Soft Computing Technique in the Water Sector: Artificial Neural Network Approach

Abstract

Soft computing techniques are general statistical tools used in various ways to fit the data-based process and satisfy the dire need of a specific application. Artificial neural network (ANN) is one of the soft computing techniques and could be applied effectively in water sector management. This study provides an overall view about a type of soft computing techniques that is an ANN, which covers its history, few of their kinds with training algorithm, model testing, shortcoming with training, and finally its effective application in one of the water sector managements. There are numerous ANNs existing today based on various neural network structures and training algorithms and it is not feasible to cover all, so in this study only three types of ANN which are generally used, namely, feed-forward backpropagation, generalized regression, and radial basis function neural networks are discussed in detail with their 84training algorithms. And finally, the development, application of rainfall-runoff model and its implementation using ANN in water sector are discussed.