摘要
体积分数检测中,常常存在时间响应曲线出现延迟或提前的情况,传统方法(如神经网络)无法进行矫正,继而影响检测的准确性。通过对样本的时间响应曲线的观察分析,发现该曲线的最大曲率点与样本体积分数存在某种联系。基于此,提出了一种体积分数定量分析模型。通过对16组不同体积分数的NH3和丙酮的实验,发现使用该模型对时间响应曲线进行矫正后,测量结果的一致性得到很大的提高。当变异系数分别大于0.17,0.5时,检测结果一致性改善的比例分别可达87.5%,88.9%,进一步证明了该方法的有效性。
In volume fraction detection, time response curve often occur time delay or advance case, traditional methods ( e. g. neural networks) cannot correct it, thereby affect accuracy of detection. Based on observation and analysis on time response curve, !t is found that there exists a relation between the maximum curvature point of the curve and volume fraction of sample. Based on this, propose a quantitative volume fraction analysis model for gas. By experiment of 16 groups different volume fraction of ammonia and acetone respectively, it is found that after using this model to correct time response curve is improved consistency of the measurement results greatly. When the coefficient of variation are greater than 0.17 and 0.5 respectively, the corresponding improvement percentage of test results consistency are 87.5 % and 88. 9 % , which further verifies the effectiveness of this method.
出处
《传感器与微系统》
CSCD
北大核心
2014年第1期31-34,共4页
Transducer and Microsystem Technologies
基金
国家科技支撑计划资助项目(2012BAI19B03)
国家自然科学基金资助项目(81271930
81171414)
教育部高校博士点基金资助项目(20090191110030)
中央高校基本科研业务费资助项目(CDJXS10231179
CDJSX102300)
关键词
化学传感器阵列
气体检测
定量分析
时间响应曲线
chemical sensor array
gas detection
quantitative analysis
time response curve