基于压力信号采集的输气管道混合模型的建立

为监测输气管道的运行状态，提出一种基于机理模型和神经网络模型的混合建模方法. 机理主模型是基于气体在管道中流动的连续性方程、运动方程和气体状态方程而建立的；神经网络模型用来补偿机理模型建模过程中的简化处理及因忽略某些动态参数变化带来的误差，提高了混合模型建模精度，为下一步进行气体管道的泄漏检测和定位奠定基础. 为避免流量计检测精度较低的缺点，实验中用高精度压力传感器取代流量计，统一采集压力信号，提高检测精度. 基于实验采集压力数据，将机理模型和混合模型输出的精度进行比较. 结果表明混合模型的精度得到了较大提高. 

Hybrid Modeling of Gas Transport Pipeline Based on Pressure Signals

To monitor the operating state of gas pipelines, a hybrid model has been established on the basis of the combination of mechanism model and neural network model. Firstly, the mechanism model was built based on the basic transport flow equations with the consideration of the mass balance condition, momentum balance condition and state equation. The mechanism model acts as the primary model in hybrid modeling. Then the neural network model was used to compensate the error of the experiential model which was raised by simplification and ignorance of some dynamic parameters during the modeling. The neural network acts as the compensatory model to improve the modeling precision. In addition, for increasing the detection precision, the high-precision pressure sensor was used to sample the gas pipeline signal instead of the low-precision flow-meter. Finally, the precision contrast between mechanism model and hybrid model shows that the detection precision achieved by using hybrid model is improved obviously.