基于神经网络软测量的动态流量测量方法研究

文章以圆管中的粘性流体为研究对象,分析了管道中压差、温度、速度的变化对动态流量的影响,设计出了用于软测量建模的RBF神经网络结构;与此同时,还设计了动态流量测量管,并在伺服阀动态性能试验台上采集了相关数据,其间采用超声波类时差法测量了流速信息;利用采集的数据基于NeuroSolution软件对RBF网络进行了学习训练、检验和测试,最后建立了动态流量的软测量模型;通过将流量预测曲线与实测曲线相对比,结果发现该软测量模型具有很高的逼近精度,它可为动态流量的测量提供一条新的途径.

Research on Dynamical Flow Measuring Method Based on Neural Network Soft Measurement

In this article, the affect caused by the change of the fluid flow pressure difference, temperature and velocity in the pipeline to the dynamic flow is analyzed with the viscous fluid in the round pipe as a researching object and the structure of RBF neural network used in soft measurement modeling is de- signed, at the same time, the dynamic flow testing tube designed and the relation datum are collected from the servo-valve performance test-stand; during this period, the information of flow velocity is detected by means of ultrasonic time-like difference way; with collected datum and basing on NeuroSolution soft- ware, RBF neural network is trained, detected and tested; finally the dynamic flow soft measuring model is set up. Through comparison between the flow predictive curve and measured curve, it is found that the soft measuring method has excellent approximant precision and can offer one new method for measuring dynamic flow.