Development of Soft Computing Models in Estimating River Water Quality by Using Different Input Combination

The exploitation of natural water resources requires recognition of the quantity and, in particular, its quality. It is important to study the quality and quantity of flow in the river in order to evaluate its locative changes for its various uses. Usually the flow crossing the river is a source of water supply in various sectors of consumption, including drinking, agriculture and industry. Therefore, knowing the changes in the quality of river flow can have a significant impact on management and planning at harvest time and water consumption, especially drinking. Various studies have been done to predict and study water quality, but in terms of the quality of surface water, less attention has been paid to smart modeling. The superiority of smart models is determined in solving nonlinear and bulky problems that cannot be solved with high precision. Najah et.al (422: 2009) also emphasized the ability of neural networks to predict Malaysian ink's river water quality indices and the ability to estimate electrical conductivity (EC) and total dissolved solids (TDS) values and opacity in this basin. Kunwar et.al (95: 2009) has also used perceptron neural networks to model the quality parameters of the biological oxygen demand (BOD) and dissolved oxygen (DO) of Gottmy river in India and has emphasized its proper efficiency.The main objective of the present research is to construct a soft calculation model for estimating the salinity of the Nisa river flow at the site of the Yalkhary hydrometric station using various input scenarios which in areas such as the present study, there is the problem of data deficits, information, as well as lack of facilities and enough cost, can be done by using an estimation model with acceptable water quality accuracy.