Skip to main content
Infrastructure Resilience Conference 2018

Full Program »

A Tool for Resilient Wind Power Generation: Conditional Kernel Density Estimation Considering Autocorrelation

Wind power is one of the most promising renewable energy. Since wind power production can be significantly affected by weather conditions, the rapid natural change of weather conditions such as wind speed and direction make wind power production unstable. This potential risk to power grid indicates the necessity to better estimate the relationship between weather inputs and power outputs (i.e., power curve) to improve critical system resilience. Power curve has no close form. Not only is there quite a few highly interactive weather factors, but also power output shows large variance, heteroscedasticity and different distributional shape under different input. All these features increase the estimation complexity and indicate the importance of using conditional density estimation. In this paper, we try to improve short term power curve conditional density estimation accuracy by further detected data feature. A key feature we find is that the short term fluctuation of power production shows time dependent pattern. We propose a conditional density estimation model with time dependent structure. With weather inputs changing from time to time, we provide a time dependent rule to allow power output change between difference distributions corresponding to different inputs. In this way, time dependency is expected to provide extra information on more accurate point and density power curve estimation. Wind power data from real cases are used to show its effectiveness. We also notice that this time dependent rule can help avoid using highly interactive multiple inputs, thus avoid curse of dimensionality. With improved short term power curve estimation, we can better monitor, control and arrange back-up unit for wind power, providing resilient electrical grid with secure integration of wind energy.

Yuchen Shi
Future Resilience System/ National University of Singapore, Department of Industrial System Engineering &Management
Singapore

Nan Chen
Future Resilience System/ National University of Singapore, Department of Industrial System Engineering &Management
Singapore

 

Powered by OpenConf®
Copyright ©2002-2016 Zakon Group LLC