Chapter 25 - Hybrid extreme learning machine optimized bat algorithm based on ensemble empirical mode decomposition for modeling dissolved oxygen in river

Abstract

In the present chapter, we use the empirical mode decomposition (EMD), the ensemble EMD (EEMD), and the complete ensemble EMD with adaptive noise (CEEMDAN) for dissolved oxygen (DO) prediction. First, based on water temperature ( T w ), DO was modeled using three machines learning models, namely, extreme learning machine (ELM), the ELM optimized Bat algorithm (Bat-ELM), and relevance vector machine (RVM). Second, river T w was decomposed using EMD, EEMD, and CEEMDAN into several intrinsic mode functions (IMF), which were used as input to the ELM, Bat-ELM, and RVM. The performances of the models were evaluated using the root mean square error (RMSE), mean absolute error (MAE), coefficient of correlation ( R ), and the Nash-Sutcliffe efficiency (NSE). From the obtained results, the models based on EMD, EEMD, and CEEMDAN estimated DO highly more accurate than the single models, with mean RMSE, MAE, R , and NSE of 0.835°C, 0.571°C, 0.965, and 0.930 against the values of 2.788°C, 2.232°C, 0.511, and 0.250, respectively.