摘要
提出一种基于最小二乘支持向量机(LS-SVM)的传感器非线性校正及温度补偿的新方法,并给出了相应的过程和算法。在该方法中,LS—SVM被用作构建逆模型,并通过该模型映射传感器非线性特性,同时实现了传感器的温度补偿和非线性校正。通过实际电容式压力传感器校正的实验结果表明:所提模型建模速度比SVM模型高1-2个数量级,补偿误差仅为SVM模型的20%左右。因此,该学习速度快、补偿精度高、抗噪声干扰能力强,适合传感器温度补偿及校正。
A novel method for temperature compensation and nonlinear correction of transducer based on least squares support vector machine (LS-SVM) is presented. The design steps and algorithm are also addressed. LS-SVM is used to build as an inverse model,by which the transducer nonlinear characteristic is mapped. The temperature compensation and nonlinear correction are realized synchronously. The experiment results of applying in the capacitor pressure sensor(CPS) show that the speed of this LS-SVM building model is 1 -2 order of magnitude,while the compensation errors is 20 % of SVM model. As a result, the method is faster in learning speed, high in accuracy and robust in noise resistance, it is more suitable for sensor's correction and temperature compensation.
出处
《传感器与微系统》
CSCD
北大核心
2007年第3期76-79,共4页
Transducer and Microsystem Technologies
关键词
最小二乘支持向量机
回归
传感器
温度补偿
校正
least squares support vector machine ( LS-SVM )
regression
sensor
temperature compensation
correction