基于WKFCM⁃SMOTE和随机森林的风电机组故障诊断

Wind Turbine Fault Diagnosis Based on WKFCM⁃SMOTE and Random Forest

针对风电机组运行数据中故障数据占比小,不平衡数据集影响故障诊断精度与诊断结果的问题,提出一种基于加权模糊核C均值(WeightedKernelFuzzyC⁃means,WKFCM)算法改进的合成少数类过采样技术(SyntheticMinorityOvers⁃amplingTechnique,SMOTE)算法,结合随机森林算法实现风电机组故障诊断。使用随机森林的袋外误差进行数据特征排序和选取,采用WKFCM⁃SMOTE算法进行故障数据集扩充,基于随机森林算法搭建故障诊断模型,并对模型参数进行网格搜索优化。试验结果表明,基于该模型的风电机组故障诊断比传统方法准确率更高。

The proportion of fault data in wind turbine operation data is small,and the unbalanced data set affects the fault diagno-sis accuracy and diagnosis results.An improved synthetic minority oversampling technique(SMOTE)algorithm based on weigh-ted kernel fuzzy C-means(WKFCM)algorithm was proposed,which is combined with random forest algorithm to realize wind tur-bine fault diagnosis.The out-of-bag error of random forest was used for data feature sorting and selection.The WKFCM-SMOTE was used to expand the fault data set.The fault diagnosis model was built based on random forest algorithm,and the model parame-ters were optimized by grid search.The test results show that the fault diagnosis of wind turbine based on proposed model is more accurate than the traditional method.