Department of Electrical Engineering National Kaohsiung University of Science and Technology Kaohsiung City, Taiwan

---

Abstract: The project has installed a disaster prevention smart solar microgrid at Dawu community, Pingtung County, Kaohsiung City which is 1,000 meters above sea level. The energy management system (EMS) in the disaster prevention smart solar microgrid needs to forecast the complementary grid power in order to improve the stability and reliability of the whole system and maximize the profit of operation during peak hours. An advanced meter infrastructure (AMI) has been installed in parallel with the smart solar microgrid which collected important parameters of the system in real-time. In this paper, a back propagation neural networks are applied to predict the complementary grid power by utilizing the load power, the battery voltage, the battery charging/discharging current and the voltage and current from the solar panel. The 1-min data in six months from 1st Nov. 2018 to 30th Apr. 2019 are collected by the AMI and utilized as input to train the back propagation neural network. The accuracy of the forecasting complementary grid power will be analyzed and evaluated with different method of normalization, the time interval of input data and the number of neural network nodes in the hidden layer. The simulation outputs show that the RMSE achieves the lowest value of 0.0014 in November and 0.0232 in average. The predicting complementary grid power reaches the best MAPE value of 0.1845% in January and 4.3876% in average. In addition, the coefficient of determination achieves the highest value of 0.9998 in February and 0.9598 in average. These achieved simulation outcomes have proved the great effect of varying input parameters on the predicting complementary grid power. The simulation outputs are utilized in the EMS in the disaster prevention smart solar microgrid for predicting higher accurate complementary grid power, improving stability and maximizing profitability of the whole system.

---

Full Paper.pdf

Copyright © 2020 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.