Support vector regression (SVR) models have been designed to predict the concentration of chemical oxygen demand in sequential batch reactors under high temperatures. The complex internal interaction between the sludge characteristics and their influent were used to develop the models. The prediction becomes harder when dealing with a limited dataset due to the limitation of the experimental works. A radial basis function algorithm with selected kernel parameters of cost and gamma was used to developed SVR models. The kernel parameters were selected by using a grid search method and were further optimized by using particle swarm optimization and genetic algorithm. The SVR models were then compared with an artificial neural network. The prediction results R2 were within >90% for all predicted concentration of COD. The results showed the potential of SVR for simulating the complex aerobic granulation process and providing an excellent tool to help predict the behaviour in aerobic granular reactors of wastewater treatment.
Aerobic granular sludge (AGS) is one of the treatment methods often used in wastewater systems. The dynamic behavior of AGS is complex and hard to predict especially when it comes to a limited data set. Theoretically, support vector machine (SVM) is a good prediction tool in handling limited data set. In this paper, an improved SVM using optimization approaches for better predictions is proposed. Two different types of optimization are built which are particle swarm optimization (PSO) and genetic algorithm (GA). The prediction of the models using SVM-PSO, SVM-GA and SVM-Grid Search are developed and compared prior to several feature analysis for verification purposes. The experimental data under hot temperature of 50˚C obtained from sequencing batch reactor is used. From simulation results, the proposed SVM with optimizations improve the prediction of chemical oxygen demand compared to the conventional grid search method and hence provide better prediction of effluent quality using AGS wastewater treatment systems.
To comply with growing demand for high effluent quality of Domestic Wastewater Treatment Plant (WWTP), a simple and reliable prediction model is thus needed. The wastewater treatment technology considered in this paper is an Aerobic Granular Sludge (AGS). The AGS systems are fundamentally complex due to uncertainty and non-linearity of the system makes it hard to predict. This paper presents model predictions and optimization as a tool in predicting the performance of the AGS. The input-output data used in model prediction are (COD, TN, TP, AN, and MLSS). After feature analysis, the prediction of the models using Support Vector Machine (SVM) and Feed-Forward Neural Network (FFNN) are developed and compared. The simulation of the model uses the experimental data obtained from Sequencing Batch Reactor under hot temperature of 50˚C. The simulation results indicated that the SVM is preferable to FFNN and it can provide a useful tool in predicting the effluent quality of WWTP.
The main challenges to achieving a reliable model which can predict well the process are the nonlinearities associated with many biological and biochemical processes in the system. Artificial intelligent approaches revolved as better alternative in predicting the system. Typical measured variables for effluent quality of wastewater treatment plant are pH, and mixed liquor suspended solids (MLSS). This paper presents an adaptive neuro-fuzzy inference system (ANFIS) and feed-forward neural network (FFNN) modeling applied to the domestic plant of the Bunus regional sewage treatment plant. ANFIS and feedforward neural network techniques as nonlinear function approximators have demonstrated the capability of predicting nonlinear behaviour of the system. The data for the period of two years and nine months sampled weekly (140 week samples) were collected and used for this study. Simulation studies showed that the prediction capability of the ANFIS model is somehow better than that of the FFNN model. The ANFIS model may serves as a valuable prediction tool for the plant.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.