基于RBFNN的铂电阻温度传感器非线性补偿

Nonlinearity compensation of platinum resistor temperature sensor based on RBFNN

针对铂电阻温度传感器应用中存在的非线性问题,提出了应用径向基函数神经网络(RBFNN)强非线性逼近能力进行铂电阻温度传感器非线性补偿的方法。介绍了非线性补偿的原理和网络训练方法。结果表明:这种非线性补偿模型具有误差小、精度高、可在线标定和鲁棒性强等优点,与基于BP神经网络的非线性补偿模型相比,大大缩短了网络训练时间,从而方便了铂电阻温度传感器在测控系统中的应用。

A method to compensate nonlinearity of platinum resistor temperature sensor is presented using nonlinearity compensation model founded by RBFNN to aim at its nonlinear problem. The principle of nonlinearity compensation and training method of neural network are introduced. The results show that nonlinearity compensation model has character of small error, high precision, on-line scaling, strong robustness and_fast network training speed compared with the compensating method of BP neural network model. It makes convenient for platinum resistor to be applied in the field of measurement and control.