| The detection of natural gas pipeline defects is an important part of pipeline integrity management and a key measure to ensure the safe operation of natural gas pipelines.However,there are many noises and weak defect characteristics in pipeline defect signals,which results in poor pipeline diagnosis accuracy.In order to accurately diagnose pipeline defects and reduce the occurrence rate of pipeline accidents,study the noise reduction theory,characteristic parameter distribution rules and diagnostic methods of pipeline defect signals.(1)Based on the theory of wavelet threshold noise reduction,a compromise function of soft and hard thresholds is established,and a fruit fly optimization algorithm is introduced to adaptively select wavelet thresholds for noise reduction.Compared with the traditional noise reduction method,FOA-WTD reduces the signal reconstruction error,completely preserves the characteristic of the signal,and is suitable for the noise reduction processing of pipeline defect signals.(2)The standard deviation method was used to quantify the number of decompositions of CEEMD and the amplitude of white noise addition,and the number of decompositions of the signal was determined by the amplitude of white noise addition.The CEEMD decomposition method based on standard deviation can improve the decomposition accuracy of pipeline defect signals,avoid the modal aliasing phenomenon of signal decomposition,and prepare for signal feature extraction.(3)By extracting the characteristic parameters of the CEEMD method,the energy entropy values of the crack and the circular hole are both increasing.Compared with the wavelet packet and EMD methods,the characteristic parameters based on the CEEMD have obvious distribution rules.Through experimental analysis,the energy entropy value has the characteristics of strong anti-noise ability and good stability,and can be used as the input feature matrix of FCM cluster analysis.(4)Using simulated annealing algorithm and genetic algorithm to optimize FCM,input the energy entropy matrix of CEEMD to complete the diagnosis of pipeline defect signal.The research shows that the comprehensive diagnosis accuracy of laboratory pipeline defect signals based on CEEMD-FCM reaches 88%,and the comprehensive diagnosis accuracy of pipeline defect signals of natural gas compressor stations is 83%,which realizes the accurate diagnosis of natural gas pipeline defect signal. |
