Pretražite po imenu i prezimenu autora, mentora, urednika, prevoditelja

Napredna pretraga

Pregled bibliografske jedinice broj: 636228

Electric Load Forecasting using Multivariate Meta- learning


Matijaš, Marin
Electric Load Forecasting using Multivariate Meta- learning, 2013., doktorska disertacija, Fakultet elektrotehnike i računarstva, Zagreb


CROSBI ID: 636228 Za ispravke kontaktirajte CROSBI podršku putem web obrasca

Naslov
Electric Load Forecasting using Multivariate Meta- learning

Autori
Matijaš, Marin

Vrsta, podvrsta i kategorija rada
Ocjenski radovi, doktorska disertacija

Fakultet
Fakultet elektrotehnike i računarstva

Mjesto
Zagreb

Datum
24.05

Godina
2013

Stranica
143

Mentor
Krajcar, Slavko

Ključne riječi
Artificial Neural Networks; Demand Forecasting; Electric Load; Electricity Consumption; Retail Electricity; Estimation; Gaussian Processes; General Load Forecaster; Least Squares Support Vector Machines; Meta-learning; Meta-features; Power; Prediction; Short-Term Load Forecasting; STLF; Support Vector Regression; SVM; Time-Series

Sažetak
Meta-learning is a powerful approach for model selection. So far it has only been applied to univariate time-series forecasting. In this dissertation, meta-learning is used for multivariate time-series forecasting. It is organized as a bi-level meta-learning with an ensemble for classification on a higher, meta- level comprising of 7 algorithms. On the lower, forecasting level, 7 different regression algorithms are used. The proposed approach is applied to load forecasting. This approach enables load forecasting model selection independent of the type of load forecasting problems. The meta- learning system built on 65 different load forecasting tasks returns forecasting error equal or lower than 10 well-known forecasting algorithms on 4 load forecasting tasks, for a recurrent real- life simulation. In the dissertation, an approach for definition and selection of metafeatures for problems involving load as a time-series have been proposed. New metafeatures of fickleness, traversity, granularity and highest ACF are introduced. The meta-learning framework is parallelized, component-based and easily extendable.

Izvorni jezik
Engleski

Znanstvena područja
Matematika, Elektrotehnika, Računarstvo



POVEZANOST RADA


Ustanove:
Fakultet elektrotehnike i računarstva, Zagreb

Profili:

Avatar Url Marin Matijaš (autor)

Avatar Url Slavko Krajcar (mentor)

Citiraj ovu publikaciju

Matijaš, Marin
Electric Load Forecasting using Multivariate Meta- learning, 2013., doktorska disertacija, Fakultet elektrotehnike i računarstva, Zagreb
Matijaš, M. (2013) 'Electric Load Forecasting using Multivariate Meta- learning', doktorska disertacija, Fakultet elektrotehnike i računarstva, Zagreb.
@phdthesis{phdthesis, author = {Matija\v{s}, M.}, year = {2013}, pages = {143}, keywords = {Artificial Neural Networks, Demand Forecasting, Electric Load, Electricity Consumption, Retail Electricity, Estimation, Gaussian Processes, General Load Forecaster, Least Squares Support Vector Machines, Meta-learning, Meta-features, Power, Prediction, Short-Term Load Forecasting, STLF, Support Vector Regression, SVM, Time-Series}, title = {Electric Load Forecasting using Multivariate Meta- learning}, keyword = {Artificial Neural Networks, Demand Forecasting, Electric Load, Electricity Consumption, Retail Electricity, Estimation, Gaussian Processes, General Load Forecaster, Least Squares Support Vector Machines, Meta-learning, Meta-features, Power, Prediction, Short-Term Load Forecasting, STLF, Support Vector Regression, SVM, Time-Series}, publisherplace = {Zagreb} }
@phdthesis{phdthesis, author = {Matija\v{s}, M.}, year = {2013}, pages = {143}, keywords = {Artificial Neural Networks, Demand Forecasting, Electric Load, Electricity Consumption, Retail Electricity, Estimation, Gaussian Processes, General Load Forecaster, Least Squares Support Vector Machines, Meta-learning, Meta-features, Power, Prediction, Short-Term Load Forecasting, STLF, Support Vector Regression, SVM, Time-Series}, title = {Electric Load Forecasting using Multivariate Meta- learning}, keyword = {Artificial Neural Networks, Demand Forecasting, Electric Load, Electricity Consumption, Retail Electricity, Estimation, Gaussian Processes, General Load Forecaster, Least Squares Support Vector Machines, Meta-learning, Meta-features, Power, Prediction, Short-Term Load Forecasting, STLF, Support Vector Regression, SVM, Time-Series}, publisherplace = {Zagreb} }




Contrast
Increase Font
Decrease Font
Dyslexic Font