检索结果-南通市图书馆

Pr.tection of Low Voltage CIGr. Distr.bution Networ.

在线全文

学校读者我要写书评

暂无评论

Smar. Gr.d and r.newable Ener.y 2013年第9期4卷 489-500页

作者： Ghullam Mustafa Bhutto Bir.itte Bak-Jensen Claus Leth Bak jayakrishnan r. pillai Depar.ment of Ener.y Technology Aalborg University Aalborg Denmark Depar.ment of Ener.y Technology Aalborg University Aalborg Denmark Depar.ment of Electr.cal power Quaid-E-Awam University of Sciences & Technology Nawabshah Pakistan

High quality electr.city ser.ices ar. the pr.me objectives in the moder. power.systems ar.und the wor.d. One of the main player. to achieve this is the pr.tection of the system which needs to be fast, r.liable and cost effective. The study about the pr.tection of the Low Voltage (LV) CIGr. distr.bution gr.d and networ.s like this has been pr.posed in this paper. The main objective of this paper.is to develop pr.tection against shor. cir.uit faults which might appear.anywher. in the networ.. The pr.tection of the power.networ.s that compr.ses of r.newable ener.y gener.tion units is complicated because of the bidir.ctional flow of the cur.ent and is a challenge for.the pr.tection engineer.. The selection of the pr.tection devices in this paper.is made to pr.tect the networ. against faults in gr.d connected and island mode of oper.tion. Ultr.-fast fuses ar. pr.posed in or.er.to pr.tect the inver.er. used for.Photovoltaic (PV) and batter. applications. The disconnection of the PV solar.panels when in island mode is made by pr.posing switch disconnecting devices. ABB is cur.ently using these kinds of disconnection devices for.the pur.ose of pr.tecting solar.panels against over.voltages in the case of islanding. The over.speed pr.tection of the existing Wind Tur.ine Gener.tor.(WTG) in the CIGr. networ. in case of gr.d loss is also pr.posed in this paper.

关键词： Cur.ent Based Pr.tection Voltage Based Pr.tection Over Speed Pr.tection Fuse Under Voltage r.lay Cir.uit Br.aker Switch Disconnector

r.bust Voltage Contr.l Consider.ng Uncer.ainties of r.newable Ener.ies and Loads via Impr.ved Gener.tive Adver.ar.al Networ.

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Jour.al of Moder. Power.Systems and Clean Ener.y 2020年第6期8卷 1104-1114页

作者： Qianyu Zhao Wenlong Liao Shouxiang Wang jayakrishnan r.dhakr.shna pillai the Key Labor.tor. of Smar. Gr.d of Ministr. of Education and Tianjin Key Labor.tor. of Power.System Simulation and Contr.l Tianjin UniversityTianjinChina the Depar.ment of Ener.y Technology Aalborg UniversityAalborgDenmark

The fluctuation of output power.of r.newable ener.ies and loads br.ngs challenges to the scheduling and oper.tion of the distr.bution *** this paper.a r.bust voltage contr.l model is pr.posed to cope with the uncer.ainties of r.newable ener.ies and loads based on an impr.ved gener.tive adver.ar.al networ.(IGAN).Fir.tly,both r.al and pr.dicted data ar. used to tr.in the IGAN consisting of a discr.minator.and a *** noises sampled fr.m the Gaussian distr.bution ar. fed to the gener.tor.to gener.te a lar.e number.of scenar.os that ar. utilized for.r.bust voltage contr.l after.scenar.o ***,a new impr.ved wolf pack algor.thm(IWPA)is pr.sented to solve the for.ulated r.bust voltage contr.l model,since the accur.cy of the solutions obtained by tr.ditional methods is *** simulation r.sults show that the IGAN can accur.tely captur. the pr.bability distr.bution char.cter.stics and dynamic nonlinear.char.cter.stics of r.newable ener.ies and loads,which makes the scenar.os gener.ted by IGAN mor. suitable for.r.bust voltage contr.l than those gener.ted by tr.ditional ***,IWPA has a better.per.or.ance than tr.ditional methods in ter.s of conver.ence speed,accur.cy,and stability for.r.bust voltage contr.l.

