployed MG system to study the usefulness of the platform and the effectiveness of the proposed control strategy. The remainder of this paper is organized as follows: Section 2 presents a brief overview of existing work from literature focusing on predictive control principles. Section 3 presents the deployed MG system, and Section 4 introduces the MG equivalent model together with the predictive control strategy. Simulation results are presented in Section 5, and conclusions and perspectives are given in Section 6. MG systems are a collocation of decentralized electrical energy sources, storage devices, and loads connected to the traditional electric grid. Different control strategies have been proposed to manage power sources as a hybrid system using real-time data as input parameters. Recent research efforts have been dedicated to the integration of MPC strategies in MG systems for management and control. A MPC control scheme for wind power generation with a battery energy storage system has been presented to compensate for the intermittency and stochastic nature of power production. The proposed control strategy allows selling more electrical energy at peak demand/price times and storing it during normal periods. A MPC approach for MG control has been presented and deployed for PV power generation in a MG system with mixed storage devices, battery, and hydrogen-based. The predictive controller optimizes operational costs by considering the value of generated energy, the cost of locally stored energy, and operational constraints. The developed MPC approach has been validated in experimental MG, which includes energy storage, grid interaction, and an MPC framework. A mixed integer linear framework for MG modeling and optimization has been illustrated, and the proposed approach was investigated using experimental MG. Another interesting work has proposed a MPC approach for a PV system with storage devices to minimize the bill cost for customers by predicting information concerning PV power generation, load consumption, and electricity price. The predicted values are used to optimize the operation of the PV system and the charge/discharge of the batteries. A strategy for energy management based on MPC principles has been studied and developed for multi-MG systems to achieve the balance between power production and demand while minimizing the cost of power supplied by efficient coordination of extra energy production between MG systems. A MPC approach has been introduced for real-time management of power generated by PV panels with energy storage, allowing modification of PV and energy storage's power according to actual consumption. An electric energy system optimization approach has been proposed by solving economic and environmental dispatch problems using a MPC approach based on a dispatch algorithm that actively relies on direct control of intermittent resources for compensating fast load fluctuations. The development of a MPC approach for MG systems has been focused on the integration of RES, storage devices, and the traditional electrical grid to supply power to building loads. The proposed control strategy uses forecasted parameters as inputs, including power production, batteries state-of-charge, and electricity consumption with daily costs. Real measurements are used as forecasted inputs to the MPC strategy, collected from different sensors such as voltage and current sensors. An IoT/Big-data platform is deployed for data gathering, processing, and predictive analytics using recent machine learning algorithms.