Neural Networks - Assisting in Water Quality Modelling

The use of artificial neural networks for modelling water quality and nutrient loads from runoff events in water supply catchments in South Australia has been investigated. The artificial neural network applied to the problem was able to 'learn' non-linear relationships from a set of input-output data for. a given system. The knowledge gained from the> input-output set is then distributed across the memory units as the connection weights of the processing elements. This trained set is then used to model different input sets to determine various output sets. The use of an artificial neural network to model these nonlinear variable systems has been shown to be better than conventional statistical modelling. A short history of the development of artificial neural networks is given and some of the basic concepts are introduced.