An AR model with high-order cumulant was built by the vibration signals which were collected during the screening of the self-synchronization probability sieve. A screen efficiency predicting model was established by using the coefficients of

the AR model as the input variables of a LS-SVM model.The real-time monitoring

was made during the screening of sieve by the way combining AR model with LS-SVM.

利用采集的概率筛筛分作业时主筛体的振动信号，建立高阶累积量AR模型。将AR模型系数作为LS-SVM模型的输入变量，建立LS-SVM筛分效率预测模型。将AR模型与LS-SVM结合实现对概率筛筛分性能的实时监测。

概率筛, 筛分效率, 高阶累积量, AR模型, 最小二乘支持向量机