A MATHEMATICAL MODELING MODEL OF RIVER’S WATER POLLUTION CONSEQUENCES WITH USE OF NEURAL NETWORK, WHICH BASED ON REGRESSION PROBLEM

Abstract

This article describes a method of river’s water pollution forecast results adjusting, as a part of results intellectual processing, based on series of Harvey Jobson empirical hydrological equations. Basic principles, functioning logic and a process of contextual methods validation were reasoned. The capabilities of the neural network using the regression problem as a method of determining the coefficient that corrects the main result based on error were used to solve this goal. The method of error correction is based on the imposition of a coefficient on some result of peak concentration characteristic value, measured for a particular point. The model provides the use of series of empirical hydrological Harvey Jobson equations, as main method for forecasting the effect of water pollution, which has empirical nature, and doesn’t require detailed input data for providing measurements. This method is part of the intellectual result processing, rather than making predictions. With use of neural network, which functions on regression problem, method of forecasting water pollution results error correcting. A possibility of neural network use with another methods, which measures peak concentration in certain location on specific time.