The Application of Adaptive Wavelet Neural Network in Ship Pipeline Leakage Detection

Abstract

At present, pipeline leakage detecting technology mostly focuses on long-distance oil and gas pipeline, but does a little on the complicated pipeline system, for example ship pipeline system. The leakage detecting negative pressure wave method may easily fail to report or even misreport, because the original signal has lots of noises and the frequent adjustable pump or the reset valve of ship pipeline will produce negative pressure wave. In order to monitor the leakage of the ship pipeline, the singularity of negative pressure wave signal was analyzed through wavelet transform modulus maxima to extract the pipe working condition feature vectors effectively. From data sampling in terms of the pipe working conditions, the learning samples are obtained. Accordingly, the nonlinear mapping between adaptive neural network inputs and outputs is well established via training. Afterwards, the pipeline system leakage is detected based on input feature vectors. According to the analysis and verification from the experimental data, this method will effectively reduce the missing alarm rate and false alarm rate.