基于互相关和函数型神经网络测量声波渡越时间  

Acoustic Transit Time Measurement Based on Cross-correlation and Functional Neural Network

作　　者：吴新杰[1] 吴成东[2] 

机构地区：[1]辽宁大学物理学院,辽宁沈阳110036 [2]东北大学信息科学与工程学院,辽宁沈阳110004

出　　处：《仪表技术与传感器》2010年第11期82-84,共3页Instrument Technique and Sensor

基　　金：教育部高等学校青年骨干教师资助项目;华北电力大学电站设备状态监测与控制教育部重点实验室开放基金资助项目(2008-010)

摘　　要：由于硬件电路的影响,采样频率的提高受到了限制,直接对发射端和接收端的采样信号进行互相关所确定的声波渡越时间精度较低,为此提出一种基于互相关和函数型神经网络测量声波渡越时间的方法。首先介绍声学法测温、互相关和神经网络的基本原理,该方法是利用传统方法得到的互相关函数在峰值附近的函数值来训练函数神经网络,然后利用训练好的神经网络进行插值运算,从而得到细化后的互相关函数局部曲线,再对该曲线进行寻峰从而得到声波的渡越时间。最后通过仿真实验验证了这种方法的有效性。Because of the impact of hardware circuit,the relatively high sampling frequency has been restricted.The measurement accuracy of acoustic transit time by directly cross-correlation of the transmitted and received sampling signals is relatively low.In order to improve acoustic transit time measurement accuracy,a new measurement method based on cross-correlation and functional neural network was proposed.The basic principle of acoustic temperature measurement,cross-correlation method and functional neural network was concisely introduced.A functional neural network was trained by the function values near the peak of cross-correlation curve obtained by traditional method.And then the trained neural network was utilized to finish interpolation operation,so a new local cross-correlation function curve was gained,from which the peak of cross-correlation was found out,thus acoustic transit time was calculated.Finally the effectiveness of this method has been verified by the simulation experiments.

关 键 词：声学法测温 函数型神经网络 互相关 渡越时间 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]