Research On Pipeline Leakage Detection Based On Noise Cancellation And Elman Neural Network

Pipeline transportation has been widely used in city lives, industries and other fields with the development of national economies. Because that pipeline transportation has many advantages such as the big transport volume, low cost, strong continuity and so on, it has a very big function on economic modernization that can not be replaced. Pipeline transportation has become a separate traffic transport section-pipeline carrying trade. So the problem that how to make sure that the pipeline transportation can run safely and steadily is becoming more and more important.Based on the real situation of pipeline transportation in our country, this thesis deeply researched the techniques of pipeline leakage detection based on noise cancellation and Elman neural network. In the traditional techniques of pipeline leakage detection, the relationship between input and output which is established by the forward network which is usually be used is often static, but the actual negative pressure wave signals are usually dynamic. So the static network model can not show the dynamic characteristics of the actual negative pressure wave signals in the system accurately. Unlike the forward network, the Elman neural network which is proposed in this thesis has information delay and information feedback. So the Elman neural network can show the dynamic characteristics of the actual negative pressure wave signals in the system accurately.The main research work is described as follows:First of all, this thesis introduced the basic situation of pipeline transportation, then introduced the history, current situation and future development of the pipeline transportation in the country and abroad briefly; some common methods of pipeline leakage detection were introduced according to their types, then analyzed their theories, feature, applications and so on.Secondly, in order to remove the noises in the actual negative pressure wave signals, the noise canceling algorithm of leakage signals was proposed based on wavelet theories and noise cancellation, which provided a sound base for the leakage detection with Elman neural network.Thridly, based on the signal without noises, negative pressure wave signals were changed into vector form. The method of pipeline leakage detection was gaven based on the ability of Elman neural network to show the dynamic characteristics of signals.Finally, according to the research above, the improved Elman neural network was used in the technique of pipeline leakage detection. Compared with the Elman neural network before being improved, the improved one was proved to be better.According to the algorithm in this thesis, many simulation experiments were carried out with the software of MATLAB to make sure that the results are effective and reliable. The results proved that the fault detection algorithm in this thesis can identify the leakage fault accurately, so this method can be used in pipeline leakage detection.