Prediction of daily average seawater temperature using data-driven and deep learning algorithms

Large amounts of cooling water are required for the cooling process in coal-fired or nuclear power plants. Because the primary use of water in a power plant is to condense steam and remove waste heat as part of a Rankine cycle, seawater temperature (SWT) is of critical importance for electrical output in power plant applications installed at the seaside. Moreover, the analysis of SWT is an important criterion for researching sea life and global climate, and it also serves as an important indicator of climate change. In this paper, multilayer perceptron, which is in the class of a feed-forward artificial neural network, deep learning approach based on long short-term memory and bidirectional long short-term memory neural networks and data-driven methods, such as adaptive neuro-fuzzy inference system (ANFIS) accompanied by fuzzy c-means, ANFIS with grid partition and ANFIS with subtractive clustering methods were applied to make 1-day ahead SWT predictions. Analyses were conducted using 5-year daily mean SWTs measured by the Turkish State Meteorological Service for Antalya province between 2014 and 2018. The models were evaluated using mean absolute error (MAE), root-mean-square error (RMSE), and correlation coefficient (R). According to the daily SWT prediction, the best MAE, RMSE, and R values were obtained with the ANFIS-SC model, which were 0.1877 degrees C, 0.2683 degrees C, and 0.99814, respectively.