The Performance Analysis of Robust Local Mean Mode Decomposition Method for Forecasting of Hydrological Time Series


LATİFOĞLU L.

Iranian Journal of Science and Technology - Transactions of Civil Engineering, cilt.46, sa.4, ss.3453-3472, 2022 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 46 Sayı: 4
  • Basım Tarihi: 2022
  • Doi Numarası: 10.1007/s40996-021-00809-2
  • Dergi Adı: Iranian Journal of Science and Technology - Transactions of Civil Engineering
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, ABI/INFORM, Aerospace Database, CAB Abstracts, INSPEC, Civil Engineering Abstracts
  • Sayfa Sayıları: ss.3453-3472
  • Anahtar Kelimeler: Forecasting, Streamflow data, Robust Local Mean Decomposition (RLMD), Artificial Neural Networks (ANN), Support Vector Regression (SVR), Long Short-Term Memory Networks (LSTM), Empirical Mode Decomposition (EMD), SINGULAR SPECTRUM ANALYSIS, ARTIFICIAL-INTELLIGENCE, STREAMFLOW, WAVELET, TRANSFORM, ANN
  • Kayseri Üniversitesi Adresli: Hayır

Özet

© 2022, Shiraz University.Accurate forecasting of streamflow data over daily timescales is a critical problem for the long-term management of water resources, agricultural uses, and many more purposes. This study proposes a new hybrid approach that combines the Robust Local Mean Decomposition (RLMD) method and the Artificial Neural Network (ANN) method for the prediction of streamflow data. Monthly streamflow data were split into the training and the testing part firstly and threefold cross-validation was performed to obtain a more reliable model. After building of the model using the training data, the proposed model was tested on the testing data. Also to compare the performance of the RLMD–ANN model and the Support Vector Regression (SVR) model, Long Short-Term Memory Networks (LSTM) were used for forecasting of subband signal and streamflow data. Also, the hybrid Empirical Mode Decomposition (EMD) model and hybrid Autoregressive Integrated Moving Average (ARIMA) model were used for comparison of the proposed model. Therefore, the RLMD–ANN model was compared with RLMD–SVR, RLMD–LSTM, EMD–ANN, Additive–ARIMA–ANN, ANN, SVR, and LSTM models. The numerical results of the study were assessed concerning the Mean Square Error (MSE), Mean Absolute Error (MAE), Determination Coefficient (R2), Correlation Coefficient (R), and Kruskal–Wallis test was used to indicate whether the results are statically significant. One- to three-ahead forecast and one–two inputs were applied to the models. The mean one-ahead forecasting performance of the three folds was calculated for two inputs with MSE, MAE, R2, and R parameters as 0.0060, 0.0522, 0.7342, and 0.8532 respectively. The obtained results show that the novel RLMD–ANN model is a reliable, efficient, and high-performant model for forecasting streamflow data.