Nonparametric conditional interval forecasts for PV power generation considering the temporal dependence

Songjian Chai, Ming Niu, Zhao Xu, Loi Lei Lai, Kit Po Wong

    Research output: Chapter in Book/Conference paperConference paper

    5 Citations (Scopus)

    Abstract

    The high penetration of solar PV generations brings about significant challenges for decision-makers of power system operation due to high volatility and uncertainty it involves. In recent years, it has been demonstrated by many researchers that the probabilistic interval forecast could significantly facilitate some decision-making cases, such as storage optimization, market bidding, reserves setting, as it can provide the uncertainty information associated with the point estimations. This paper proposes a nonparametric conditional interval forecast method for PV power generation which can capture the interdependence among the real power output and their point forecasts within all forecasting horizons of interests. The proposed model is tested using the dataset of PV generation power measurements and day-ahead point forecasts in Belgium. The results based on reliability and interval score performance metrics illustrate the effectiveness of the proposed model.

    Original languageEnglish
    Title of host publicationIEEE Power and Energy Society General Meeting
    EditorsPaula Traynor
    PublisherIEEE, Institute of Electrical and Electronics Engineers
    Volume2016-November
    ISBN (Electronic)9781509041688
    DOIs
    Publication statusPublished - 10 Nov 2016
    Event2016 IEEE Power and Energy Society General Meeting, PESGM 2016 - Boston, United States
    Duration: 17 Jul 201621 Jul 2016

    Conference

    Conference2016 IEEE Power and Energy Society General Meeting, PESGM 2016
    CountryUnited States
    CityBoston
    Period17/07/1621/07/16

    Fingerprint Dive into the research topics of 'Nonparametric conditional interval forecasts for PV power generation considering the temporal dependence'. Together they form a unique fingerprint.

    Cite this