APPLICATION OF ANN AND SVM MULTICLASS MODELS USED FOR MULTI-SENSOR MONITORING AND MEASUREMENT OF SURFACE WATER QUALITY

摘要

The control, measurement and monitoring of water quality is becoming more and more interesting because of its effects on human life. Many approaches were developed in order to ameliorate the water quality monitoring process. This paper presents the application of Artificial Neural Network (ANN) and Support Vector Machines (SVM) multiclass classification techniques in control and monitoring of water quality. This study involved the interpretation of surface water quality data of Tilesdit dam (Algeria). A multi-class problem is a typical example for solving the mentioned problem. The MLP networks and the algorithm of SVM, one-against-all, are the most popular strategies for multi-class problems. In this work, the training phase is carried out using these approaches to supervise water quality from four several physicochemical parameters such as temperature, pH, Conductivity and Turbidity. These parameters of water quality indicators were collected at Tilesdit production station during three years (20092011). In order to evaluate their performances, a simulation using real dataset measured from study area station, corresponding to the recognition rates (training and test), training time and robustness, is carried out The results are compared to get the best performance evaluation of the intelligent proposed approach for of monitoring and measurement of water quality.