基于神经网络的气体管道泄漏检测与定位

针对建立气体管道准确机理模型比较困难的问题，提出基于压力时间序列的神经网络来进行气体管道泄漏检测与定位的方法。针对管道的流量测量的精度较低且安装流量计会破坏管道原流场的问题，提出采用主动施扰法检测气体管道不同位置的压力信号，利用压力时间序列神经网络获取泄漏点的位置信息和泄漏量信息，并通过实验分析泄漏位置不同和泄漏量大小对检测精度的影响。实验结果表明，该方法能实现气体管道泄漏检测与定位，并且精度较高。

Leakage detecting and localizing for gas pipeline based on neural network

To avoid the difficulty of building the gas pipeline mechanism model, the method of leakage detection and localization for gas pipeline is proposed, based on pressure time series. Because of the low accuracy in measurement and destroy of the pipeline original flow in flowmeter installation, this paper introduces the active-disturbance method to obtain the location and leakage according the pressure detection of different positions and analyzes the effect of the different leakage position and leakage quantity on detection precision. The experiment indicates the method could accomplish the higher-accuracy detection and location for gas pipeline.