Neuro based acoustic diagnosis of gas leakage in pipeline

In industry, such as oil refinery industry, there may occur various kinds of safety problems for pipelines aged after its constructions. To realize preventive maintenance of pipelines, there are large needs for the diagnosis technology of gas leakage. In this study, gas leakage sounds generated from the piping crack is analized and tried to be used for detection of the gas leakage. Sound data for analysis are generated and collected in pipeline where background noise is small. To diagnose the crack, sound data for analysis are sampled applying Fast Fourier Transform. Classification and discrimination of cracks are carried out using neural network and K-nearest neighbor methods. As a result of the experiments for classification, the size of the crack and gas pressure were successfully classified.