Research On Data Fusion In Multi-sensor Pipeline Detection

The technology of data fusion in multi-sensor is a new research field,which is a multiple-cross disciplines among the sensor technology,data processing and computer hardware and software technology.Due to the environment and the existence of many negative factors in the pipeline transportation,it is easy to cause the pipeline leak accidents,so we need to arrange a large number of sensors for testing the pipeline regularly.During the process of testing,the test results are influenced by outside and inside factors.Therefore,we should use a variety of detection techniques on the unify position of the same pipe,and we will encounter the problem of energy constrains and redundant information when we test to the pipeline,which is limited to the ability of communication and calculation.We mainly study the following writtens:Firstly,conduct nondestructive test with pipeline by connecting the Ultrasonic with Magnetic Flux Leakage sensors,which can get a more detailed and comprehensive test results.At the same time,apply the the data fusion technology to the pipeline detection system and build a model for the pipeline detection.Secondly,describe the improved data fusion algorithm of the union of F-CONFIS and the D-S evidence theory,then apply them to the pipeline detection,use the theory of F-CONFIS fusion for ultrasonic and magnetic flux leakage sensors respectively,then also exploit the D-S evidence theory to decision fusion.The data experiment shows that the algorithm has a good convergence fusion effect.Thirdly,base on dual column data acquisition agreement a new theory which t combines the grey theory with Kalman filter,then use the the optimal combination to forecast model.This algorithm makes full use of sensory data of space-time correlation to predict the data,considers the pipeline sensor network characteristics and mutation.We achieve the goal of high prediction accuracy,low communication and relatively lower computational by using the CGKDF algorithm.