A new statistical approach to enhance the performance of model-free optimal control algorithms

Eduardo R.V. Mayen, Reza J. Hamidi, Turaj Ashuri

Research output: Contribution to book or proceedingConference articlepeer-review

3 Scopus citations

Abstract

This paper presents a new statistical approach to enhance the performance of Extremum Seeking Control algorithm. Extremum seeking controls is a model-free optimization algorithm that uses dither and demodulation signals to obtain the derivative information of an input for directional search. To improve the performance and stability of the algorithm, it is needed to simplify the input signal to the system since unsteadiness in the input signal causes the algorithm to be less efficient and fail. Time-averaging and non-statistical interpolation schemes of the input change the statistical characteristics of the signal, and they are of limited help. In this research, we propose a more accurate and representative extension of the algorithm based on the Probability Density Function to preserve the statistical characteristic of the original signal. The results show that the generated signal using an interpolation based on the Probability Density Function enhances the performance of extremum seeking control algorithm in finding the optimal states of wind energy systems.

Original languageEnglish
Title of host publication2018 Multidisciplinary Analysis and Optimization Conference
PublisherAmerican Institute of Aeronautics and Astronautics Inc, AIAA
ISBN (Print)9781624105500
DOIs
StatePublished - 2018
Event19th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, 2018 - Atlanta, United States
Duration: Jun 25 2018Jun 29 2018

Publication series

Name2018 Multidisciplinary Analysis and Optimization Conference

Conference

Conference19th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, 2018
Country/TerritoryUnited States
CityAtlanta
Period06/25/1806/29/18

Fingerprint

Dive into the research topics of 'A new statistical approach to enhance the performance of model-free optimal control algorithms'. Together they form a unique fingerprint.

Cite this