{"title":"Neural Network-Based Models for Prediction of Smart Grid Stability","authors":"Kishore Bingi, B. Prusty","doi":"10.1109/i-PACT52855.2021.9696517","DOIUrl":null,"url":null,"abstract":"This paper focuses on proposing neural network-based prediction models for smart grid stability. The models are trained using the damped least-squares technique with reaction time, consumed/generated power, and elasticity coefficient as input variables to predict the stability as an output variable. The models' accuracy and effectiveness are compared numerically using R2 and mean square errors. For all the models, the selected activation functions at the hidden layer is tansig, and purelin at the output layer. The results demonstrated the neural network-based prediction models' adequate performance for training and testing phases. For the system under consideration, the feed-forward neural network's best performance is true in terms of least error and the highest R2 values.","PeriodicalId":335956,"journal":{"name":"2021 Innovations in Power and Advanced Computing Technologies (i-PACT)","volume":"87 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2021-11-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"4","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2021 Innovations in Power and Advanced Computing Technologies (i-PACT)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/i-PACT52855.2021.9696517","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 4
Abstract
This paper focuses on proposing neural network-based prediction models for smart grid stability. The models are trained using the damped least-squares technique with reaction time, consumed/generated power, and elasticity coefficient as input variables to predict the stability as an output variable. The models' accuracy and effectiveness are compared numerically using R2 and mean square errors. For all the models, the selected activation functions at the hidden layer is tansig, and purelin at the output layer. The results demonstrated the neural network-based prediction models' adequate performance for training and testing phases. For the system under consideration, the feed-forward neural network's best performance is true in terms of least error and the highest R2 values.