@inproceedings{scholars17261,
           title = {Forecasting of Wind Turbines Generated Power with Missing Input Variables},
       publisher = {Institute of Electrical and Electronics Engineers Inc.},
         journal = {2022 International Conference on Future Trends in Smart Communities, ICFTSC 2022},
           pages = {98--103},
            note = {cited By 0; Conference of 2022 International Conference on Future Trends in Smart Communities, ICFTSC 2022 ; Conference Date: 1 December 2022 Through 2 December 2022; Conference Code:186671},
             doi = {10.1109/ICFTSC57269.2022.10039884},
            year = {2022},
          author = {Sunder, M. and Abishek, R. and Maiti, M. and Bingi, K. and Devan, P. A. M. and Assaad, M.},
            isbn = {9798350334548},
        keywords = {Atmospheric pressure; Brain; Feedforward neural networks; Long short-term memory; Wind; Wind turbines, Electric wind; Feed forward neural net works; Input parameter; Input variables; Missing inputs; Missing parameter; Neural-networks; Power; Power industry; Wind speed, Forecasting},
             url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85149110412&doi=10.1109\%2fICFTSC57269.2022.10039884&partnerID=40&md5=02a6828ad78ac6f42f38ef6f7f17d770},
        abstract = {The power generated by electric wind turbines undergoes rapid changes due to continuous fluctuation of wind speed, direction, atmospheric pressure, etc. Providing the power industry with the capability to estimate these performance characteristics helps in the pre-planning of maintenance, which helps in power management by assessing the generated power for the day. However, forecasting the generated power with any missing input parameters is quite challenging. Therefore, this paper proposes a forecasting model with three types of neural networks to handle one missing input parameter to predict the wind turbine's generated power. Firstly, a Feed Forward Neural Network (FFNN) is developed to forecast generated power from all four available input parameters. Later the FFNN, along with a Long Short-Term Memory (LSTM) and Nonlinear Autoregressive (NAR) neural networks, are modeled to handle the missing input parameter. The main FFNN then uses the predicted parameter to forecast the generated power. The results from the simulation study have indicated that the proposed strategy achieved the best performance in predicting the missing input and the system's generated power. {\^A}{\copyright} 2022 IEEE.}
}