Predicting temporal rate coefficient of bar volume using hybrid artificial intelligence approaches


KANKAL M., UZLU E., NACAR S., YÜKSEK Ö.

JOURNAL OF MARINE SCIENCE AND TECHNOLOGY, cilt.23, sa.3, ss.596-604, 2018 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 23 Sayı: 3
  • Basım Tarihi: 2018
  • Doi Numarası: 10.1007/s00773-017-0495-1
  • Dergi Adı: JOURNAL OF MARINE SCIENCE AND TECHNOLOGY
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Sayfa Sayıları: ss.596-604
  • Anahtar Kelimeler: Bar volume, Neural networks, Sediment transport, Teaching-learning-based optimization, Artificial bee colony, RUBBLE-MOUND BREAKWATERS, FUZZY NEURAL-NETWORKS, SHORE SANDBAR VOLUMES, GEOMETRIC CHARACTERISTICS, GENETIC ALGORITHMS, COASTAL EROSION, IRREGULAR WAVES, PROFILES, DESIGN
  • Karadeniz Teknik Üniversitesi Adresli: Evet

Özet

To project the structures to be built in the coastal zone and to make the best use of the coastal area, the mechanism of sediment transport, including both longshore and cross-shore transport, in this region should be well known. Within this context, temporal change rate of cross-shore sediment transport is of vital importance, especially to predict the erosion quantitatively. In this study, hybrid artificial intelligence models based on physical model data were established to determine the alpha coefficient used to describe the temporal change of cross-shore sediment transport. Teaching-learning-based optimization (TLBO) and artificial bee colony (ABC) algorithms were used for training of artificial neural network (ANN) in the model setup. Then, these models were compared with the classical back propagation ANN (ANN-BP) model. Wave height and period, bed slope and sediment diameter were considered as input parameters in the models. In all models, the used data for training and testing sets were 42 and 10 of total 52 experimental data, respectively. In the end of the analyses, it has been determined that the ANN-TLBO and ANN-ABC models have resulted in better results than the BP models. Also, the smallest mean absolute error and root mean square error values for testing set have been obtained from the ANN-TLBO model with 0.0068 and 0.0081, respectively. Therefore, it has been concluded that the best model ANN-TLBO can be successfully applied to predict the alpha coefficient.