A COMPARATIVE STUDY ON ARTIFICIAL NEURAL NETWORKS FOR ENVIRONMENTAL QUALITY ASSESSMENT
Yijun Liu, Sheng He, Yao Wang, Xiumei Wang
Key laboratory of cloud computing & intelligent information processing of Changzhou City, Jiangsu University of Technology, 213001, China
The aim of this study is to use neural network tools as an environmental decision support in assessing environmental quality. A three-layer feedforward neural network using three learning approaches of BP, LM and GA-BP has been applied in non-linear modelling for the problem of environmental quality assessment. The case study shows that the well designed and trained neural networks are effective and form a useful tool for the prediction of environmental quality. Furthermore, the LM network has the fastest convergence speed and the GA-BP network outperforms the other two networks in both predictive and final classification accuracies of environmental quality.