Research On Buried Pipes Detecting Method Based On Fusing Multi-sensor Data

With the development of cities,the buried pipeline network has become more com-plex,thus the management of the network is becoming more and more important.The information of some part of the network is absent and the change of the network because of new pipeline constructions propose a requirement for the more efficient and more ac-curate technology for trenchless locating and mapping of buried utilities.Commonly,there are two research orientations to satisfy the requirement:proposing better sensor technologies and designing better data processing algorithms.Based on a car shaped device with passive magnetic field coils and a ground pene-trating radar,a detecting solution is proposed in this paper,which could realize one-time detection and present 3D map of buried pipes in nearly real time.As a part of it,a series of data processing methods are designed.The whole procedure could be divided into two steps.The first step is to interpret original data of sensors into possible spatial points of buried utilities and the second one is to fuse these points to evaluate the trend curve of pipes.For the first step,specific interpreting methods are designed for each sensor.For the second step,a novel data fusion method,named Pipe Revising and Evolving(PRE),is proposed to deal with the false-alarm problem caused by the noise of data.The main contributions include the following three aspects:-A new cable-detecting method with PMFC is designed.Combined with GPS module,the device only need scan along a scan-line once,then the location and the direction of the buried cable could be analysed.To find the time being above the cable for a group of coils,a maximum-minimum(MM)algorithm is designed.-Since the device need scan along a route crossing over the buried pipes,for the route design problem of scanning,zigzag layout is proposed since it reduces the scanning route in a unit of length and has a larger detectable orientation range than the parallel layout.-PRE is proposed for the false-alarmed problem caused by noise.The spatial points from interpreting methods contain ones reflecting real pipes,or false-alarmed ones.Based on the linear characteristic of pipes,PRE would screen hypothetical pipes with the analysis of the probability to obtain a highly reliable result.For the real-time feature,PRE is designed as an iterative format.