Marine Water Quality Monitoring Based On Multi-sensor Data Fusion

At present, the marine pollution is serious in our country. To make full use of themarine resources and develop marine economy, our country is paying more and moreattention and making the key research on evaluating water quality accurately, and putforward higher request about the real-time monitoring technology of marine waterquality and the real-time processing technology of the huge amounts of information.Realizing the comprehensive monitoring of water quality and making decisions is ofgreat significance to the comprehensive use of marine resources, and marine pollutionwarning, especially the marine ecological environment protection in our country. Thepaper depends on the technology of multi-sensor date fusion. In this paper D-S evidencetheory and fuzzy neural network are both fusion algorithm, they are applied to evaluatewater quality. The system is established to evaluate real-time water quality accurately,The main research contents include:Firstly, the paper discussed the status on the multi-sensor data fusion at home andabroad, the exiting problem, development trends, the structure and method are introducedin detail.Secondly, the paper introduced the improved BP algorithm of neural network, whichcan be used to evaluate the water quality. The paper analysis and obtained the parameterswhich influence the speed of network training, and obtain the best neural networkstructure, built neural network model. The training verified the validity of the modethrough the online analysis of the water and the output of the corresponding code.Then, the paper introduced the testing plan of the water quality monitoring, the planis based on D-S evidence theory, The paper determined the method to get basicprobability assignment. the results of the quality are obtained according multi-sensortechnology and single sensor technology based on D-S evidence theory. Experienceverified that the accuracy of multi-sensor technology is better than single sensortechnology.Finally, Combining the characteristics of water quality evaluation and the advantage of D-S evidence theory and neural network in water quality evaluation, the paper putforward fusion algorithm that it is applied to evaluate water quality, which is based onfuzzy neural network and D-S evidence theory. Two-stage fusion algorithm is applied todetect the water quality. The experience verified the feasibility of fusion algorithmdesigned in this paper, the paper compared the accuracy of different fusion methods, andpoint out the deficiency and improvements of this system.