Evolutionary product unit based neural networks for hydrological time series analysis

Abstract

Artificial Neural Networks (ANNs) are now widely used in many areas of science, medicine, finance and engineering. Analysis and prediction of time series of hydrological/and meteorological data is one such application. Problems that still exist in the application of ANN's are the lack of transparency and the expertise needed for training. An evolutionary algorithm-based method to train a type of neural networks called Product Units Based Neural Networks (PUNN) has been proposed in a 2006 study. This study investigates the applicability of this type of neural networks to hydrological time series prediction. The technique, with a few small changes to improve the performance, is applied to some benchmark time series as well as to a real hydrological time series for prediction. The results show that evolutionary PUNN produce more transparent models compared to widely used multilayer perceptron (MLP) neural network models. It is also seen that training of PUNN models requires less expertise compared to MLPs.

Time series forecasting by recurrent product unit neural networks

“…This handicap makes convenient the use of global search algorithms, such as genetic algorithms [17,18], evolutionary algorithms [19] or swarm optimisation algorithms [20], in order to find the parameters minimising the error function. PUNNs have been widely used in classification [21] and regression problems [22], but scarcely applied to TSF, with the exception of some attempts on hydrological TSA [23,24]. It is important to point out that, in TSF, there is an autocorrelation between the lagged values of the series.…”

Hourly runoff forecasting for flood risk management: Application of various computational intelligence models

“…Since then, there has been an proliferation of related research including the use of radial basis function type artificial neural networks Jayawardena et al, 2006), evolutionary product unit based neural networks for hydrological time series analysis (Karunasingha et al, 2011), water level prediction using artificial neural networks (Biswas and Jayawardena, 2014), and, river flow forecasting (Tawfik et al, 1997;Abrahart and See, 2000;Imrie et al, 2000;Birikundavyi et al, 2002;Cigizoglu, 2003;Moradkhani et al, 2004;Machado et al, 2011), among others. During the last decade or so, fuzzy logic approach has been used in hydrological applications (e.g.…”