基于PSO-SVM的回转窑筒体故障诊断

Fault Diagnosis of Rotary Kiln Shell Based on PSO-SVM

为提高回转窑筒体的故障诊断效率,提出将改进的集合经验模式分解(MEEMD)与参数优化支持向量机(SVM)结合的故障诊断方法。采用MEEMD对回转窑托轮挠度信号进行分解和重构,选取原信号分解得到的KS、KR谐波幅值和重构信号的时域特征构造特征向量输入到支持向量机诊断模型中;针对人工选取支持向量机惩罚因子和核参数可能导致的准确性差异,采用粒子群算法对支持向量机的惩罚因子和核参数进行寻优;最后采用寻优后的结果建立支持向量机故障诊断模型对样本数据进行故障诊断分类,实验结果表明该方法不仅可以更加清晰地获取回转窑筒体的故障信息,而且能实现更高精度的故障诊断。

To improve the fault diagnosis efficiency forrotary kiln,this paper proposes a fault diagnosis method combining improved empirical mode decomposition(MEEMD)and support vector machine(SVM).MEEMD was used to decompose and reconstruct the deflection signal of the rotary kiln roller.The eigenvectors which are composed of KS and KR harmonic amplitudes are obtained from the original signal decomposition.Along with the time-domain characteristics of the reconstructed signal,they were input into the support vector machine diagnostic model.Aiming at reducing the possible accuracy defects resulted from manual selection,PSO was used to select the optimal SVM penalty factors and kernel parameters.Finally,the optimum support vector machine fault diagnosis model is able to diagnose and classify the sample data.The experimental results show that the method proposed in this paper not only can obtain the fault information of the rotary kiln barrel clearly but also achieve fault diagnosis with optimum-precision.