Skip to the main content

Original scientific paper

https://doi.org/10.17559/TV-20160517081755

Predviđanje greške primjenom AR-LSSVR za elektrolitički kondenzator

Ming Yin ; College of Electronic Engineering, Naval University of Engineering, Jiefang Dadao No. 717, Qiaokou district, 430033 Wuhan, Hubei Province, China
Yanyi Xu ; College of Electronic Engineering, Naval University of Engineering, Jiefang Dadao No. 717, Qiaokou district, 430033 Wuhan, Hubei Province, China
Xiaohui Ye ; College of Electronic Engineering, Naval University of Engineering, Jiefang Dadao No. 717, Qiaokou district, 430033 Wuhan, Hubei Province, China
Shaochang Chen ; College of Electronic Engineering, Naval University of Engineering, Jiefang Dadao No. 717, Qiaokou district, 430033 Wuhan, Hubei Province, China
Hongxia Wang ; College of Electronic Engineering, Naval University of Engineering, Jiefang Dadao No. 717, Qiaokou district, 430033 Wuhan, Hubei Province, China
Feng Xie ; College of Electronic Engineering, Naval University of Engineering, Jiefang Dadao No. 717, Qiaokou district, 430033 Wuhan, Hubei Province, China


Full text: croatian pdf 989 Kb

page 783-789

downloads: 259

cite

Full text: english pdf 989 Kb

page 783-789

downloads: 306

cite


Abstract

U radu se opisuje metoda predviđanja greške na osnovu Autoregressive - Support Vector Regression Metode (AR-LSSVR) za elektrolitički kondenzator. Budući da je elektrolitički kondenzator jeftin, a velik, uveliko se primjenjuje u elektroničkim krugovima. Najprije se daje osnovni model i algoritam za predviđanje greške za AR, LSSVM i AR-LSSVR. Model AR-LSSVR kombinira prednosti algoritma LSSVR-a i modela AR te ih dopunjuje kako bi se povećala točnost predviđanja. Daje se dijagram toka predviđanja pojave greške na temelju AR-LSSVR. Konačno se AR-LSSVR model primjenjuje na Buck strujni krug. Rezultati pokazuju da je predviđanje greške elektrolitičkog kondenzatora bolje primjenom modela AR-LSSVR.

Keywords

AR-LSSVR, elektrolitički kondenzator, predviđanje greške

Hrčak ID:

183038

URI

https://hrcak.srce.hr/183038

Article data in other languages: english

Visits: 959 *