基于BP算法的传感器单气体定量识别  

Single Gas Quantitative Identification of Sensors Based on BP Algorithm

作　　者：于明岩[1] 施云波[1] 赵文杰[1] 

机构地区：[1]哈尔滨理工大学测控技术与通信工程学院测控技术与仪器黑龙江省高校重点实验室,黑龙江哈尔滨150080

出　　处：《仪表技术与传感器》2011年第8期17-20,90,共5页Instrument Technique and Sensor

基　　金：国家自然基金(60772019);哈尔滨市科技创新人才研究专项资金(2010RFXXG009);黑龙江省教育厅科学技术研究项目(11541047)

摘　　要：气体的定性识别与定量分析一直是气体检测领域的主要难点,也是气体传感器应用中的研究热点。文中在对多个气体传感器进行气体定性分析和实时数据采集基础上,通过BP网络分析方法,实现了气体的定量识别,然后对训练前后气体浓度的误差进行对比分析,仿真结果表明:平均相对误差均控制在1%以内,最大相对误差控制在2%以内,提高了对气体浓度的预测精度。Gas qualitative identification and quantitative analysis are always the main difficulties in the field of gas detection,and also the research hotspot in gas sensors application.Based on several gas sensors for gas qualitative analysis and the real-time data acquisition,this paper used BP network analysis method,realized the gas quantitative identification,and then wade comparative analysis of the error of the training pre-and post gas concentration data.The simulation results show that the average relative errors are controlled within 1%,the maximum relative errors are controlled within 2% and improve the prediction precision of the gas concentration.

关 键 词：气体传感器 BP网络 定量分析 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]