论文部分内容阅读
水电机组振动故障诊断中常面临样本稀缺及分布不均匀、不平衡等问题,严重影响诊断结果。针对此类问题提出一种基于模糊K近邻(K nearest neighbor,KNN)支持向量数据描述(support vector data description,SVDD)的故障诊断模型。首先利用核变换将故障样本映射到高维特征空间,并采用SVDD提取不平衡故障样本域的边界支持向量样本,构建基于相对距离模糊阈值和KNN的决策规则,最终在此基础上建立机组故障诊断模型。用该模型对经过不平衡处理的国际标准测试数据样本进行测试实验,并与支持向量机(support vector machine,SVM)及目前应用较多的SVDD模型的分类结果进行对比,结果表明该模型可有效解决不平衡样本分类倾斜性问题。最后,将模型用于某水电厂机组振动故障诊断,取得了较高的诊断精度,证明了该方法的有效性。
Vibration fault diagnosis of hydropower units often face the scarcity of samples and uneven distribution, imbalance and other issues, seriously affecting the diagnostic results. Aiming at these problems, a fault diagnosis model based on K nearest neighbor (KNN) support vector data description (SVDD) is proposed. Firstly, the fault samples are mapped to high-dimensional feature space by using kernel transform, and SVDD is used to extract the boundary support vector samples of the unbalanced fault samples domain to construct decision rules based on relative distance fuzzy thresholds and KNN. Finally, based on this, model. The model is used to test the unbalanced international standard test data samples and compared with the support vector machine (SVM) and the classification results of the SVDD models that are currently used more and more. The results show that the model is effective Solve the problem of unbalanced sample classification tilt. Finally, the model is used to diagnose the vibration of a hydropower plant vibration and a high diagnostic precision is obtained, which proves the effectiveness of the method.