热电偶级联系统动态性能评估方法研究

Study on dynamic characteristics evaluation method of the cascade thermocouple system

为了在缺乏激励信号先验知识的情况下实现热电偶动态性能的精确评估,提出一种由3只不同偶丝直径热电偶组成的级联式温度测试系统的时间常数评估方法。该方法基于总体最小二乘Tikhonov正则化算法对热电偶级联系统数学模型进行辨识及时间常数解算。对于热电偶偶丝直径较为接近的测试系统,基于奇异值分解进行正则化方法优化,可以避免热电偶信息矩阵共线性造成的时间常数病态估计,有效拓展了该方法的适用性。通过数值仿真进行了理论验证,并在一种气体温度动态校准装置中进行了实验验证,结果表明奇异值分解正则化时间常数估计方法相比总体最小二乘Tikhonov正则化算法和普通最小二乘算法,辨识精度分别至少提高了75%和3.5倍,证明了该方法的有效性。

To realize the precise evaluation of dynamic characteristics of the thermocouple in the circumstance of lacking the prior knowledge of excitation temperature signal, an evaluation method for the time constant of the temperature measurement system is proposed, which utilizes three thermocouples with different wire diameters. Based on the Tikhonov regularization method, the mathematical model can be identified and the time constant can be calculated. For the cascade system with a consistent thermocouple wire diameter, an optimized regularization method with the singular value decomposition is proposed. The ill-conditioned estimation can be avoided, which is due to the collinearity of the thermocouple information matrix. In this way, the validity of the proposed method is extended effectively. It is verified by the numerical simulation in the perspective of theory. Experiments are implemented with a gas temperature dynamic calibration equipment. Compared with the total least square Tikhonov regularization algorithm and the least squares algorithm, experimental results show that the identification precision of the optimized regularization method of the time constant estimation with the singular value decomposition is improved 75% and 3.5 times at least. The validity of the proposed method is verified.