Leakage Detection Of Natural Gas Pipeline Based On Rough Set And Neural Networks

With the continuous development of natural gas pipeline transportation,environmental pollution, economic loss and harm to people’s property safety caused by pipeline leakage become more serious.And,how to reduce or even avoid these problems become increasingly concerned.So natural gas pipeline leak detection is still a current important research topics. Conventional leak detection technology to pipeline leak detection have the disadvantages of low efficiency and inaccurate diagnosis,and these methods do not apply to natural gas pipeline leak detection. In this paper, according to the actual circumstance of natural gas pipeline operation, combines rough set and neural network, to build the model is suitable for natural gas pipeline leak detection. This article main research content is as follows:This paper first summarizes the pipeline leak detection development situation at home and abroad, then introduces the existing detection technology. Based on the comprehensive comparison of various leak detection methods,this paper put forward the natural gas pipeline leak detection method that rough set and neural network combined.In this paper, based on the research of the basic theory of neural network, construct a RBF neural network model.For Daqing natural gas pipeline operation condition,this paper select eight property parameters as input parameters of RBF neural network, and determine whether a leak signal or not.In order to solve the problem that the neural network input space dimension is bigger and the network structure is complex,in this paper,integrated the advantages and disadvantages of rough set and neural network,and make them to combined,build a system of RBF neural network with rough sets as the front system.Make the data as training samples and testing samples which are treated by rough sets for the network,and determine whether the input signal for leakage signal. In order to further improve the judgement precision of the neural network,PSO optimization algorithm are introduced in detail in this paper, an improved PSO algorithm is put forward on the basis of the standard PSO algorithm,which is used to optimize the weights of RBF neural network. Build RBF neural network model optimized by the improved PSO algorithm after rough sets system, and applied to gas pipeline leak detection.This paper introduces the principle of natural gas pipeline detection based on sound,and uses wavelet analysis to remove noise from the original sound signal. Using the correlation method to determine the time difference between sound propagation to upstream of the sensor and downstream sensors, combining with the sound velocity in gas pipeline to find the location of the leakage source.