Research pick: Predicting windpower - "Wind weather prediction based on multi-output least squares support vector regression optimised by bat algorithm"

A computer algorithm based on how bats fly at night tracking flying insect prey with their bio-sonar could help meteorologists predict wind patterns more reliably, according to new research published in the International Journal of Embedded Systems. The work could have implications for the optimal running wind turbines for sustainable power generation.

Dingcheng Wang, Yiyi Lu, Beijing Chen, and Youzhi Zhao of the School of Computer and Software at Nanjing University of Information Science and Technology, in Nanjing, China, explain how wind power has come to the fore as one of the most important alternatives to electricity generation without the need to burn fossil fuels. However, it depends on steady winds. The stability of wind turbines is also susceptible to gusting and winds that are too fast-moving.

The team has now tested a bat algorithm model of wind direction and speed that in simulations shows that a multi-output least-squares support vector machine prediction is the most effective approach to prediction. Such predictions would not only help operators ensure the safety of the wind turbines by shutting them down at appropriate times but allow them to manage the output in the context of other power supplies feeding into the local or national electricity grids.

Wang, D., Lu, Y., Chen, B. and Zhao, Y. (2020) ‘Wind weather prediction based on multi-output least squares support vector regression optimised by bat algorithm’, Int. J. Embedded Systems, Vol. 12, No. 2, pp.137–145.