关键词： r.bust voltage contr.l uncer.ainty gener.tive adver.ar.al networ. wolf pack algor.thm

A r.view of Gr.ph Neur.l Networ.s and Their.Applications in Power.Systems

在线全文

学校读者我要写书评

暂无评论

Jour.al of Moder. Power.Systems and Clean Ener.y 2022年第2期10卷 345-360页

作者： Wenlong Liao Bir.itte Bak-Jensen jayakrishnan r.dhakr.shna pillai Yuelong Wang Yusen Wang the Depar.ment of Ener.y Technology Aalborg UniversityAalborgDenmark the State Gr.d Tianjin Chengxi Electr.c Power.Supply Br.nch TianjinChina the School of Electr.cal Engineer.ng and Computer.Science KTH Royal Institute of TechnologyStockholmSweden

Deep neur.l networ.s have r.volutionized many machine lear.ing tasks in power.systems,r.nging fr.m patter. r.cognition to signal *** data in these tasks ar. typically r.pr.sented in Euclidean ***,ther. is an incr.asing number.of applications in power.systems,wher. data ar. collected fr.m non-Euclidean domains and r.pr.sented as gr.ph-str.ctur.d data with high-dimensional featur.s and inter.ependency among *** complexity of gr.ph-str.ctur.d data has br.ught significant challenges to the existing deep neur.l networ.s defined in Euclidean ***,many publications gener.lizing deep neur.l networ.s for.gr.ph-str.ctur.d data in power.systems have *** this paper.a compr.hensive over.iew of gr.ph neur.l networ.s(GNNs)in power.systems is ***,sever.l classical par.digms of GNN str.ctur.s,e.g.,gr.ph convolutional networ.s,ar. *** applications in power.systems such as fault scenar.o application,time-ser.es pr.diction,power.flow calculation,and data gener.tion ar. r.viewed in ***,main issues and some r.sear.h tr.nds about the applications of GNNs in power.systems ar. discussed.

关键词： Machine lear.ing power.system deep neur.l networ. gr.ph neur.l networ. ar.ificial intelligence

Data-dr.ven Missing Data Imputation for.Wind Far.s Using Context Encoder

在线全文

学校读者我要写书评

暂无评论

Jour.al of Moder. Power.Systems and Clean Ener.y 2022年第4期10卷 964-976页

作者： Wenlong Liao Bir.itte Bak-Jensen jayakrishnan r.dhakr.shna pillai Dechang Yang Yusen Wang Depar.ment of Ener.y Technology Aalborg UniversityAalborgDenmark College of Infor.ation and Electr.cal Engineer.ng China Agricultural UniversityBeijingChina School of Electr.cal Engineer.ng and Computer.Science KTH Royal Institute of TechnologyStockholmSweden

High-quality datasets ar. of par.mount impor.ance for.the oper.tion and planning of wind ***,the datasets collected by the super.isor. contr.l and data acquisition(SCADA)system may contain missing data due to var.ous factor. such as sensor.failur. and communication *** this paper.a data-dr.ven appr.ach is pr.posed to fill the missing data of wind far.s based on a context encoder.CE),which consists of an encoder.a decoder.and a *** deep convolutional neur.l networ.s,the pr.posed method is able to automatically explor. the complex nonlinear.char.cter.stics of the datasets that ar. difficult to be modeled *** pr.posed method can not only fully use the sur.ounding context infor.ation by the r.constr.cted loss,but also make filling data look r.al by the adver.ar.al *** addition,the cor.elation among multiple missing attr.butes is taken into account by adjusting the for.at of input *** simulation r.sults show that CE per.or.s better.than tr.ditional methods for.the attr.butes of wind far.s with hallmar. char.cter.stics such as lar.e peaks,lar.e valleys,and fast ***,the CE shows str.nger.gener.lization ability than tr.ditional methods such as auto-encoder.K-means,k-near.st neighbor.back pr.pagation neur.l networ.,cubic inter.olation,and conditional gener.tive adver.ar.al networ. for.differ.nt missing data scales.

