基于BP网络的高精度超声波液位计研究  

Research on high accuracy ultrasonic liquid lever meter based on BP neural network

作　　者：陈洁[1] 焦振宇[1] 

机构地区：[1]江苏信息职业技术学院,无锡214153

出　　处：《电子测量技术》2009年第9期30-33,共4页Electronic Measurement Technology

摘　　要：分析了超声波速度在不同的传输介质、环境下的变化,以及由此导致的超声波回波测距时对液位测量的影响。实验得出超声波传播速度与温度、湿度是一个非线性函数,利用BP神经网络能很好地逼近非线性函数的特点,建立神经网络模型来逼近该函数,利用样本数据在Matlab环境下训练得出权值与阈值并保存留用。在应用系统编程时构造相同的BP网络,编写一个前向计算函数,即可获得系统当时环境下的超声波传播速度。应用该方法后,回波测距的精度提高了2到3个数量级。Analysis of Ultrasonic propagation speed in different media , environment changes, and the result of liquid level mesure being affected by Speed value when ultrasonic echo distance measurement. Experiments could educe conclusion that Ultrasonic propagation speed is a nonlinear function in relation to the temperature, humidity, however, based on The BP neural network can being a good approximation of the nonlinear function, establish neural network model to approximate the function, the sample data was trained by dint of the Matlab,weights and thresholds were obtained. The structure of applications is the same as the BP network which found, to compile a function which could calculate Ultrasonic propagation speed along, namely obtain Ultrasonic propagation speed at that time. Application of this method to ultrasonic echo distance measurement,the accuracy is up 2 to 3 grades.

关 键 词：BP神经网络 超声波回波测距 超声波传播速度 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]