常压塔汽油干点观测方法研究

STUDY OBSERVING METHOD FOR GASOLINE ENDPOINT OF CRUDE UNIT

提出了一种基于系统机理方程的常压塔汽油干点神经网络观测方法。其步骤是 :首先建立系统的机理方程 ,并基于此方程构造观测器观测出常压塔一些不可测量但可观测的量 ;然后将观测量与可测量一起作为输入信号、将汽油干点作为输出信号构成神经网络观测汽油干点。与实际化验值对比 ,汽油干点误差的绝对值在 2℃以内 ,表明这一观测方法有较高的精度。

An observing method of gasoline endpoint based on equations of a system's mechanism and artificial neural network (ANN) is presented. The steps are as follows:First, the equations for describing the mechanism of a system are established, and an observer is constructed on the basis of the equations.The observer is used to observe some signals that can not be measured but can be observed. Next, the observed data and the measurable data are taken as the input signal, and the gasoline endpoint is taken as output signals to construct an artificial neural network. The artificial neural network is used to observe the endpoint of gasoline. In comparison with the actual data, the absolutely errors of observed data of gasoline endpoint are within ±2℃, which shows the high accuracy of this observing method.