RBFNN versus FFNN for daily river flow forecasting at Johor River, Malaysia

Abstract

Streamflow forecasting can have a significant economic impact, as this can help in water resources management and in providing protection from water scarcities and possible flood damage. Artificial neural network (ANN) had been successfully used as a tool to model various nonlinear relations, and the method is appropriate for modeling the complex nature of hydrological systems. They are relatively fast and flexible and are able to extract the relation between the inputs and outputs of a process without knowledge of the underlying physics. In this study, two types of ANN, namely feed-forward back-propagation neural network (FFNN) and radial basis function neural network (RBFNN), have been examined. Those models were developed for daily streamflow forecasting at Johor River, Malaysia, for the period (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008). Comprehensive comparison analyses were carried out to evaluate the performance of the proposed static neural networks. The results demonstrate that RBFNN model is superior to the FFNN forecasting model, and RBFNN can be successfully applied and provides high accuracy and reliability for daily streamflow forecasting.

A polynomial neural network for river’s water-level prediction

“…Nevertheless, despite of the low values of the coefficient of determination (R 2 ), having a system that approximates the water level of a river's flow with a mean error of 1 cm, is considered as a fair enough approximator for the purposes of this document. While in many studies, such as, e.g., in [12,14,15], only the past time series are used in order to forecast the water level of rivers, and some others use the weather data of the past days together with the time series (e.g., in [16]), the present study follows a completely different approach. The present study uses data provided by sensors installed on areas of the river, together with weather data which can be provided in real-time by national organizations; this methodology was followed because the authors believe that the rainfall in areas near the river is a parameter easy to retrieve, is of crucial importance on the resulting water level, but other studies usually do not use it.…”

“…Zhu et al [13] model the daily temperature of a river using an ANN and a neuro-fuzzy system using as input parameters the air temperature, the river flow discharge and the calendar components (the day, the month and the year). Yaseen et al [14] compare feed-forward neural network (FFNN) and radial basis function neural networks (RBFNN) to forecast the daily streamflow of river; an input of the ANN's the use the river's streamflow of previous days. Kasiviswanathan et al [15] use ANN's and wavelet-based neural networks in order to forecast a river's streamflow and predict floods based on streamflow data of previous days.…”

Prediction of the Seasonal Changes of the Chloride Concentrations in Urban Water Reservoir

“…They are connected by the so-called synapses which are related with appropriate weighting factors. The most commonly used network model is the three-layer ANN model, in which we distinguish the input layer, hidden layer and output layer [38][39][40][41][42][43][44][45][46].…”

“…To define the final model, time series regression analysis was performed using the multilayer perceptron (MLP) ANN model with the maximum number of hidden layers defined as 10, and linear, logistic, tanh, exponential, and sinusoidal functions were used as activation functions for hidden and output neurons [38][39][40][41][42][43][44][45][46].…”

Hourly Water Level Forecasting at Tributary Affected by Main River Condition

“…Moreover, researchers have applied various techniques of neural network models to obtain better prediction accuracy [15]. The many types of neural network model applications include typical backpropagation ANN (BP-ANN) [16], feedforward neural network (FFNN) using multilayer perceptron (MLP) and radial basis function network (RBF) [17], and recurrent neural network (RNN) [18] as dynamic models. Theoretical improvement trials enable researchers to apply ANN models for various applications.…”