非线性传感特性的神经网络学习校正方法及应用

非线性误差是传感器的主要误差之一 ,用硬件补偿或用查表、曲线拟合等方法复杂且精度低。研究了一种神经网络校正方法 ,并提出了一种改进的小脑模型神经网络 ,仿真和应用表明 ,可使非线性误差减小到近似为零。

A Method for Learning and Correcting the Sensor's Nonlinear Characteristic Using a Neural Network

The nonlinear error is one of the sensor's major errors. The methods compensated by hardware or table look up, curve fitting and so on are complicate and the accuracy is lower. A method of correcting the sensor's nonlinear error using a neural network is shown, and a improved CMAC neural network is proposed. The results of simulation experiment and application show that the nonlinear error of the sensor system is decreased approximately to zero after correcting with this method.