Towards a particle-flow framework for uncertainty quantification, with applications in wind plant system dynamics and control

Abstract

The method of particle flow, originally developed for solving Bayes' formula, is extended to provide a general transformation between two probability distributions. It is shown that this can enable the use of a chaos expansion for uncertain or stochastic dynamic systems. The approach is demonstrated on a simple example. The method is potentially relevant for the real-time control of wind plants. For example, it could be used to obtain a probabilistic estimate of the wind field inside a wind farm using a combination of measurements from the turbines and modelling. Time lags and wake effects make this problem non-Gaussian, which the particle-flow method is well-suited to handle. It remains to be seen, however, whether there is a compelling reason to use a chaos expansion for stochastic dynamic analysis. Functions implementing the methods have been programmed in the Julia language.

Towards a particle-flow framework for uncertainty quantification, with applications in wind plant system dynamics and control