PSO-based Adaptive Artificial Neural Networks for Wind Speed Forecasting
碩士 === 國立成功大學 === 電機工程學系碩博士班 === 97 === The power generated by wind turbines varies with the continuous fluctuation of wind speed. Consequently, wind energy conversion systems (WECS) cannot be dispatched like conventional power generation units. If the short-term wind speed can be forecast more accu...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | zh-TW |
Published: |
2009
|
Online Access: | http://ndltd.ncl.edu.tw/handle/42427138922696859528 |
id |
ndltd-TW-097NCKU5442048 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-097NCKU54420482016-05-04T04:17:06Z http://ndltd.ncl.edu.tw/handle/42427138922696859528 PSO-based Adaptive Artificial Neural Networks for Wind Speed Forecasting 粒子群最佳化之適應性類神經網路風速預測系統 Jian-Shun Wen 溫建順 碩士 國立成功大學 電機工程學系碩博士班 97 The power generated by wind turbines varies with the continuous fluctuation of wind speed. Consequently, wind energy conversion systems (WECS) cannot be dispatched like conventional power generation units. If the short-term wind speed can be forecast more accurately, power generation may be scheduled more efficiently and costs can be reduced accordingly. The proportion of wind power is expected to increase in the electric power system, and accurate forecasting of wind energy will become a basis of electric power pricing. The aim of the thesis is at building an artificial neural network (ANN) wind-speed forecasting system and testing the system built with different time step sizes as inputs to the ANN. The particle swarm optimization (PSO) algorithm is used to improve the learning rate of the ANN and enable online adjusting of the system. In order to verify and evaluate the proposed short-term wind forecasting model, data are collected for real time-series wind speed and wind power generation from Taipower Penghu Chungtun (from January to March 2004) and for wind speed data from the Central Weather Bureau, Hsinchu Station. The results of the proposed method are compared with the Persistence model widely used by the industry based on the errors between the real wind speed and the forecasted wind speed. The results of the forecast wind speed are also employed to provide a range of forecast wind generation output as references for reserves of conventional power generation units. Hong-Tzer Yang 楊宏澤 2009 學位論文 ; thesis 61 zh-TW |
collection |
NDLTD |
language |
zh-TW |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國立成功大學 === 電機工程學系碩博士班 === 97 === The power generated by wind turbines varies with the continuous fluctuation of wind speed. Consequently, wind energy conversion systems (WECS) cannot be dispatched like conventional power generation units. If the short-term wind speed can be forecast more accurately, power generation may be scheduled more efficiently and costs can be reduced accordingly. The proportion of wind power is expected to increase in the electric power system, and accurate forecasting of wind energy will become a basis of electric power pricing.
The aim of the thesis is at building an artificial neural network (ANN) wind-speed forecasting system and testing the system built with different time step sizes as inputs to the ANN. The particle swarm optimization (PSO) algorithm is used to improve the learning rate of the ANN and enable online adjusting of the system. In order to verify and evaluate the proposed short-term wind forecasting model, data are collected for real time-series wind speed and wind power generation from Taipower Penghu Chungtun (from January to March 2004) and for wind speed data from the Central Weather Bureau, Hsinchu Station. The results of the proposed method are compared with the Persistence model widely used by the industry based on the errors between the real wind speed and the forecasted wind speed. The results of the forecast wind speed are also employed to provide a range of forecast wind generation output as references for reserves of conventional power generation units.
|
author2 |
Hong-Tzer Yang |
author_facet |
Hong-Tzer Yang Jian-Shun Wen 溫建順 |
author |
Jian-Shun Wen 溫建順 |
spellingShingle |
Jian-Shun Wen 溫建順 PSO-based Adaptive Artificial Neural Networks for Wind Speed Forecasting |
author_sort |
Jian-Shun Wen |
title |
PSO-based Adaptive Artificial Neural Networks for Wind Speed Forecasting |
title_short |
PSO-based Adaptive Artificial Neural Networks for Wind Speed Forecasting |
title_full |
PSO-based Adaptive Artificial Neural Networks for Wind Speed Forecasting |
title_fullStr |
PSO-based Adaptive Artificial Neural Networks for Wind Speed Forecasting |
title_full_unstemmed |
PSO-based Adaptive Artificial Neural Networks for Wind Speed Forecasting |
title_sort |
pso-based adaptive artificial neural networks for wind speed forecasting |
publishDate |
2009 |
url |
http://ndltd.ncl.edu.tw/handle/42427138922696859528 |
work_keys_str_mv |
AT jianshunwen psobasedadaptiveartificialneuralnetworksforwindspeedforecasting AT wēnjiànshùn psobasedadaptiveartificialneuralnetworksforwindspeedforecasting AT jianshunwen lìziqúnzuìjiāhuàzhīshìyīngxìnglèishénjīngwǎnglùfēngsùyùcèxìtǒng AT wēnjiànshùn lìziqúnzuìjiāhuàzhīshìyīngxìnglèishénjīngwǎnglùfēngsùyùcèxìtǒng |
_version_ |
1718255567169912832 |