Mindaugas Jankauskas, A. Serackis, Raimondas Pomornacki, Van Khang Hyunh, Martynas Šapurov, A. Baskys
{"title":"Short-term Wind Energy Forecasting with Advanced Recurrent Neural Network Models: A Comparative Study","authors":"Mindaugas Jankauskas, A. Serackis, Raimondas Pomornacki, Van Khang Hyunh, Martynas Šapurov, A. Baskys","doi":"10.1109/AIEEE58915.2023.10134882","DOIUrl":null,"url":null,"abstract":"For effective energy management and the integration of renewable energy sources into the grid, accurate wind power forecasting is crucial. The objective of this work is to construct and assess a variety of machine learning models to forecast the generation of electricity from wind farms using meteorological data, such as wind speed, wind direction, air temperature, humidity, and atmospheric pressure. We tested six models, of which three were based on a recurrent neural network approach. The most pertinent elements for precise predictions were found using feature selection algorithms after preprocessing the meteorological and wind farm power generating data. The results of the study indicate that the best option to predict the generation of electricity from wind farms using meteorological data is the model based on bidirectional LSTM. The study also emphasizes how crucial feature selection and hyperparameter adjustment are for enhancing model performance. Future work may investigate different machine learning models, include more features or data sources, and evaluate how well the models can be used in energy management systems in the real world.","PeriodicalId":149255,"journal":{"name":"2023 IEEE 10th Jubilee Workshop on Advances in Information, Electronic and Electrical Engineering (AIEEE)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2023-04-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2023 IEEE 10th Jubilee Workshop on Advances in Information, Electronic and Electrical Engineering (AIEEE)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/AIEEE58915.2023.10134882","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
For effective energy management and the integration of renewable energy sources into the grid, accurate wind power forecasting is crucial. The objective of this work is to construct and assess a variety of machine learning models to forecast the generation of electricity from wind farms using meteorological data, such as wind speed, wind direction, air temperature, humidity, and atmospheric pressure. We tested six models, of which three were based on a recurrent neural network approach. The most pertinent elements for precise predictions were found using feature selection algorithms after preprocessing the meteorological and wind farm power generating data. The results of the study indicate that the best option to predict the generation of electricity from wind farms using meteorological data is the model based on bidirectional LSTM. The study also emphasizes how crucial feature selection and hyperparameter adjustment are for enhancing model performance. Future work may investigate different machine learning models, include more features or data sources, and evaluate how well the models can be used in energy management systems in the real world.