Autors: Pavlatos, C., Makris, E., Fotis, G., Vita, V., Mladenov, V. M. Title: Enhancing Electrical Load Prediction Using a Bidirectional LSTM Neural Network Keywords: forecastingelectricitydemand;bidirectionalLSTM;short-termpre Abstract: Preciseanticipationofelectricaldemandholdscrucialimportancefortheoptimaloperation ofpowersystemsandtheeffectivemanagementofenergymarketswithinthedomainofenergy planning. Thisstudybuildsonpreviousresearchfocusedontheapplicationofartificialneural networkstoachieveaccurateelectricalloadforecasting. Inthispaper,animprovedmethodology isintroduced,centeringaroundbidirectionalLongShort-TermMemory(LSTM)neuralnetworks (NN).TheprimaryaimoftheproposedbidirectionalLSTMnetworkistoenhancepredictiveperformancebycapturingintricatetemporalpatternsandinterdependencieswithintimeseriesdata. Whileconventionalfeed-forwardneuralnetworksaresuitableforstandalonedatapoints,energy consumptiondataarecharacterizedbysequentialdependencies,necessitatingtheincorporation ofmemory-basedconcepts. ThebidirectionalLSTMmodel isdesignedtofurnishtheprediction frameworkwiththecapacitytoassimilateandleverageinformationfrombothprecedingandforthcomingtimesteps.Thisaugmentationsignificantlybolsterspredictivecapabilitiesbyencapsu References Issue
|
Вид: статия в списание, публикация в издание с импакт фактор, публикация в реферирано издание, индексирана в Scopus и Web of Science