关键词： Data-dr.ven missing data imputation wind far. deep lear.ing context encoder

Scenar.o Gener.tions for.r.newable Ener.y Sour.es and Loads Based on Implicit Maximum Likelihood Estimations

在线全文

学校读者我要写书评

暂无评论

Jour.al of Moder. Power.Systems and Clean Ener.y 2022年第6期10卷 1563-1575页

作者： Wenlong Liao Bir.itte Bak-Jensen jayakrishnan r.dhakr.shna pillai Zhe Yang Yusen Wang Kuangpu Liu AAU Ener.y Aalborg UniversityAalborgDenmark School of Electr.cal Engineer.ng and Computer.Science KTH Royal Institute of TechnologyStockholmSweden

Scenar.o gener.tions for.r.newable ener.y sour.es and loads play an impor.ant r.le in the stable oper.tion and r.sk assessment of integr.ted ener.y *** paper.pr.poses a deep gener.tive networ. based method to model time-ser.es cur.es,e.g.,power.gener.tion cur.es and load cur.es,of r.newable ener.y sour.es and loads based on implicit maximum likelihood estimations(IMLEs),which can gener.te r.alistic scenar.os with similar.patter.s as r.al *** tr.ining the model,any number.of new scenar.os can be obtained by simply inputting Gaussian noises into the data gener.tor.of *** pr.posed appr.ach does not r.quir. any model assumptions or.pr.or.knowledge of the for. in the likelihood function being made dur.ng the tr.ining pr.cess,which leads to str.nger.applicability than explicit density model based *** extensive exper.ments show that the IMLEs accur.tely captur. the complex shapes,fr.quency-domain char.cter.stics,pr.bability distr.butions,and cor.elations of r.newable ener.y sour.es and ***,the pr.posed appr.ach can be easily gener.lized to scenar.o gener.tion tasks of var.ous r.newable ener.y sour.es and loads by fine-tuning par.meter. and str.ctur.s.

关键词： r.newable ener.y sour.e scenar.o gener.tion implicit maximum likelihood estimation(IMLE) deep lear.ing gener.tive networ.

Ultr.-shor.-ter. Inter.al Pr.diction of Wind Power.Based on Gr.ph Neur.l Networ. and Impr.ved Bootstr.p Technique

在线全文

学校读者我要写书评

暂无评论

Jour.al of Moder. Power.Systems and Clean Ener.y 2023年第4期11卷 1100-1114页

作者： Wenlong Liao Shouxiang Wang Bir.itte Bak-Jensen jayakrishnan r.dhakr.shna pillai Zhe Yang Kuangpu Liu AAU Ener.y Aalborg UniversityAalborg 9220Denmark Key Labor.tor. of Smar. Gr.d of Ministr. of Education School of Electrical and Information EngineeringTianjin UniversityTianjin 300072China

r.liable and accur.te ultr.-shor.-ter. pr.diction of wind power.is vital for.the oper.tion and optimization of power.***,the volatility and inter.ittence of wind power.pose uncer.ainties to tr.ditional point pr.diction,r.sulting in an incr.ased r.sk of power.system *** r.pr.sent the uncer.ainty of wind power.this paper.pr.poses a new method for.ultr.-shor.-ter. inter.al pr.diction of wind power.based on a gr.ph neur.l networ.(GNN)and an impr.ved Bootstr.p ***,adjacent wind far.s and local meteor.logical factor. ar. modeled as the new for. of a gr.ph fr.m the gr.ph-theor.tic ***,the gr.ph convolutional networ.(GCN)and bi-dir.ctional long shor.-ter. memor.(Bi-LSTM)ar. pr.posed to captur. spatiotempor.l featur.s between nodes in the *** obtain highquality pr.diction inter.als(PIs),an impr.ved Bootstr.p technique is designed to incr.ase cover.ge per.entage and nar.ow PIs *** simulations demonstr.te that the pr.posed method can captur. the spatiotempor.l cor.elations fr.m the gr.ph,and the pr.diction r.sults outper.or. popular.baselines on two r.al-wor.d datasets,which implies a high potential for.pr.ctical applications in power.systems.

关键词： Wind power gr.ph neur.l networ.(GNN) bidir.ctional long shor.-ter. memor.(Bi-LSTM) pr.diction inter.al Bootstr.p technique