串联双卡尔曼滤波器在PCR非接触测温中的应用

Sequential dual Kalman filtering approach for non-contact temperature measurement in real-time PCR instrument

传统PCR仪对试剂样品进行接触式测温时,存在测量迟滞大、无法直接测量试剂温度以及只能测量局部点温度的缺点,采用红外温度传感器对样品进行非接触测量可以克服以上局限。但红外温度传感器测量噪声较大,为此提出一种串联式双卡尔曼滤波算法来对红外温度传感器采集到的温度数据进行实时滤波。滤波过程中迭代卡尔曼滤波器(IEKF)与线性卡尔曼滤波器(KF)顺序运行,前者利用少量观测数据对热交换模型中的参数进行在线辨识,后者利用辨识后的系统对采样数据进行实时滤波,该算法结合了IEKF非线性估计收敛快与KF计算简单、实时性高的优点。最后建立平滑度、偏差度和准确度指标来评价该算法与KF、EKF、FIR数字滤波器的滤波效果。实验表明,该算法与其他滤波器输出具有相当的平滑度,且偏差度为0.510 6,远小于其他滤波器,准确度指标也由滤波估计前6.871 1提高到0.150 2。

Traditionally,PCR instrument detects the temperature of reagent by contact measurement,which has several limitations such as thermal hysteresis,reagent contamination and point-measurement.Thus the infra red thermometer is applied to non-contact measure the temperature of reagent.But the infra red thermometer has the defect of relatively high measurement noise.Therefore,we propose a sequential dual Kalman Filters（Seq-DKF）,which consists of an Iterated Extended Kalman Filter（IEKF） and a linear Kalman Filter（KF）,to remove the measurement noise from infra red thermometer signals in real-time.In this algorithm,IEKF and KF executes sequentially.The former filter rapidly identifies the parameter of heat exchange model by using a small number of observations.The latter filters the temperature signal based on the identified model.After specifying the heat exchange model and the Seq-DKF,we apply,the algorithm to the PCR instrument for testing.Finally,The other three filters are employed for comparison to demonstrate the effectiveness of the Seq-DKF algorithm.With the Seq-DKF estimation,the accuracy of temperature is enhanced to 0.150 2 from 6.871 1.