Pipeline Blockage Detection Based on Sparse Characterization Classification Enhanced by VMD

Regarding the active acoustical method of determining the blocking status in buried pipe using low frequency sound waves as excitation, it has difficulties in distinguishing the different degrees of blocking and eliminating the interference from the pipe components such as lateral connection, a novel approach to identify the blocking fault using the Sparse Representation for Classification (SRC) enhanced by variational mode decomposition was proposed in this paper. Firstly, the acoustic signals collected in the pipeline were decomposed by variational mode decomposition, then sample entropy characteristic parameters were extracted from each component. Secondly, the feature vectors that can effectively represent the signals were used to construct the dictionary and the sparse representation classifier. The experimental results had shown that the method could effectively identify the blocking status in different degrees, the lateral connection could also be discriminated accurately in the pipeline under a small number of samples, which has certain practical value for engineering applications.