Partial blockage detection in underground pipe based on guided wave&semi-supervised learning

Aiming at the detection problem of blockage in urban water supply pipelines and drainage pipelines, also the problem to distinguish commonly used pipe components such as lateral connection from the actual blocking conditions. A method based on dual-tree complex wavelet transform and Safe Semi-Supervised Support Vector Machine for blockage recognition is proposed in this paper. The first step of this method is to decompose the acoustic signals obtained from the pipeline by the dual-tree complex wavelet transform, and then convert the acquired components into Sound pressure level. Secondly, the pulse factor and the average acoustic energy density are extracted respectively from the effective components as acoustical features. Finally, the S4VM classifier is applied to cluster and label the untrained data, furthermore the different degree of the blocking is able to identify as well as the pipe components.