中文摘要：

本文通过对粘性流体在圆管中的层流和湍流流量方程对比研究发现，动态流量主要与管道中摩擦导致的压头损失、管道中最大的流速、流体温度变化有关，依据这一原理设计了基于径向基函数人工神经网络（RBFNN）的软测量模型。在伺服阀动态性能实验台上构建了数据采集系统，在新型动态流量测量管上安装超声波、压差、温度传感器来采集各种信息，其中流速信息采用一种新颖的超声波类时差法获取，用于标定的实际流量由无载液压缸的速度传感器获取。基于NeuroSolution软件中的RBF网络模块组成软测量RBFNN，选用部分采集数据作为学习样本对RBFNN进行训练，建立了动态流量的软测量模型。利用采集的数据的测试样本对RBFNN进行测试，通过流量预测曲线和实际曲线的对比，验证了该软测量模型具有很高的逼近精度。该软测量方法为动态流量的测量提供了一条新的途径。

英文摘要：

By comparing the laminar flow and the turbulent flow of viscous fluid in circular pipe, it is found that the dynamic flow is mainly related to the head loss caused by the friction, the largest flow rate and the fluid temperature. According to this principle, a soft sensor model based on the radial basis function neural network （RBFNN） is designed. A data acquisition system is built on the dynamic performance test-bed of servo valve. The ultrasonic, pressure and temperature sensors have been installed on the new dynamic flow measuring pipe to collect relevant informa- tion. The flow rate is measured by use of a new ultrasonic time difference-like method. The actual flow for the calibration is gained from the speed sensor on the no-load hydraulic cylinder. The soft sensing RBFNN has been constructed by use of the RBF network module of the NeuroSolution software. The RBFNN trained by use of learning samples from the collected data. In this way, the soft sensing model for dynamic flow testing has been established. The RBFNN has been tested by use of testing samples from the collected data. By comparing the flow predicting curve and the actual flow curve, it shows that the soft sensing model has a high approximating precision. The soft sensing method provides a new way for dynamic flow measuring.