Neural-Based Ensembles and Unorganized Machines to Predict Streamflow Series from Hydroelectric Plants

Abstract

Estimating future streamflows is a key step in producing electricity for countries with hydroelectric plants. Accurate predictions are particularly important due to environmental and economic impact they lead. In order to analyze the forecasting capability of models regarding monthly seasonal streamflow series, we realized an extensive investigation considering: six versions of unorganized machines—extreme learning machines (ELM) with and without regularization coefficient (RC), and echo state network (ESN) using the reservoirs from Jaeger’s and Ozturk et al., with and without RC. Additionally, we addressed the ELM as the combiner of a neural-based ensemble, an investigation not yet accomplished in such context. A comparative analysis was performed utilizing two linear approaches (autoregressive model (AR) and autoregressive and moving average model (ARMA)), four artificial neural networks (multilayer perceptron, radial basis function, Elman network, and Jordan network), and four ensembles. The tests were conducted at five hydroelectric plants, using horizons of 1, 3, 6, and 12 steps ahead. The results indicated that the unorganized machines and the ELM ensembles performed better than the linear models in all simulations. Moreover, the errors showed that the unorganized machines and the ELM-based ensembles reached the best general performances.

Previsão do Preço do Açúcar: Uma Análise Comparativa entre Modelo Linear e Redes Neurais

“…Por fim, destacamos que para os resultados constantes nas Tabelas II e III aplicou-se o teste de Friedman [38] para averiguarmos se os resultados são significativamente diferentes. Os p-valores encontrados foram da ordem de 10 −6 , o que indica que modificar o preditor leva a resultados distintos [29]. Para trabalhos futuros, recomenda-se o desenvolvimento de um modelo híbrido para correc ¸ão de erro.…”

“…O treinamento se resume a encontrar a soluc ¸ão de um problema de mínimos quadrados, uma tarefa com soluc ¸ão determinística e baixo custo computacional. A forma mais tradicional é lanc ¸ar mão do algoritmo da pseudo-inversa de Moore-Penrose [29].…”

Comparative study of forecasting approaches in monthly streamflow series from Brazilian hydroelectric plants using Extreme Learning Machines and Box & Jenkins models

“…In the same way, when comparing the performances by different error metrics, it can be noticed that, in some cases, the best MSE is different from the smallest MAE, NSE, or CC. As the function minimized in the optimization and calculation of the coefficients is based on the MSE, we adopted this metric in the real domain to determine the best performances (Belotti et al, 2020a;Siqueira et al, 2014).…”

“…In this manner, for each of the 12 months of the year, an AR model will be adjusted independently. For this, the terms of the Yule-Walker equations are rewritten for the periodic case in matrix form following Equation ( 8) (Belotti et al, 2020a):…”

“…The prediction of streamflow series can be considered a nonlinear static mapping task (Belotti et al, 2020a;Siqueira et al 2012b;Yaseen et al, 2017). The Multilayer Perceptron (MLP) is an architecture of ANN widely used in tackling this problem (Siqueira et al, 2014).…”