At 10:00 on June 11th, Prof. Duan Dongliang from the University of Wyoming in the United States participated in academic exchanges at the invitation of Professor Hu Xiaoya and gave an academic report entitled “Learning-Based Very-Short-Term Solar Prediction”. The report was hosted by and more than 30 teachers and students from our college and the School of Electronial Information and Communications attended the meeting.
Professor Duan Dongliang studied in the Department of Electronics and Information Engineering in Huazhong University of Science and Technology from 2002 to 2006 and obtained a bachelor's degree in engineering. From 2006 to 2010, he studied at the Department of Electronic Engineering of the University of Florida and obtained a master's degree in science in 2009. Besides, he studied in 2010-2012 in electrical engineering in Colorado State University and received a Ph.D. He has been teaching at the University of Wyoming in the United States since 2012 and is currently an Associate Professor (Lifetime). His research work mainly involves research in the fields of smart grid, wireless communication, and signal processing. His research projects include the program of US Department of Energy (project amount: $5.5 million) and the US National Science Foundation program. He has published more than 40 high-level academic papers, and was selected for the best paper award competition by the IEEE PES General Meeting in 2017 and received the best paper award in IEEE International Conference on Communication Systems in 2018. He has been an organizer and reviewer for many academic conferences such as ICNC, ICC, and Globecom. He is currently a senior editor at IET Communications and has served as a reviewer for top journals in the fields of smart grid, wireless communications and signal processing (e.g. IEEE Trans. on Reviewer of Power Systems, IEEE Trans. on Smart Grid, IEEE Trans. on Wireless Communications, IEEE Trans. on Communications, IEEE Trans. on Signal Processing, etc.).
In this report, Professor Duan first pointed out the current severe situation of global energy shortage and showed the necessity of exploring renewable energy development. Then, he expounded the shortcomings of the current medium-term and short-term forecasting methods of solar energy and the instability of ultra-short-term forecasting. Next, he introduced the combination of information extraction of the sky image and environmental data learning to achieve ultra-short-term prediction of solar illumination. Among them, the movement of clouds was the main factor to cause the high variability of solar activity. Professor Duan introduced the convolutional neural network (CNN) to analyze the sky image to determine whether there was cloud interference. On this basis, a long-short-term memory network (LSTM) prediction framework utilized other relevant meteorological data (such as temperature, wind speed, etc.) to predict environmental trends. By comparison with the existing schemes, he proved that the proposed prediction framework could effectively maintain the stability of the system and the prediction accuracy was also significantly improved. Finally, at the end of the report, the teachers and students on the scene gave very positive affirmations to the wonderful report and had a full exchange and discussion with Prof. Duan on the doubts and difficult points in the report.
Mr. Duan’s report was profound but in simple language. The teachers and students had frequent interactions and benefited a lot. The atmosphere was warm. Finally, Mrs. Hu Xiaoya expressed her gratitude to Professor Duan for his wonderful report.
