developing a proper model for online estimation of the 5-day biochemical oxygen demand based on artificial neural network and support vector machine

recently, hardware sensors are widely used in monitoring and measurement of water quality parameters. constraint of the instrument to measure some water quality parameters such as the 5-day biochemical oxygen demand (bod5), which are time consuming, causes efforts are diverted to the use of software sensors for online prediction of bod5. the main goal of this research is developing an appropriate software sensor based on artificial neural network (ann) and supported vector machines (svm) models for online prediction of bod5 in the sefidrood river. for this purpose, appropriate models with ann and svm are developed by considering bod5 as a function of other water quality variables. in the development of ann model the role of various training functions such as levenberg-marquardt (lm), resilient back-propagation (rp) and scaled conjugate gradient (scg) algorithms on optimization of ann parameters is evaluated. also for optimization of svm parameters, two-step grid search algorithm is conducted. the results of this research indicated that superior performance of ann model with lm algorithm (ann (lm) model) than the other two algorithms i.e. rp and scg. besides svm model had a suitable performance in bod5 prediction, so that pearson correlation coefficient (r) in the test step of the model obtained as 0.95. finally, the further investigation for selection of the best model between ann (lm) and svm based on developed discrepancy ratio statistic is executed. results of ddr statistic indicated superior performance of svm model than ann (lm) for online prediction of bod5 in the sefidrood river.