西林瓶内氧气含量在线检测的基线快速校正方法

Fast Baseline Correction Method for Online Detection of Oxygen Content in Vials

构建了一个开放单光路短光程激光检测平台,提出了一种基于线型的Levenberg-Marquardt(L-M)算法,以洛伦兹线型和高斯线型的二阶导数作为L-M算法的拟合函数,对氧气体积分数为0%～21%的样本进行校正。结果表明:在基于洛伦兹线型的L-M算法校正中,氧气含量-二次谐波峰值的拟合系数最高,为0.9995;以洛伦兹线型为优化依据,对含氧气体积分数为1%的样品进行了多次测量,校正处理前后预测的氧气体积分数最大偏差分别为0.38%和0.22%,预测的均方根误差分别为0.25%和0.16%。

A laser detection platform with an open single path and short optical distance is constructed, and a Levenberg-Marquardt(L-M) algorithm based on linetype is proposed. The method uses second order derivative of Gaussian linetype and Lorenz linetype as the fitting function of L-M algorithm. The samples with oxygen volume fraction of 0%-21% are corrected. The results show that the correlation coefficients of oxygen content and peak is the highest in L-M algorithm correction based on Lorentz linetype(0.9995). The sample with oxygen volume fraction of 1% is measured several times on the basis of Lorentz linetype. Compared with the measured results before and after the correction, the predicted maximum deviations of oxygen volume fraction are 0.38% and 0.22%, respectively, and the root mean square errors of prediction are 0.25% and 0.16%, respectively.