多维传感器标定的支持向量机复合式方法

Support vector machine complex method for multi-dimensional sensor calibration

提出了多维传感器标定的支持向量机复合式标定方法,解决了多维传感器标定中的参数不确定性和耦合非线性因素.对一个多维传感器进行了支持向量机复合标定的仿真研究,将这一方法应用到六维力传感器测量系统的标定试验中.结果表明,与传统标定方法相比,支持向量机复合标定的方法在不增加数据样本的情况下,能够显著提高传感器标定的精度;同时也比纯支持向量机标定有更强的泛化能力,复合标定的模型还可以为传感器的设计提供依据.

A support vector machine （SVM） complex calibration approach was proposed to solve the parameters uncertainty and coupling nonlinearity. The simulation of the calibra- tion of a multi-dimensional sensor was conducted by the SVM complex calibration method, which was then utilized to calibrate a six-dimensional force sensor system. It can be seen from the results that, as compared with other traditional calibration methods, the SVM complex calibration method can improve the calibration accuracy significantly, without in- creasing data samples; additionally, the complex calibration method also has better general- ization performance in comparison with the SVM black box modeling, indicating that it can provide a basis for the design of sensors.