Photovoltaic power forecasting using total sky imager

Fluctuation and intermittent are existed in photovoltaic (PV) power generation. Massive PV grid-connection may bring adverse effect to the running of power system. The forecast of output power of the PV power plant, will balance the dispatch of the conventional energy (thermal power, hydropower, etc...

Full description

Saved in:
Bibliographic Details
Main Author: Ye, Penghao
Other Authors: Gooi Hoay Beng
Format: Final Year Project
Language:English
Published: 2014
Subjects:
Online Access:http://hdl.handle.net/10356/61283
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:Fluctuation and intermittent are existed in photovoltaic (PV) power generation. Massive PV grid-connection may bring adverse effect to the running of power system. The forecast of output power of the PV power plant, will balance the dispatch of the conventional energy (thermal power, hydropower, etc) and the photovoltaic power, thereby ensures the power quality. Irradiance, the main factor that affects the output power of PV power generation, leads to the fluctuation of the output power due to its sudden change, and brings huge difficulty to the PV power prediction. The traditional forecasting method, which is largely based on physical and statistical models, may bring about the lag of the prediction results. Thus it is unable to provide accurate prediction results continuously. Relevant researches have indicated that the main reason for the sudden change of irradiance is the cloud coverage against the sunlight. The prediction of cloud coverage above the solar panel can be achieved. This is done by all-day sky observation and image acquisition using a ground-based remote sensing cloud detection equipment. Through the image analysis technique, it is possible to extract the feature of the clouds’ future movement. One meteorological instrument, Total Sky ImagerTM, is designed to capture the cloud coverage images. Once the images acquired, they can be used for correcting the irradiance data. Then the corrected data may be input to train a specialized artificial neural network (ANN) model. When the input data accumulated to a certain amount, the well-trained model can be applied in the forecast of the solar power output.