Direct solution of the parametric stochastic distribution control problem

Puya Afshar, Amin Nobakhti, Hong Wang

    Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

    Abstract

    The Stochastic Distribution Control (SDC) problem is a generalised form of the minimum variance control problem where non-Gaussian noise distributions are encountered. The problem has been previously solved using two alternative approaches. When it is assumed that the output Probability Distribution Function (PDF) is measurable, then a parameterized controller is obtained. If on the other hand this assumption is removed (which corresponds to most practical cases), then the controller found is no longer parameterisable (i.e. it is a control action sequence). Both these approaches have thus far been solved using local Newtonian methods. In this paper a third alternative is presented which combines the desirable features of the previous two methods by finding a parameterized controller, without having to assume that the output PDF is directly measurable at the same time. In addition, global direct search algorithms are used to avoid convergence to local solutions. The approach is demonstrated on a SISO nonlinear system corrupted by non-Gaussian input noise. ©2009 IEEE.
    Original languageEnglish
    Title of host publicationProceedings of the IEEE Conference on Decision and Control|Proc IEEE Conf Decis Control
    Pages2616-2621
    Number of pages5
    DOIs
    Publication statusPublished - 2009
    Event48th IEEE Conference on Decision and Control held jointly with 2009 28th Chinese Control Conference, CDC/CCC 2009 - Shanghai
    Duration: 1 Jul 2009 → …

    Conference

    Conference48th IEEE Conference on Decision and Control held jointly with 2009 28th Chinese Control Conference, CDC/CCC 2009
    CityShanghai
    Period1/07/09 → …

    Keywords

    • Differential evolution
    • Distribution control
    • Non-Gaussian stochastic systems
    • PI control

    Fingerprint

    Dive into the research topics of 'Direct solution of the parametric stochastic distribution control problem'. Together they form a unique fingerprint.

    Cite this