Study Of Ground Mineral Water Quality Prediction Based On Support Vector Machine

Ground water quality prediction is one of the most important factors among all the ground water quality assessment. The result can be used to represent ground water quality and pollution status as well as prediction of future water quality. There have been many mathematical models that are used to predict and assess water quality. However, there is few generalized acceptable model. There is arbitrariness in choosing the models for assessment. Although these methods dominate in practical application, people still find many deficiencies.In this thesis, the current groundwater quality prediction methods were systematically summarized and the essential issues and main contents of statistical learning theory are elaborated. Based on some of the defects in evaluation and prediction methods, Support Vector Machine (SVM) is introduced. A brief revision of SVM development and research status quo is made and the strong points of SVM method are summed up. SVM is a novel powerful machine learning method developed in the framework of Statistical Learning Theory (SLT). SVM solves practical problems such as small samples, nonlinearity, over learning, high dimension and local minima, which exit in most of learning methods, and has high generalization.In this paper, we studied SVM as regression techniques for natural mineral water quality parameters prediction. A new SVM algorithm is put forward to solve the prediction problem of concentration of H2SO3 and Nitrite regarding the supervision of natural mineral water quality parameter average concentration as a time series prediction question. The supervision data of pollution material concentration is performed regression estimate analysis using SVM. Compared with the result of BP network method, the prediction model established by SVR method can make full use of the distribution features and the prediction result fits the practical condition better. The result of the experiment shows that the method based on SVR method is more performance compared to BP Neural Network. Finally, we put forward several protection measures to solve the reflecting problems of underground natural mineral water quality prediction in order to promote the continuous improvement of underground mineral water, and realize the sustainable development and utilization.The innovation of this thesis lies in the following: First of all, this paper puts forward SVM technique as regression techniques for natural mineral water quality parameters prediction for the first time in domestic, enriches and expands the theory and application scope of SVM. Secondly,it predicts the changes in key indicators of underground natural mineral water during a period of time, which is as the supervision of management and mining of underground natural mineral water. Thirdly, it offers several protective suggestions for the sustainable development of underground natural mineral water, which is as reference for the same types of underground natural mineral water.