基于BP神经网络的铂铑热电偶的非线性校正

The Nonlinearity Correction of Platinum-rhodium Thermocouple Based on BP Neural Network

热电偶是工业中广泛使用的测温元件,由于其自身的物理特性的限制,其输出热电势与被测温度之间存在着一定的非线性,这将增大温度的测量误差,从而影响测量精度。针对这一问题,提出了基于BP神经网络的非线性校正算法,以热电偶的热电动势为输入信号,以与其相对应的温度值作为输出信号,给出了前馈型BP神经网络的结构和训练权值的方法。通过实验仿真结果对比可知,此方法降低了热电偶测温的非线性误差,便于操作,大大提高了铂铑热电偶在温度测量中的测量精度。

Thermocouple temperature measurement device,which is widely used in industry,has a certain nonlinear re-lation between the output thermal potential and the measured temperature due to its physical characteristics, which will increase the temperature measurement error and affect the measurement accuracy. In order to solve this problem,a non-linear correction algorithm was proposed based on Back Propagation (BP) neural network with the thermoelectromotive force as the input signal and the corresponding temperature values as the output signal, and the structure of the feed forward BP neural network and methods of training weights can be got. By comparison with the simulation,the results show that the nonlinear error of thermocouple temperature measurement is reduced. It is relatively simple to implement and the temperature measurement accuracy of platinum-rhodium thermocouple is greatly enhanced.