The System Design Of MFL Detection And The Method Research Of Data Processing

As one of the main energies in the world, oil has been regarded as the lifeblood of the economy. Because pipeline is one of the safest, most reliable and most efficient transports of oil, it has been developed rapidly in all countries. However, pipeline prones to wear, corrosion and leakage phenomenon with the increase of in-service time and a variety of environmental changes and vandalism, which can cause environmental pollution, economic loss and life-threatening. In order to avoid huge losses caused by the oil pipeline leak, people need to make regular checks on the pipeline and the best way is magnetic flux leakage. Due to technical limitations and adverse environmental impacts, data obtained by detection often can’t be used for analyzing pipeline defects directly. The original data need to be processed to become tidier and smoother. According to the existing equipment, processing algorithm and the characteristics of MFL signal, this thesis designs a complete pipeline magnetic flux leakage detection system. This thesis consists of the following three parts:Firstly, this thesis designs a data acquisition system. According to the system requirements, this thesis designs a pipeline magnetic flux leakage detection device based on NI Compact-RIO, including the analysis of sensor indicators and the design of related circuit. Besides, this thesis designs a data acquisition and storage procedures. In the selection of data acquisition rate, this thesis designs a dynamic sampling rate algorithm due to different operating conditions of the pipeline detector, which can get more information and save more storage space.Secondly, this thesis disposes magnetic flux leakage signal with basic mining method. The basic mining consists of data cleaning, data compensation and data reduction. In data cleaning, this thesis cleans the magnetic flux leakage signal according to Grubbs test method. In data compensation, this thesis compensates for the wild-point with interpolation method, and compensates for the magnetic flux leakage signal with two-dimensional interpolation, which can make the signal smoother. In data reduction, this thesis normalizes magnetic flux leakage and enables the realization of compensation theory about velocity effect.Finally, this thesis disposes magnetic flux leakage signal with different filtering methods. According to different sources of magnetic flux leakage signal, this thesis presents a filtering method based on classification and compensation. In signal classification, this thesis uses an improved Regional Growth Rules method to classify signals accurately. In magnetic flux leakage signal filtering, this thesis uses fast watershed algorithm based on adaptive threshold to extract feature points, and then compensate feature points in order to compensate for the distortion caused by filtering. In the normal magnetic flux leakage signal filtering, this thesis uses fast sliding mean method to improve the speed of the non-critical signal filtering. In the magnetic flux leakage signal reconstruction, this thesis uses interpolation to make signal smoother in connection.