LS-SVM算法在无缝线路锁定轨温检测中的应用

提出一种改进的基于磁巴克豪森噪声（MBN）技术的无缝线路实际锁定轨温检测方法。该方法使用MBN技术检测钢轨温度应力，使用轨温计检测表面轨温，并采用最小二乘支持向量机（LS-SVM）算法建立线路的钢轨温度应力随钢轨表面轨温与实际锁定轨温差值变化的预测模型，最后将预测模型得到的温差和钢轨的表面轨温用于计算实际锁定轨温。通过对河北保定的一段无缝线路进行钢轨温度应力及表面轨温检测，证明该方法能提高检测的精度。

Application of LS-SVM algorithm in detecting the stress-free temperature of continuous welded rail

Magnetic Barkhausen noise (MBN) technology has been used to detect the actual stress-free temperature of continuous welded rail. The method uses MBN technology to detect the rail temperature stress, and uses the rail thermometer to detect the surface rail temperature. Using the least squares support vector machine (LS-SVM) algorithm, a predictive model has been established to predict the relationship between the rail temperature stress and the difference between the surface rail temperature and the actual stress-free temperature. Finally, the temperature difference obtained by the predictive model and the surface rail temperature of the rail have been used to calculate the actual stress-free temperature. The rail temperature stress and surface rail temperature have been tested on a section of continuous welded rail in Baoding, Hebei Province, and the results show that the method improves the accuracy of detection.