基于随机RF的集成SVM故障诊断改进算法
|
摘要
针对工业系统监控数据不均衡导致的故障状态难以被识别问题,提出一种基于随机旋转森林的集成支持向量机(RRFESVM)故障诊断算法,通过将监控数据进行属性随机分割、组合、PCA变换和样本有放回重采样,组建多个新训练子集并使用支持向量机算法进行训练,得到多个支持向量机故障诊断基分类器,集成得到强分类器,由此既保证基分类器之间的差异性,又保证故障诊断精度和分类器性能稳定性,从而解决故障诊断易偏置问题,提高作为少数类的故障状态实时诊断准确率。亚轨道飞行器再入过程实验与TE化工过程实验都表明RRFESVM故障诊断算法能够有效提升不均衡数据情况下的实时故障诊断性能。
The system failure states are hard to be identified with imbalanced monitoring dataset.A RRFESVM(Random Rotation Forest based Ensemble SVMs) fault diagnosis algorithm was proposed.After the attributes of monitoring data were segmented and combined randomly,the resample step was implemented and Principle Component Analysis(PCA) was used for feature transformation.Then a number of new training subsets were obtained and Support Vector Machine(SVM) fault diagnosis sub-classifiers can be trained with these new training subsets accordingly.Eventually a voting-integrated strong RRFESVM fault diagnosis classifier was developed.In this way,the diversity of sub-classifiers,the fault diagnosis accuracy improvement and the classifier performance stability requirement can be achieved.The experimental results of suborbital reusable launch vehicle return course and Tenessee Eastman(TE) chemical process show that the proposed RRFESVM fault diagnosis algorithm can improve the diagnosis performance on the imbalanced dataset.
The system failure states are hard to be identified with imbalanced monitoring dataset.A RRFESVM(Random Rotation Forest based Ensemble SVMs) fault diagnosis algorithm was proposed.After the attributes of monitoring data were segmented and combined randomly,the resample step was implemented and Principle Component Analysis(PCA) was used for feature transformation.Then a number of new training subsets were obtained and Support Vector Machine(SVM) fault diagnosis sub-classifiers can be trained with these new training subsets accordingly.Eventually a voting-integrated strong RRFESVM fault diagnosis classifier was developed.In this way,the diversity of sub-classifiers,the fault diagnosis accuracy improvement and the classifier performance stability requirement can be achieved.The experimental results of suborbital reusable launch vehicle return course and Tenessee Eastman(TE) chemical process show that the proposed RRFESVM fault diagnosis algorithm can improve the diagnosis performance on the imbalanced dataset.
引文
[1] Schapire R E.The strength of weak learn ability[J].Machine learning,1990,5(2):197-227.
[2] 邵良杉,马寒.基于旋转森林的分类器集成算法研究[J].计算机工程与应用,2015,51(23):149-154.
[3] Wang X Z,Xing H J,Li Y,et al.A Study on Relationship Between Generalization Abilities and Fuzziness of Base Classifiers in EnsembleLearning[J].IEEE Transactions on Fuzzy Systems,2015,23(5):1638-1654.
[4] Wang B Y,Pineau J.Online bagging and boosting for imbalanced data streams[J].IEEE Transactions on Knowledge and Data Engineering,2016,28(12):3353-3366.
[5] Fu Q,Jing B,He J P,et al.Fault Feature Selection and Diagnosis of Rolling Bearings Based on EEMD and Optimized Elman_AdaBoost Algorithm[J].IEEE Sensors Journal,2018,18(12):5024-5034.
[6] 周玉臣,方可,马萍,等.基于集成学习的复杂仿真模型验证方法[J].系统工程与电子技术,2018,40.
[7] Rodriguez JJ,Kuncheva L I.Rotation forest:a new classifier ensemble method[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2006,28(10):1619-1630.
[8] 王亚辉,李新国,王谦.亚轨道飞行器再入过程执行机构故障仿真[J].飞行力学,2014,32(5):446-449.
[9] Andreas B,Lawrence R,Mohieddine J.Revision of the Tennessee Eastman Process model[J].IFAC-Paper On Line,2015,48(8):309-314.
[10] 王鲜芳,王岁花,杜昊泽,等.基于模糊粗糙集和支持向量机的化工过程故障诊断[J].控制与决策,2015,30(2):353-356.
[2] 邵良杉,马寒.基于旋转森林的分类器集成算法研究[J].计算机工程与应用,2015,51(23):149-154.
[3] Wang X Z,Xing H J,Li Y,et al.A Study on Relationship Between Generalization Abilities and Fuzziness of Base Classifiers in EnsembleLearning[J].IEEE Transactions on Fuzzy Systems,2015,23(5):1638-1654.
[4] Wang B Y,Pineau J.Online bagging and boosting for imbalanced data streams[J].IEEE Transactions on Knowledge and Data Engineering,2016,28(12):3353-3366.
[5] Fu Q,Jing B,He J P,et al.Fault Feature Selection and Diagnosis of Rolling Bearings Based on EEMD and Optimized Elman_AdaBoost Algorithm[J].IEEE Sensors Journal,2018,18(12):5024-5034.
[6] 周玉臣,方可,马萍,等.基于集成学习的复杂仿真模型验证方法[J].系统工程与电子技术,2018,40.
[7] Rodriguez JJ,Kuncheva L I.Rotation forest:a new classifier ensemble method[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2006,28(10):1619-1630.
[8] 王亚辉,李新国,王谦.亚轨道飞行器再入过程执行机构故障仿真[J].飞行力学,2014,32(5):446-449.
[9] Andreas B,Lawrence R,Mohieddine J.Revision of the Tennessee Eastman Process model[J].IFAC-Paper On Line,2015,48(8):309-314.
[10] 王鲜芳,王岁花,杜昊泽,等.基于模糊粗糙集和支持向量机的化工过程故障诊断[J].控制与决策,2015,30(2):353-356.