摘要
鉴于相同条件下传感器的输出特性本质上服从某一未知分布及统计学习理论中支持向量机方法解决非线性问题的能力,提出了一种通过学习机构造出反映传感器输出特性的回归函数进行动态补偿的方法。该方法无需被补偿传感器结构特性的先验知识,且提高了泛化能力。实验表明:补偿后的传感器具有期望的输出特性。
Whereas that output character of sensor submits to some kind of unaware distribution essentially in the same conditions and the support vector machine method based on the statistical learning theory has the ability to solve nonlinear problems. A new approach that uses the regressing function constructed by learning machine and reflectin~ the output character of the sensor to dynamically compensate the output of sensor is put forward. Not having known the configurable character of sensor forward, using the method is available and extends its application scope. The experiments show that the compensated sensor has the anticipant output character.
出处
《传感器技术》
CSCD
北大核心
2005年第12期19-21,25,共4页
Journal of Transducer Technology
关键词
支持向量机
粗糙集
动态补偿
输出非线性
support vector machine（SVM）
rough sets
dynamic compensation
output nonlinear
