一种基于流形学习和KNN算法的柴油机工况识别方法
|
摘要
不同负荷状态下的柴油机振动、温度、转速等信号显著不同,而机组故障信号特征往往被淹没在随负荷变化而剧烈变化的信号中,因此变负荷状态下的柴油机故障监测诊断难度较大,一直困扰着柴油机的实际故障诊断工作。提出一种基于流形学习和KNN算法的柴油机工况识别方法,为柴油机变负荷工况下故障监测预警打下基础。方法融合机组的多源信号特征构建特征向量,通过流形学习t-分布邻域嵌入算法(t-SNE)实现特征向量的维数约简和敏感特征提取,采用K最近邻分类算法(KNN)完成柴油机运行负荷状态的自动分类。正常及故障状态下多组柴油机监测数据的处理结果验证了方法的有效性和实用性。
Under different load conditions, the signals of vibration, temperature, speed, etc., of diesel engines are significantly different, and the fault signal characteristics of the engine unit are often submerged in signals that change drastically with the load change. Therefore, the diesel engine's fault monitoring and diagnosing under variable load conditions is difficult and the drastically changed signals always troubles the actual fault diagnosis. This paper presents an operating mode identification method for diesel engines based on manifold learning and KNN algorithm, which lays a foundation for fault monitoring and early warning of diesel engines under variable load conditions. The method combines the multi-source signal features of the unit to construct the feature vector. The feature reduction and sensitive feature extraction of the feature vector is achieved through the manifold learning t-distributed Stochastic Neighbor Embedding(t-SNE)algorithm. The K-Nearest Neighbor(KNN) classification algorithm is used to complete the automatic classification of diesel engine's operating load status. The diesel engine signal under normal and fault conditions verifies the effectiveness and practicality of this method.
Under different load conditions, the signals of vibration, temperature, speed, etc., of diesel engines are significantly different, and the fault signal characteristics of the engine unit are often submerged in signals that change drastically with the load change. Therefore, the diesel engine's fault monitoring and diagnosing under variable load conditions is difficult and the drastically changed signals always troubles the actual fault diagnosis. This paper presents an operating mode identification method for diesel engines based on manifold learning and KNN algorithm, which lays a foundation for fault monitoring and early warning of diesel engines under variable load conditions. The method combines the multi-source signal features of the unit to construct the feature vector. The feature reduction and sensitive feature extraction of the feature vector is achieved through the manifold learning t-distributed Stochastic Neighbor Embedding(t-SNE)algorithm. The K-Nearest Neighbor(KNN) classification algorithm is used to complete the automatic classification of diesel engine's operating load status. The diesel engine signal under normal and fault conditions verifies the effectiveness and practicality of this method.
引文
[1]周平,刘东风,吴千.柴油机油液监测综合评价方法研究[J].润滑油,2010,25(02):35-38.
[2] YADAV G, GANAI P, TIWARI S, et al. An investigation for prior failure of engine component through spectroscopy oil analysis program[J]. Applied Mechanics&Materials, 2014, 592-594(19):1362-1365.
[3]梅检民,张玲玲,肖云魁,等.基于高阶累积量的轴承并发故障振动信号分析[J].内燃机学报,2011,29(4):327-331.
[4]沈虹,赵红东,梅检民,等.基于角域四阶累积量切片谱的柴油机连杆轴承故障特征提取[J].振动与冲击,2014,33(11):90-94.
[5]程利军,张英堂,李志宁,等.基于阶比跟踪及共振解调的连杆轴承故障诊断研究[J].内燃机工程,2012,33(5):67-73.
[6]马发民,张林,王锦彪.维数约简算法简述[J].软件工程,2017,20(8):7-13.
[7] BREGLER C, OMOHUNDRO S M. Nonlinear manifold learning for visual speech recognition[C].//Int. Conf.Computer Vi-sion, 1995.
[8] VAN DER MAATEN L, HINTON G. Visualizing data using t-SN-E[J]. Journal of Machine Learning Research,2008, 9(2579-2605):85.
[9] VAN DER MAATEN L. Accelerating t-SNE using treebased algorithms[J]. Journal of Machine Learning Research, 2014, 15:1-21.
[10] SILVA V D, TENENBAUM J B. Global versus local method-s in nonlinear dimensionality reduction[C].//Proc. of C-onference on Advances in Neural Information Processing Systems, 2003.
[11] ROWEIS S T, SAUL L K. Nonlinear dimensionality reduction by locally linear embedding[J]. Science, 2000,290(5500):2323-2326.
[12] HINTON G E, ROWEIS S T. Stochastic neighbor embedding[C].//Advances in Neural Information Processing Systems, 2002:833-840.
[2] YADAV G, GANAI P, TIWARI S, et al. An investigation for prior failure of engine component through spectroscopy oil analysis program[J]. Applied Mechanics&Materials, 2014, 592-594(19):1362-1365.
[3]梅检民,张玲玲,肖云魁,等.基于高阶累积量的轴承并发故障振动信号分析[J].内燃机学报,2011,29(4):327-331.
[4]沈虹,赵红东,梅检民,等.基于角域四阶累积量切片谱的柴油机连杆轴承故障特征提取[J].振动与冲击,2014,33(11):90-94.
[5]程利军,张英堂,李志宁,等.基于阶比跟踪及共振解调的连杆轴承故障诊断研究[J].内燃机工程,2012,33(5):67-73.
[6]马发民,张林,王锦彪.维数约简算法简述[J].软件工程,2017,20(8):7-13.
[7] BREGLER C, OMOHUNDRO S M. Nonlinear manifold learning for visual speech recognition[C].//Int. Conf.Computer Vi-sion, 1995.
[8] VAN DER MAATEN L, HINTON G. Visualizing data using t-SN-E[J]. Journal of Machine Learning Research,2008, 9(2579-2605):85.
[9] VAN DER MAATEN L. Accelerating t-SNE using treebased algorithms[J]. Journal of Machine Learning Research, 2014, 15:1-21.
[10] SILVA V D, TENENBAUM J B. Global versus local method-s in nonlinear dimensionality reduction[C].//Proc. of C-onference on Advances in Neural Information Processing Systems, 2003.
[11] ROWEIS S T, SAUL L K. Nonlinear dimensionality reduction by locally linear embedding[J]. Science, 2000,290(5500):2323-2326.
[12] HINTON G E, ROWEIS S T. Stochastic neighbor embedding[C].//Advances in Neural Information Processing Systems, 2002:833-840.