Industrial Oil Pipeline Leakage Detection Based on Extreme Learning Machine Method

Abstract

Pipeline transportation plays a significant role in modern industry, and it is an important way to transport many kinds of oils and natural gases. Industrial oil pipeline leakage will cause many unexpected circumstances, such as soil pollution, air pollution, casualties and economic losses. An extreme learning machine (ELM) method is proposed to detect the pipeline leakage online. The algorithm of ELM has been optimized based on the traditional neural network, so the training speed of ELM is much faster than traditional ones, also the generalization ability has become stronger. The industrial oil pipeline leakage simulation experiments are studied. The simulation results showed that the performance of ELM is better than BP and RBF neural networks on the pipeline leakage classification accuracy and speed.