基于GA⁃BP神经网络的氢气传感器的浓度补偿研究  

Concentration compensation research of hydrogen concentration sensor based on GA⁃BP neural network

作　　者：王雅坤 张宝林 王兆成 周传君[2] 郭仕佳 马琬雲 WANG Yakun;ZHANG Baolin;WANG Zhaocheng;ZHOU Chuanjun;GUO Shijia;MA Wanyun(School of Electronic and Information Engineering,Hebei University of Technology,Tianjin 300401,China;China Spacesat Co Ltd.,Beijing 100094,China)

机构地区：[1]河北工业大学电子信息工程学院,天津300401 [2]航天东方红卫星有限公司,北京100094

出　　处：《传感器与微系统》2024年第11期33-36,共4页Transducer and Microsystem Technologies

摘　　要：为解决环境因素致使氢气浓度传感器测量精度误差较大,导致氢燃料电池车辆因氢气泄漏检测不精确而产生爆炸风险的问题,提出了一种基于遗传算法反向传播(GA⁃BP)神经网络的氢气浓度传感器的浓度补偿方法。首先,利用BP神经网络对氢气浓度进行初步预测;然后,通过GA在寻优方面的优势进行浓度补偿,解决了BP神经网络局部陷入极值的问题。实验结果表明:基于GA⁃BP神经网络的氢气浓度传感器的浓度补偿方法对热导型氢气浓度传感器的预测准确度达到99.98%,最大相对误差值为0.2%,氢气浓度传感器的测量准确度提高了50.8%,为氢燃料汽车行业的发展奠定了基础。In order to solve the problem that environmental factors make the measurement precision of the hydrogen concentration sensor have a large error,which leads to the risk of explosion of hydrogen fuel cell vehicles due to inaccurate detection of hydrogen leakage,a concentration compensation method based on genetic algorithm back propagation(GA⁃BP)neural network for the hydrogen concentration sensor is proposed.Firstly,BP neural network is utilized to make preliminary prediction of hydrogen concentration.Secondly,concentration compensation is carried out through the advantage of GA in optimizing,which solves the problem of the BP neural network locally falling into the extreme value.The experimental results show that the concentration compensation method of hydrogen concentration sensor based on GA⁃BP neural network has a prediction accuracy of 99.98%for the thermally conductive hydrogen concentration sensor,and the maximum relative error value is 0.2%,and the measurement accuracy of the hydrogen concentration sensor is improved by 50.8%,which lays a foundation for the development of the hydrogen⁃fueled automobile industry.

关 键 词：氢气浓度传感器 反向传播神经网络 遗传算法 遗传算法—反向传播神经网络 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]