Design And Implementation Of Data Fusion Method Based On D-S Evidence Theory

Multi-source data fusion technology can integrate incomplete information collected by different data sources and process them accordingly to make the advantages of discrete data complement each other,thereby weakening the uncertainties in data sources.Data fusion technology is now used in such areas as target identification,automation,situational assessment and geosciences,and have been gradually expanded in areas such as social safety,remote sensing images,pollution detection,and climate analysis.Nowadays,the rapid development of related technologies such as mobile Internet technology and big data is accompanied by the emergence and growth of explosive data volume.At the same time,the development of electronic devices makes the data sources complex and diverse,and how to quickly and effectively integrate large amounts of data dealing with the right conclusions and decisions becomes a problem.D-S evidence theory has been widely used as one of the classical data fusion methods.However,there are still many problems to be improved in the application of D-S evidence theory,such as evidence source independence,focal element explosion,basic probability assignment acquisition and conflict of evidence and other issues.The paradox problem caused by evidence conflict is the most serious.When there is a high degree of conflict or complete conflict between evidence,the data fusion method based on D-S evidence theory will inevitably have invalid or wrong conclusions after using D-S combination rules to fuse.There are some differences between the fusion result and the conventional thinking,which makes D-S evidence theory cannot be more fully applied.Based on the detailed analysis of the existing problems of D-S evidence theory,this paper combines D-S evidence theory with data fusion,proposes the data fusion method based on improved D-S evidence theory by introducing the Bhattacharyya distance,putting forward the confidence level of evidence and improving the rule of evidence combination,which effectively solves the evidence conflict problem in D-S evidence theory.The experimental results show that the improved data fusion algorithm can get the result of data fusion well,and has a high accuracy and credibility.The traditional data fusion system in the face of large-scale data sets,due to its computing and storage capacity constraints and cannot achieve the desired results.In order to meet the demand of data fusion under big data,we design and implement a data fusion system based on Hadoop platform by combining data fusion technology with the distributed storage and processing technology based on big data.This paper describes the requirements of the data fusion system analysis,the basic system structure,system processing flow and system function modules,and the data fusion algorithm proposed in this paper is well combined with the MapReduce programming model of the Hadoop platform.Finally,the data fusion system based on Hadoop platform is tested.The test results show that the data fusion system can well realize the functions of data acquisition and preprocessing,data storage,data fusion and user management in requirement analysis,and can effectively,accurately accomplish the task of data fusion.