神经网络改进算法在超声波流量测量中的应用  

Study on Ultrasonic Flow Measurement of Perturbed Flow

作　　者：李芳[1] 李艾华[1] 吴朝军 

机构地区：[1]第二炮兵工程学院502教研室,陕西西安710025 [2]二炮装备部驻北京地区军事代表局,北京100085

出　　处：《计算机测量与控制》2008年第2期163-164,175,共3页Computer Measurement &Control

摘　　要：由于液压系统中管路结构造成的紊流现象对传统超声波流量计的影响,文章在非充分发展流体的截面流速分布特点的基础上,提出采用多通道超声波测量方法获取流速信息,并将BP神经网络系统应用于流态识别和流量计算,通过对BP网络进行节点和阈值优化,提高了神经网络的识别精度;采用此方法不仅提高了超声波流量计对于非充分发展流体的检测能力,而且降低流量计在测量安装时对直管短长度的要求,增强了仪器的适应能力,具有广泛的工业应用前景。According to the influence caused by pipeline structure in hydraulic system to traditional ultrasonic flowmeter of perturbed flow, a method of ultrasonic multipath measuring was adopted to obtain flow velocity basedon the characteristic of deficient evolution flow velocity in section. Anda BP neural networks was applied to fluid recognition. BP neural networks has been optimized to improve measurement precision and adaptability of ultrasonic flowmeter by node and threshold optimization. The method has improved the detection ability of ultrasonic flowmeter to deficient evolution fluid. The requirement of straight pipe length in flowmeter installation was depressed. The adapting ability of instrument was enhanced. It will be widely used in industry field.

关 键 词：超声波 流量测量 紊流神经网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]