If a man gives no thought about what is distant, he will find sorrow near at hand.
Being aware of what is going to happen in future can always make people feel safe. This is why weather forecasting is so important in people’s daily life. It is the same situation when talking about wind power prediction for electric power systems, especially for the modern smart grids.
As large-scale of energy cannot be stored, the balance between generation and consumption of electricity has to be kept to ensure the safety and stability of power system operation. However, the fast development and large-scale integration of wind power into smart grids have broken this balance, due to the intermittency and stochasticity of natural wind, which makes it quite difficult to make an accurate prediction.
The emergence of smart grids has brought both challenges and opportunities in the development of wind power prediction techniques. The reliability, flexibility, and efficiency of grid operation put forward higher requirements for accurate and efficient prediction. Nevertheless, the development of high-performance communication & computing technologies makes it possible to further enhance the prediction paradigms in smart grids.
Wind power prediction will be carried out based on massive data collected from weather prediction center, meteorological stations, meters in wind farms and power systems. The problem is how to organize and make full use of these data. Researchers have improved prediction by recognizing various patterns from data, such as spatial correlation and smoothing effects among regional wind farms, similarity searching and matching in history time series. Extracting concise and essential information from redundant and disordered data can be helpful in boosting both efficiency and accuracy of wind power prediction.
Another important point is that, as a smart grid requires, the wind power prediction should have the ability of self-adaptation. The characteristics of wind power output are constantly evolving with the complex meteorological systems. A prediction model couldn’t perform well all the time unless it keeps adjusting itself in real-time to adapt to the changing environment. The time-varying issue can also be solved by combining several individual models that present different performance in different situations. So, there is always an excellent model standing out to cope with its corresponding situation, which makes the combined prediction better than any individual prediction.
All the advanced technologies, e.g., big data, artificial intelligence, can be tried to find their possibility in wind power prediction. Let the wind power prediction model see more, learn more, think itself and finally be smart enough to play its role well in the smart grid environment.
L. Ye & Y. N. Zhao
College of Information and Electrical Engineering
China Agricultural University (CAU), China