Prediction of synchronous closing time of permanent magnetic actuator for vacuum circuit breaker based on PSO-BP

One of the key points of phase-controlled technology is to predict the circuit breaker's operation time. In this paper, it took control voltage and ambient temperature as the main input variables to forecast the closing time through building backward propagation (BP) neural network model. In vi...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE transactions on dielectrics and electrical insulation Ročník 24; číslo 6; s. 3321 - 3326
Hlavní autoři: Hou, Chunguang, Yu, Xiao, Cao, Yundong, Lai, Changxue, Cao, Yuchen
Médium: Journal Article
Jazyk:angličtina
Vydáno: New York IEEE 01.12.2017
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Témata:
Accuracy
Actuator
Actuators
Ambient temperature
Circuit breakers
closing time
Magnetic circuits
Mathematical models
Model accuracy
Neural networks
Particle swarm optimization
vacuum circuit breaker
ISSN:1070-9878, 1558-4135
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:One of the key points of phase-controlled technology is to predict the circuit breaker's operation time. In this paper, it took control voltage and ambient temperature as the main input variables to forecast the closing time through building backward propagation (BP) neural network model. In view of the high accuracy requirements for the closing time's prediction, it is difficult to meet the requirement by relying solely on the BP neural network. Basing on BP neural network, it uses particle swarm optimization (PSO) algorithm to optimize the model for improving the prediction accuracy. Through calculation and analysis, the PSO-BP algorithm is more accurate than BP neural network in the prediction accuracy, which controls the error within 0.2%. The prediction error meets the requirements of synchronous controlled technology.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1070-9878
1558-4135
DOI:10.1109/TDEI.2017.006475