基于SVM回归的GMI磁传感器信号处理方法

Signal Processing of GMI Magnetic Sensor Based on SVM Regression

为克服典型非晶丝材料的巨磁阻抗(giant magneto impedance,GMI)效应的非线性特性所导致的局限性问题,提出了一种基于支持向量机(support vector machine,SVM)回归的GMI磁传感器多参数数据处理方法,利用支持向量机SVM作为识别工具,以敏感材料的的阻抗模值和阻抗角信息作为磁场识别参数,将被测磁场强度值作为输出参数,进行SVM模型建立和性能验证。结果表明:该方法能很好地克服敏感材料的非线性特性的影响,处理误差在?0.007Oe以内。

In order to overcome the limitations of the nonlinear properties of giant magneto-impedance （GMI） effects of typical amorphous wire materials. A multi-parameter data processing method of GMI magnetic sensor based on support vector machine （SVM） regression is proposed. Using SVM as a recognition tool, the impedance modulus and impedance phase information of sensitive materials are used as magnetic field identification parameter, the measured magnetic field strength value is taken as the output parameter, and then the SVM model is established and the performance is verified. The results show that the method can overcome the influence of the nonlinearity of the sensitive material, and the processing error is within ？0.007 Oe.