| The water pollution and the shortage of water resources has caused extensiveconcern of human beings, so it is very important to detect pollution in water as soon aspossible and localize the pollution source accurately. This paper has reviewed state of artresearch about localization of water pollution. Based on the diffusion model of acontinuous offshore pollution source, this thesis deals with the localization algorithm byusing spatial-temporal sampling data. The main concentrations are as follows:Firstly, based on the diffusion model of the pollutant source, features of the dynamicconcentration field generated by an offshore pollutant source are analyzed in detail, thenspatial-temporal sampling is applied to sample water concentration both on time andspace, so packet sampling data are obtained. It reduces redundant data and decreasescomputing scale compare with sampling only on time.Secondly, after completing spatial-temporal sampling, the spatial-temporal filteringmethod is presented to localize the offshore continuous pollution source, which usesunscented Kalman filter to process both spatial sampling data and temporal samplingdata. Compared with the temporal filtering localization method, the proposed method haslower computation cost, better localization accuracy and localization efficiency, betterrobust, and localization results have less dependence of initial values.Thirdly, in simulations, GMS software is employed to simulate the process of anoffshore pollution source in a lake, which demonstrates the diffusion law of the pollutionsource, temporal distribution and spatial distribution of the dynamic concentration field.Concentration data are also obtained by GMS. The spatial-temporal filtering method andthe temporal filtering only method are compared by matlab. The initial value effect ofboth localization methods are studied. Finally, localization effects of the number ofsensor nodes and initial mass flow rate under different sensor nodes are analyzed. |
PDF Full Text Request