LS-SVM技术在曳引机故障预测中的应用

The application of LS-SVM method in traction machine fault prediction

利用支持向量机采用的结构风险最优化准则、预测能力强、鲁棒性好等优点,研究了最小二乘支持向量机(Least Squares Support Vector Machines,LS-SVM)回归算法在曳引机故障预测中的应用。提出了一种自动搜寻最优参数方法,对参数和进行寻优,避免了人工选择的盲目性,提高了算法的效率。通过将LS-SVM和RBF神经网络进行对比实验,得出在相同训练样本条件下,LS-SVM可以取得比RBF更好的预测精度和预测速度,更加适合于现场实际应用。最后将LS-SVM模型用于曳引机振动信号的时域分量预测中,预测的平均相对误差小于5%,取得了较高的预测精度。

Made use of the advantages of support vector machines that structural risk optimization criteria,strong predictability and good robust,it adopt Least Squares Support Vector Machines（LS-SVM ）to research Traction Machine fault prediction. Considering the inefficiency of manual selection,we adopt searching optimal parameters automatically to find optimize parameter and. An contrast test has been made which proves LS-SVM can have a good performance with the same training samples and get a better result than RBF neural network in forecasting precision and speed. So it is suitable for the worksite application. Finally,this method is used to predict time field components of traction machine vibration signals and result shows that the average relative errors are all less than5%,which have an accurate result.