改进BP网络在智能传感器设计中的应用  

Improvement of the application of BP network in intelligent sensor design

作　　者：张涛[1] 吕欣美[2] 

机构地区：[1]郑州轻工业学院计算机与通信工程学院,河南郑州450002 [2]中州大学就业指导中心,河南郑州450044

出　　处：《铜业工程》2010年第2期82-85,共4页Copper Engineering

摘　　要：传感器大多数都对环境(比如温度)有一定的敏感度,这样就会使传感器的零点和线性发生偏移,从而造成输出值随环境温度的变化而变化,再加上气压、以及气体流量等因素,导致测量出现附加误差。本文将利用神经网络来处理各种环境因素而产生的误差,将低成本的微控制器与传感器结合起来,设计出了能够自动补偿环境影响的智能传感器。并针对硬件平台的局限性,根据学习网络的学习特性,做了相应的优化改进,实现了传感器高精度快速误差补偿。Usually,environment temperature is not stable in the industrial field,and most of sensors have a certain sensitivity to temperature,which makes the zero and sensitivity of sensors to change,and so the sensors output changes with the environment temperature,therefore,temperature compensation is always key in the industrial measuring and control field.This paper introduces a new method of improving traditional gas sensor,where BP neural network is applies.BP neural network has advantages in nonlinear mapping,self-study,and generalization,which can construct double input single output net model by training a mass of sample data.High precision temperature compensation of sensors can be realized with this improved arithmetic.

关 键 词：BP网络 线性 智能 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]