| Water is the source of our lives;hence the environmental of water,more or less,influences the living standard of human beings,in that the quality of drinking water has a direct impact on the health of people drinking them.As the technology of biological investigation and microanalysis develops,people are now embracing an increasingly deeper notion of the various harmful elements in water.Systematic study of many measures like comprehensive evaluation,analysis,prediction and early warning has extremely realistic meaning for ensuring our health,maintaining social stability and promoting economic development.Drinking water system is a complex system with multiple indices,wher e drinking water quality is studied from various aspects in a comprehensi ve way.This paper does research respectively on comprehensive assessm ent,analysis,prediction and early warning system design,applying data mining and machine learning as main research methods.The research tas ks are as follows:1.The geometric mean weighting method and model based on information gain ratio are proposed for comprehensive assessment of drinking water quality;use GIS and interpolation method as auxiliary methods to show distribution of condition of drinking water quality intuitively.2.Use association analysis of data mining to comprehensively analyze drinking water;find the association rules to find the relationship between indices and drinking water quality according to different assessment models.3.Build model with indices of drinking water to predict the index which is hard to obtain using SVR;import deep belief network in deep learning to pre-train the data;better express the characteristics of data and make the accuracy of prediction higher;compare the results with the results got through original data and analyzing the reason.4.The early warning system is designed by combining comprehensive assessment,analysis and prediction.Supervision and monitoring platform of drinking water quality in Shanghai is shown. |
PDF Full Text Request