基于OBF神经网络的温度传感器非线性补偿方法

Approach to Nonlinear Compensation of Temperature Transducer Based on OBF Neural Network

针对热敏电阻温度传感器应用中存在的非线性问题,提出应用递推最小二乘法训练正交基(OBF)神经网络进行非线性补偿的方法。研究结果表明,与RBF神经网络非线性补偿模型和BP神经网络非线性补偿模型相比,该正交基神经网络非线性补偿模型具有误差小,精度高,训练次数少的优点,故为一种有效的非线性补偿方法,在测控领域具有实用价值。

This paper presents a method to compensate nonlinear of thermistor temperature transducer using Recursive Least Square（RLS） algorithm to training nonlinear compensation model based on Orthogonal Basis Function（OBF） neural network,according to the nonlinear problem existed in the thermistor temperature transducer. The results show that the proposed method has high precision and fast network training speed compared with the compensation methods based on RBF neural network model and BP neural network model. The nonlinear compensation approach is effective and practical in measurement and control system.