基于主成分分析和深度森林算法的S700K转辙机故障诊断

PCA-gcForest-based Fault Diagnosis of S700K Switch Machine

摘要:

针对目前转辙机故障诊断准确性不高、效率低等问题，提出了一种基于主成分分析（PCA）和深度森林（gcForest）算法的故障诊断方法。对于S700K转辙机11种故障模式下的电流、功率曲线，采用主成分分析进行电流特征值特征简约，然后使用嵌入简约特征值的改进深度森林模型提高数据处理能力，增强模型内在特征代表性。结果表明，改进深度森林模型故障诊断准确率为97.62%，验证了该方法的有效性和优越性。

Abstract:

To overcome the shortage of the existing fault diagnosis methods such as low accuracy and efficiency， a fault diagnosis method based on principal component analysis （PCA） and multi-grain cascade forest （gcForest） algorithm was proposed. PCA was used to simplify the current eigenvalue for 11 fault modes of S700K switch machine. And an improved gcForest model with the simpler eigenvalue embedded was used to improve the data processing capability and enhance the inherent feature representativeness of the model. The experimental results show that the fault diagnosis accuracy of the improved gcForest model is 97.62%， which verifies the effectiveness and superiority of the method.