【作者】 高嵩； 唐耀庚； 李兰君；

【机构】 南华大学电气工程学院；

【摘要】 曲线拟合是传感器非线性校正的重要方法,但对于输出特性具有明显分段性、各段特性差异较大的传感器,采用单一模型难以取得满意的拟合结果。提出基于多支持向量回归机(MSVR)的曲线拟合方法,将输入样本空间分割成多个子空间,为每一子空间建立一个支持向量回归机(SVR)来映射该局部空间的非线性关系,合成各SVR的输出即为传感器全量程输出特性。将所提出的方法用于热偶规真空传感器输出特性拟合,仿真结果表明了该方法的有效性。更多还原

【Abstract】 It is an important way of nonlinearity correction for sensors by fitting curves to data,but it is difficult to fit curve with only one model for those whose output characteristic is complex and composed of different parts. A navel method of curve fitting based on Multiple Support Vector Regression (MSVR) was proposed. The input samples were split into some sub-spaces,mapping relation was set to each sub-space by using a Support Vector Regression (SVR) which may has parameters different from others. The output of each SVR is the output characteristic of corresponding area,the sum of all SVR is the total output characteristic of the sensor. The proposed approach was applied to fit the output characteristic curve for the thermocouple vacuum sensor,the simulation results show that this approach is valid.更多还原