A Control Strategy Based on Power Forecasting for Micro-Grid Systems

Weather conditions influence RES production, while occupancy affects power consumption. To address this issue, accurate short-term forecasts are necessary for seamless integration of RES with traditional electrical grids in MG systems. A forecasting model is presented for predicting power production and consumption in MG systems, along with battery state of charge (SoC). A control strategy is implemented to balance Demand/Response by considering forecasted and real-time values. The model is based on data collected from a real MG system, and simulation results demonstrate the effectiveness of power forecasting for MG control. The MG system under consideration contains different RES, such as photovoltaic (PV) systems and wind turbines, connected to the traditional electrical grid. The PV power production has priority in satisfying power demand over wind turbine or battery power, as the site receives more radiation during the day than wind. The MG system includes an internet of things (IoT)/Big-data platform for storing and analyzing data, with sensors installed to collect data for training forecasting algorithms and comparing obtained forecasts with observed values. The remainder of this paper is organized as follows: Section 2 describes existing work from literature, while Section 3 introduces different forecasting methods classification used in the literature and the deployed method used in this work, which is an ARIMA model. The ARIMA model is a statistical method used for time series forecasting, which combines autoregressive (AR), moving average (MA), and differencing (D) components. The performance of a MG depends on the climate in which it is located, as weather variables have a significant impact on RES production and electricity consumption. Several approaches have been investigated in the literature for short-term forecasting of RES production and load consumption. For instance, authors in [6] presented a simple approach for short-term forecasting of the power produced by a photovoltaic system connected to the grid, using a one-year database of solar irradiance, cell temperature, and power produced by one megawatt of photovoltaic panels. In [7], tools for short-term load and wind power forecasting were implemented, and two loads predictors based on an autoregressive with exogenous variables model (ARX) and an artificial neural network (ANN) were designed. The authors concluded that the ARX model describes the future load for short prediction horizons better than the ANN, but for larger prediction horizons, the ANN outperforms the ARX model. The work presented in this paper develops an energy management platform that allows managing micro-grid systems using a machine learning method to forecast the control inputs parameters. Unlike the work presented in [3], which is based on real collected data to control the MG system, this present work is based on the same MG platform to develop an intelligent and predictive control strategy for MG systems. The classification of power forecasting techniques is presented in Section 3, which includes statistical methods, machine learning methods, and hybrid methods. The deployed method used in this work is the ARIMA model, which is a statistical method used for time series forecasting. The ARIMA model combines autoregressive (AR), moving average (MA), and differencing (D) components to forecast future values of a time series. The simulation results presented in Section 4 demonstrate the effectiveness of the power forecasting method and a simple control strategy is proposed to balance the Demand/Response power in MG by considering the forecasted input parameters. The results show that the proposed method can accurately forecast power production and consumption in MG systems, and the control strategy can effectively balance Demand/Response power. In conclusion, the integration of RES in MG systems poses significant challenges due to the variability and uncertainty of production and consumption. Accurate short-term forecasts are necessary for seamless integration of RES with traditional electrical grids in MG systems. The proposed forecasting model and control strategy can effectively balance Demand/Response power in MG systems, and the results demonstrate the effectiveness of the proposed method.