基于COA-GRU的低成本气体传感器数据修正方法
Data Correction Method for Low-cost Gas Sensors Based on COA-GRU Algorithm
李炳伟 1叶树霞 1齐亮 1张永韡 1冯锦 1陈宇霆1
作者信息
摘要
针对低成本气体传感器在受到温度、湿度、压力、气体交叉干扰等影响时检测精度低的问题,提出了一种长鼻浣熊-门控循环单元神经网络(COA-GRU)的修正模型,用于提高传感器检测精度.首先,根据低成本传感器的非线性特性构建了GRU修正模型;其次,利用COA算法解决修正模型的多局部极值以及参数组合寻优问题;最后,利用低成本传感器组以及H200D气体检测装置的实测数据对该方法进行了仿真实验.结果表明,使用COA-GRU修正模型后,SO2、CO、NO2、CO2 传感器的平均绝对误差分别降低了72.0%、28.4%、29.6%、13.5%,能够有效提高低成本传感器的检测精度.
Abstract
This paper proposed a correction model based on coati optimization algorithm-gated recurrent unit(COA-GRU)to improve the sensor detection accuracy and to address the low detection accuracy issue of low-cost gas sensors caused by factors such as temperature,humidity,pressure,and gas cross-interference.Firstly,a GRU correction model was constructed based on the non-linear characteristics of low-cost sensors.Secondly,the COA algorithm was utilized to solve the problems of multiple local ex-trema and parameter combination optimization in the correction model.Finally,simulation experiments were conducted using real data obtained from a group of low-cost sensors and the H200D gas detection device.The results show that after using the COA-GRU correction model,the average absolute errors of SO2,CO,NO2,and CO2 sensors were reduced by 72.0%,28.4%,29.6%,and 13.5%,respectively,which can effectively improve the detection accuracy of low-cost sensors.
关键词
气体传感器/长鼻浣熊门控循环单元/修正模型/检测精度Key words
gas sensors/coati optimization algorithm-gated recurrent unit/corrected model/detection accuracy引用本文复制引用
基金项目
国家重点研发计划(51875270)
江苏省科技计划(BY2020031)
出版年
2024
NETL
NSTL
万方数据