Prediction of form roughness coefficient in alluvial channels using efficient hybrid approaches

Roushangar K., Saghebian S. M., Kırca V. Ş. Ö., Ghasempour R.

SOFT COMPUTING, cilt.24, sa.24, ss.18531-18543, 2020 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 24 Sayı: 24
  • Basım Tarihi: 2020
  • Doi Numarası: 10.1007/s00500-020-05090-5
  • Dergi Adı: SOFT COMPUTING
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Applied Science & Technology Source, Compendex, Computer & Applied Sciences, INSPEC, zbMATH
  • Sayfa Sayıları: ss.18531-18543
  • Anahtar Kelimeler: ANN, Dune bedform, Bedform resistance, MLP-FFA, SUPPORT VECTOR MACHINE, BED FORM, FIREFLY ALGORITHM, LOAD, SEDIMENT, MODEL
  • İstanbul Teknik Üniversitesi Adresli: Evet

Özet

In general, total roughness coefficient in open channels includes both grain resistance and bedform resistance. Due to the nonlinearity of the roughness coefficient, an accurate prediction of the bedform roughness is difficult. In this study, the capability of artificial neural network and multilayer perceptron (MLP) with firefly algorithm (MLP-FFA) were assessed in predicting the form resistance in channels with dune bedform. In this regard, different input combinations based on flow, bedform, and sediment characteristics were developed in order to determine the best combination. Five different experimental data series were applied to train and test the models. It was found that in predicting the form resistance, the model which took the advantages of both flow and sediment characteristics yielded to better outcomes. It was observed that the bedform characteristics led to an improvement in models accuracy. The results of the sensitivity analysis showed that the Reynolds number and the relative discharge were more effective parameters in the modeling process. Also, investigating the dune geometry (i.e., relative dune height) showed that the densimetric Froude number was the most significant variable.