基于LPV模型GRNN输气管道音波定位算法

Acoustic Location Algorithm in Gas Pipelines Based on the GRNN of LPV Model Approach

针对输气管道泄漏检测及定位问题以及管道内气体可压缩、检测难等特点,建立了输气管道线性变参数(LPV)模型,并设计了广义回归神经网络(GRNN),以理论时间差为模型输入,以对应的管道各点位置为期望输出.采用音波法对输气管道进行泄漏故障诊断与定位.结合具体实例并采用现场数据进行仿真研究,结果表明:采用基于LPV模型的GRNN输气管道泄漏故障音波定位算法是一种有效的方法,可使预测值准确地跟踪真实值,实验结果为输气管道泄漏故障检测与定位的工业应用提供了可靠的依据.

Due to the problems of leakage detection and location in gas pipelines as well as the measurement difficulties for the gas in pipelines,agas pipeline linear parameter varying（LPV） model was established,and the generalized regression neural network（GRNN）was designed,which took the theoretic time difference as input model,and the location of different point in pipes as output.Acoustic identification method was used for gas pipeline leakage fault diagnosis and location.Simulation was carried out using field data,and the results showed that acoustic location algorithm research in gas pipelines based on the GRNN of LPV model approach is an effective method,which can make the predictive value accurately track real value.The experimental results can provide the reliable basis for the industrial application of gas pipeline leak detection and location.