Application of systematic clustering-back propagation network to evaluation of groundwater quality evaluation

Based on the fuzzy comprehensive evaluation and gray clustering method, the water environmental quality assessment of the BP neural network model are built by the application of neural network theory. We will combine the system clustering with neural network, overcome the study sample no pick on the characteristics, and greatly improve the learning efficiency and the precision of the network. We have researched over-fitting of neural network and determined objective function values (system accuracy) at the end of the network learning by the new network quantitative evaluation index (approximation degree). We have assessed parts of the groundwater environmental quality assessment of Pan Shi city and Compared groundwater evaluation method of GB/ T14848–93 by the model. Results show that the method is feasible.