气固两相流中基于神经网络的固相质量流量检测方法的研究

The Measurement of Solid-phase Mass Flowrate of Two-phase Flow by Using Neural Networks Method

用神经网络的自学习算法对采样数据进行辨识得出固相质量流量的黑箱模型，实现固相质量流量的在线测量。实验结果，模型最大测量误差为１０％

As to measurement of solid-phase mass flowrate in gas-solid two-phase flow system, a modeling method based on artificial neural networks (ANN) technology is proposed instead of traditional mathematical method.This provides a new way for mo deling and measurement of parameters difficult to measure in complex systems. Based on the basic theories of pneumatics and semi-empiricals of solid-phase flowrate given by traditional mathematical models,two important and measurable parameters,i.e.gas-phase mass flowrate and pipeline pressure drop are used as input parameters of our ANN model.The target parameter,i.e.,the solid-phase mass flowrate,is expected to be given by the output of the ANN model.Back-Propagation (BP) structure and its modified learning algorithm is introduced to establish the black-box model of the solid-phase mass flowrate.The established black-box model is then used as the on-line measurement model of solid-phase mass flowrate.Both simulation results and experimental results on a pneumatic conveying system of powder prove the feasibility of the ANN modeling method.Experimental results show the average error of the model is 4 2% and the maximum error 10%.