基于RBF神经网络和LS-SVM组合模型的磁浮车间隙传感器温度补偿

对磁浮车悬浮间隙传感器温度漂移产生机理进行了研究,提出采用组合预测的方法建立传感器温度特性逆模型进行温度补偿,根据传感器温度特性分别建立径向基函数神经网络(RBF-NN)和最小二乘支持向量机(LS-SVM)温度补偿系统模型,通过在探头内布置PT1000铂热电阻检测探头温度,依据温度信号对传感器进行温度误差补偿。仿真结果表明组合模型能较好地拟合温度逆特性,组合补偿模型的输出不受工作温度的影响,全量程最大误差为0.14 mm,在工作间隙范围内误差小于0.05 mm,且组合模型的补偿误差优于单一模型补偿效果,该方法可有效消除温度漂移效应,并提高传感器的检测准确度,能够满足磁浮车悬浮控制系统要求。

Temperature Compensation of Maglev Train Gap Sensor Based on RBF Neural Network and LS-SVM Combined Model

The temperature drift mechanism of the maglev train gap sensor is analyzed and a method is proposed to solve the temperature drift problem. In this method,the combined model of the temperature inverse characteristic is designed to compensate the temperature drift error. The radial basis function neural network ( RBF-NN)and the least squares support vector machine( LS-SVM)combined temperature compensation model is established with the temperature characteristic of the gap sensor. A PT1000 temperature sensor is embedded in the probe in order to provide the reference temperature. The combined model compensates the temperature drift error of the gap senor according to the temperature signal. The simulation results show that the inverse temperature characteristic can be fitted well by the combined model. The output of the compensator is independent of the tooth-groove position. The simulation studies show that this compensator can provide correct gap data with the error less than 0. 14 mm in the full scale and less than 0. 05 mm in the normal work gap. The precision of the combined model is better than that of any single model. The precision of the sensor is increased with this method and the compensated output of the gap sensor may meet the requirement of levitation control system.