An Optimized Recurrent Neural Network for Metocean Forecasting

• Main Authors: Alqushaibi, A., Abdulkadir, S.J., Rais, H.M., Al-Tashi, Q., Ragab, M.G.
• Format: Conference or Workshop Item
• Published: Institute of Electrical and Electronics Engineers Inc. 2020
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85097537252&doi=10.1109%2fICCI51257.2020.9247681&partnerID=40&md5=5e47ef54d0f84273a9527a3fe3a0db9c; http://eprints.utp.edu.my/29858/

Metocean data plays a crucial role in planning and constructing offshore projects. the success of many offshore projects depends on the accuracy of metocean data analyzing and forecasting. And analyzing metocean data requires a tremendous effort to validate the data and determine the transformation of the metocean data conditions. Hence the wind plays an important role in the climate changes, recurrent neural network approaches such as vanilla recurrent neural network (VRNN), long short-term memory (LSTM), and Gated recurrent units (GRU) are used and compared to yield an accurate wind speed forecasting. The highest wind speed forecasting accuracy contribute to the minimization of cost and helps avoiding the operational faulty risk. Different models for estimating the hourly wind speed one hour ahead and one day ahead has been developed according to literature. However, this research compares the mentioned Artificial Neural Networks and selects the outstanding performance model to process the metocean data. The training and validation data of this work has been collected from free oceanic websites. © 2020 IEEE.