Anzeige
Erschienen in:
2019 | OriginalPaper | Buchkapitel
A Chained Neural Network Model for Photovoltaic Power Forecast
verfasst von : Carola Gajek, Alexander Schiendorfer, Wolfgang Reif
Erschienen in: Machine Learning, Optimization, and Data Science
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
Photovoltaic (PV) power forecasting is an important task preceding the scheduling of dispatchable power plants for the day-ahead market. Commercially available methods rely on conventional meteorological data and parameters to produce reliable predictions. These costs increase linearly with a rising number of plants. Recently, publicly available sources of free meteorological data have become available which allows for forecasting models based on machine learning, albeit offering heterogeneous data quality. We investigate a chained neural network model for PV power forecasting that takes into account varying data quality and follows the business requirement of frequently introducing new plants. This two-step model allows for easier integration of new plants in terms of manual efforts and achieves high-quality forecasts comparable to those of raw forecasting models from meteorological data.